all files

.gitattributes

@@ -33,3 +33,13 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+<<<<<<< HEAD
+*.keras filter=lfs diff=lfs merge=lfs -text
+ai_model/models/*.keras filter=lfs diff=lfs merge=lfs -text
+ai_model/models/*.pickle filter=lfs diff=lfs merge=lfs -text
+ai_model/alphabet_model/*.keras filter=lfs diff=lfs merge=lfs -text
+ai_model/alphabet_model/*.pickle filter=lfs diff=lfs merge=lfs -text
+ai_model/llm/glosses/train.csv filter=lfs diff=lfs merge=lfs -text
+ai_model/words/WLASL_v0.3.json filter=lfs diff=lfs merge=lfs -text
+=======
+>>>>>>> 2d9ae59c354470381bb789385838571677cfd5a3

README.md

@@ -1,4 +1,16 @@
 ---
+<<<<<<< HEAD
+title: Handsup Model
+emoji: 😻
+colorFrom: purple
+colorTo: red
+sdk: gradio
+sdk_version: 5.42.0
+app_file: app.py
+pinned: false
+short_description: where ai_models will be hosted
+python_version: "3.11.0"
+=======
 title: NewHandsupModel
 emoji: 😻
 colorFrom: gray
@@ -8,6 +20,7 @@ sdk_version: 5.44.1
 app_file: app.py
 pinned: false
 short_description: 'new model will be here '
+>>>>>>> 2d9ae59c354470381bb789385838571677cfd5a3
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

ai_model/README.md

@@ -0,0 +1,21 @@
+# Installations 
+Author: Karabo
+
+## Istallations in Windows
+1. python3 -m pip install opencv-python
+2. opencv-python-headless
+3. python3 -m pip install mediapipe
+4. python3 -m pip install tensorflow
+5. python3 -m pip install flask
+6. python3 -m pip install python-dotenv
+7. python3 -m pip install flask-cors
+8. python3 -m pip install collections
+9. python3 -m pip install datasets
+10. python3 -m pip install transformers
+11. python3 -m  pip install torch torchvision torchaudio
+12. python3 -m  pip install 'accelerate>=0.26.0'
+13. python3 -m pip install peft
+14. python3 -m pip install python-dotenv
+15. python3 -m pip install openai
+
+

ai_model/jz_model/JZLSTM.py

@@ -0,0 +1,85 @@
+import os
+import pickle
+import numpy as np
+import tensorflow as tf
+from sklearn.model_selection import train_test_split
+from sklearn.preprocessing import LabelEncoder
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import LSTM, Dense
+from tensorflow.keras.callbacks import TensorBoard
+from tensorflow.keras.utils import to_categorical
+from sklearn.metrics import accuracy_score
+
+DATADIR = os.path.join('J_Z')
+actions = ['J','Z']
+sequenceLength = 20
+
+x, y = [], []
+
+for action in actions:
+    actionPath = os.path.join(DATADIR, action)
+    
+    for sequence in os.listdir(actionPath):
+        sequencePath = os.path.join(actionPath, sequence)
+        frames = sorted(os.listdir(sequencePath), key=lambda x: int(os.path.splitext(x)[0]))
+
+        if len(frames) < sequenceLength:
+            continue
+
+        sequenceData = []
+        for frame in frames[:sequenceLength]:
+            framePath = os.path.join(sequencePath, frame)
+            keypoints = np.load(framePath)
+            keypoints = keypoints[:63]
+            sequenceData.append(keypoints)
+            
+        x.append(sequenceData)
+        y.append(action)
+
+x = np.array(x)
+y = np.array(y)
+
+featureLength = x.shape[2]
+
+labelEncoder = LabelEncoder()
+yInt = labelEncoder.fit_transform(y)
+yCategorical = to_categorical(yInt)
+
+with open("labelEncoder.pickle", "wb") as f:
+    pickle.dump(labelEncoder, f)
+
+print("Label mapping:")
+for i, label in enumerate(labelEncoder.classes_):
+    print(f"{i}: {label}")
+
+XTrain, XTest, yTrain, yTest = train_test_split(x, yCategorical, test_size=0.2, random_state=42)
+
+# logDir = os.path.join('Logs')
+# tbCallback = TensorBoard(log_dir=logDir)
+
+model = Sequential([
+    tf.keras.Input(shape=(sequenceLength, featureLength)),
+    LSTM(64, return_sequences=True, activation='relu'),
+    LSTM(128, return_sequences=True, activation='relu'),
+    LSTM(64, return_sequences=False, activation='relu'),
+    Dense(64, activation='relu'),
+    Dense(32, activation='relu'),
+    Dense(yCategorical.shape[1], activation='softmax')
+])
+
+model.compile(optimizer='Adam', loss='categorical_crossentropy', metrics=['categorical_accuracy'])
+
+model.fit(XTrain, yTrain, epochs=30)
+
+model.summary()
+
+model.save('JZModel.keras')
+
+yPred = model.predict(XTest)
+predictedInts = np.argmax(yPred, axis=1)
+actualInts = np.argmax(yTest, axis=1)
+
+predictedLabels = labelEncoder.inverse_transform(predictedInts)
+actualLabels = labelEncoder.inverse_transform(actualInts)
+
+print("Test Accuracy:", accuracy_score(actualLabels, predictedLabels))

ai_model/jz_model/collectLandmarks.py

@@ -0,0 +1,64 @@
+import os
+import numpy as np
+import cv2
+import mediapipe as mp
+
+actions = ['J']
+noSequences = 30
+sequenceLength = 20
+startFolder = 0
+
+DATADIR = os.path.join('J_Z')  
+
+for action in actions:
+    for sequence in range(startFolder, startFolder + noSequences):
+        os.makedirs(os.path.join(DATADIR, action, str(sequence)), exist_ok=True)
+
+def mediapipeDetection(image, model):
+    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+    image.flags.writeable = False
+    results = model.process(image)
+    image.flags.writeable = True
+    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+    return image, results
+
+def drawStyledLandmarks(image, results):
+    mpDrawing = mp.solutions.drawing_utils
+    mpDrawingStyles = mp.solutions.drawing_styles
+
+    if results.right_hand_landmarks:
+        mpDrawing.draw_landmarks(image, results.right_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+
+def extractKeypoints(results):
+    rh = np.array([[res.x, res.y, res.z] for res in results.right_hand_landmarks.landmark]).flatten() if results.right_hand_landmarks else np.zeros(21*3)
+    return rh
+
+
+cap = cv2.VideoCapture(0)
+with mp.solutions.holistic.Holistic(min_detection_confidence=0.5, min_tracking_confidence=0.3) as holistic:
+
+    for action in actions:
+        for sequence in range(startFolder, startFolder + noSequences):
+            print(f"Collecting for {action}, sequence {sequence}")
+            for frameNum in range(sequenceLength):
+
+                ret, frame = cap.read()
+                image, results = mediapipeDetection(frame, holistic)
+                drawStyledLandmarks(image, results)
+
+                if frameNum == 0:
+                    cv2.putText(image, 'STARTING COLLECTION', (120,200), cv2.FONT_HERSHEY_SIMPLEX, 1, (0,255,0), 4, cv2.LINE_AA)
+                    cv2.imshow('OpenCV Feed', image)
+                    cv2.waitKey(1000)  
+                else:
+                    cv2.putText(image, f'Collecting {action}, Video #{sequence}', (10, 15), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,0,255), 2, cv2.LINE_AA)
+                    cv2.imshow('OpenCV Feed', image)
+
+                keypoints = extractKeypoints(results)
+                np.save(os.path.join(DATADIR, action, str(sequence), str(frameNum)), keypoints)
+
+                if cv2.waitKey(10) & 0xFF == ord('q'):
+                    break
+
+cap.release()
+cv2.destroyAllWindows()

ai_model/llm/converterUsage.py

@@ -0,0 +1,6 @@
+from glossToEnglish import translateGloss
+
+glossInput = "APPLE TABLE ON"
+translation = translateGloss(glossInput)
+
+print("Translation:", translation)

ai_model/llm/glossToEnglish.py

@@ -0,0 +1,17 @@
+import os
+import dotenv
+from openai import OpenAI
+
+dotenv.load_dotenv()
+
+client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+
+def translateGloss(gloss: str, model: str = "o3-mini") -> str:
+
+  response = client.chat.completions.create(
+      model=model,
+      messages=[
+          {"role": "user", "content": f"Translate this asl gloss to English: '{gloss}, output as 'English sentence : '"}
+      ]
+  )
+  return response.choices[0].message.content

ai_model/models/detectLettersModel.py

@@ -0,0 +1,89 @@
+import pickle
+import numpy as np
+import tensorflow as tf
+from sklearn.preprocessing import LabelEncoder
+from sklearn.metrics import accuracy_score
+from tensorflow.keras.callbacks import EarlyStopping
+
+early_stop = EarlyStopping(patience=2, restore_best_weights=True)
+
+dataDictTrain = pickle.load(open('../processed_data/trainData.pickle', 'rb'))
+
+cleanedData = []
+cleanedLabels = []
+for i, item in enumerate(dataDictTrain['data']):
+    if isinstance(item, (np.ndarray, list)) and len(item) == 42:
+        cleanedData.append(np.array(item, dtype=np.float32))
+        cleanedLabels.append(dataDictTrain['labels'][i])
+
+xTrain = np.array(cleanedData)
+yTrainRaw = np.array(cleanedLabels)
+
+dataDictTest = pickle.load(open('../processed_data/testData.pickle', 'rb'))
+xTestRaw = np.array(dataDictTest['data'], dtype=np.float32)
+yTestRaw = np.array(dataDictTest['labels'])
+
+xTrain = xTrain.reshape(-1, 42, 1)
+xTest = xTestRaw.reshape(-1, 42, 1)
+
+labelEncoder = LabelEncoder()
+labelEncoder.fit(yTrainRaw)
+
+yTrainEncoded = labelEncoder.transform(yTrainRaw)
+yTestEncoded = labelEncoder.transform(yTestRaw)
+
+#second data set for finetuning
+dataDictTrain2 = pickle.load(open('../processed_data/fineTuneData.pickle', 'rb'))
+
+cleanedData2 = []
+cleanedLabels2 = []
+for i, item in enumerate(dataDictTrain2['data']):
+    if isinstance(item, (np.ndarray, list)) and len(item) == 42:
+        cleanedData2.append(np.array(item, dtype=np.float32))
+        cleanedLabels2.append(dataDictTrain2['labels'][i])
+
+xTrain2 = np.stack(cleanedData2)
+yTrainRaw2= np.array(cleanedLabels2)
+
+xTrain2 = xTrain2.reshape(-1, 42, 1)
+
+labelEncoder2 = LabelEncoder()
+labelEncoder2.fit(yTrainRaw2)
+
+yTrainEncoded2 = labelEncoder2.transform(yTrainRaw2)
+
+model = tf.keras.models.Sequential([
+    tf.keras.layers.Conv1D(32, kernel_size=3, activation='relu', input_shape=(42, 1)),
+    tf.keras.layers.MaxPooling1D(pool_size=2),
+    tf.keras.layers.Conv1D(64, kernel_size=3, activation='relu'),
+    tf.keras.layers.Flatten(),
+    tf.keras.layers.Dense(64, activation='relu'),
+    tf.keras.layers.Dense(len(labelEncoder.classes_), activation='softmax')
+])
+
+model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
+model.fit(xTrain, yTrainEncoded, epochs=15, batch_size=32, validation_split=0.2)
+
+for layer in model.layers[:-2]:
+    layer.trainable = False
+
+model.compile(optimizer=tf.keras.optimizers.Adam(1e-5),loss='sparse_categorical_crossentropy',metrics=['accuracy'])
+
+yTrainEncoded2 = labelEncoder2.transform(yTrainRaw2)
+
+model.fit(xTrain2, yTrainEncoded2, epochs=10, batch_size=32, validation_split=0.2, callbacks = [early_stop])
+
+testLoss, testAccuracy = model.evaluate(xTest, yTestEncoded)
+print(f"\nTest accuracy: {testAccuracy * 100:.2f}%")
+
+predictions = model.predict(xTest)
+predictedIndex = np.argmax(predictions, axis=1)
+predictedLabels = labelEncoder.inverse_transform(predictedIndex)
+
+print("\nPredictions:")
+for i, pred in enumerate(predictedLabels):
+    print(f"Image {i}: Predicted = {pred}, Actual = {yTestRaw[i]}")
+
+model.save("detectLettersModel.keras")
+with open("labelEncoder.pickle", "wb") as file:
+    pickle.dump(labelEncoder, file)

ai_model/models/detectNumbersModel.py

@@ -0,0 +1,57 @@
+import pickle
+import numpy as np
+import tensorflow as tf
+from sklearn.preprocessing import LabelEncoder
+from sklearn.metrics import accuracy_score
+
+dataDictTrain = pickle.load(open('../processed_data/numTrainData.pickle', 'rb'))
+
+cleanedData = []
+cleanedLabels = []
+for i, item in enumerate(dataDictTrain['data']):
+    if isinstance(item, (np.ndarray, list)) and len(item) == 42:
+        cleanedData.append(np.array(item, dtype=np.float32))
+        cleanedLabels.append(dataDictTrain['labels'][i])
+
+xTrain = np.array(cleanedData)
+yTrainRaw = np.array(cleanedLabels)
+
+dataDictTest = pickle.load(open('../processed_data/numTestData.pickle', 'rb'))
+xTestRaw = np.array(dataDictTest['data'], dtype=np.float32)
+yTestRaw = np.array(dataDictTest['labels'])
+
+xTrain = xTrain.reshape(-1, 42, 1)
+xTest = xTestRaw.reshape(-1, 42, 1)
+
+labelEncoder = LabelEncoder()
+labelEncoder.fit(yTrainRaw)
+
+yTrainEncoded = labelEncoder.transform(yTrainRaw)
+yTestEncoded = labelEncoder.transform(yTestRaw)
+
+model = tf.keras.models.Sequential([
+    tf.keras.layers.Conv1D(32, kernel_size=3, activation='relu', input_shape=(42, 1)),
+    tf.keras.layers.MaxPooling1D(pool_size=2),
+    tf.keras.layers.Conv1D(64, kernel_size=3, activation='relu'),
+    tf.keras.layers.Flatten(),
+    tf.keras.layers.Dense(64, activation='relu'),
+    tf.keras.layers.Dense(len(labelEncoder.classes_), activation='softmax')
+])
+
+model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
+model.fit(xTrain, yTrainEncoded, epochs=15, batch_size=32, validation_split=0.2)
+
+testLoss, testAccuracy = model.evaluate(xTest, yTestEncoded)
+print(f"\nTest accuracy: {testAccuracy * 100:.2f}%")
+
+predictions = model.predict(xTest)
+predictedIndex = np.argmax(predictions, axis=1)
+predictedLabels = labelEncoder.inverse_transform(predictedIndex)
+
+print("\nPredictions:")
+for i, pred in enumerate(predictedLabels):
+    print(f"Image {i}: Predicted = {pred}, Actual = {yTestRaw[i]}")
+
+model.save("detectNumbersModel.keras")
+with open("numLabelEncoder.pickle", "wb") as file:
+    pickle.dump(labelEncoder, file)

ai_model/pre_processing/extractVideoLandmarks.py

@@ -0,0 +1,67 @@
+import cv2
+import mediapipe as mp
+import json
+import os
+
+def extractLandmarksFromVideo(videoPath, letter, outputDir='landmarks'):
+    mpHands = mp.solutions.hands
+    hands = mpHands.Hands(static_image_mode=False, max_num_hands=1, min_detection_confidence=0.5)
+    cap = cv2.VideoCapture(videoPath)
+
+    landmarksData = []
+
+    while cap.isOpened():
+        success, frame = cap.read()
+        if not success:
+            break
+
+        frame = cv2.flip(frame, 1)
+        imageRGB = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+
+        results = hands.process(imageRGB)
+
+        frameLandmarks = []
+        if results.multi_hand_landmarks:
+            handLandmarks = results.multi_hand_landmarks[0]
+            for lm in handLandmarks.landmark:
+                frameLandmarks.append({
+                    "x": lm.x,
+                    "y": lm.y,
+                    "z": lm.z,
+                })
+
+        if frameLandmarks:
+            landmarksData.append(frameLandmarks)
+
+    cap.release()
+    hands.close()
+
+    os.makedirs(outputDir, exist_ok=True)
+    outputFilename = f"{letter.upper()}.json"
+    outputPath = os.path.join(outputDir, outputFilename)
+    with open(outputPath, "w") as f:
+        json.dump({"frames": landmarksData}, f, indent=2)
+
+    print(f"Saved landmarks to {outputPath}")
+
+def processAllVideosInDirectory(videoDirectory, outputDirectory='landmarks'):
+    if not os.path.isdir(videoDirectory):
+        print(f"Error: Video directory '{videoDirectory}' not found.")
+        return
+
+    os.makedirs(outputDirectory, exist_ok=True)
+
+    for filename in os.listdir(videoDirectory):
+        if filename.lower().endswith(('.mp4')): 
+            videoPath = os.path.join(videoDirectory, filename)
+            letter = os.path.splitext(filename)[0]
+            print(f"Processing video: {filename} for letter: {letter}")
+            extractLandmarksFromVideo(videoPath, letter, outputDirectory)
+        else:
+            print(f"Skipping non-video file: {filename}")
+
+if __name__ == "__main__":
+    videoInputDirectory = "../video_data" 
+    landmarkOutputDirectory = "../processed_data/landmarks" 
+    
+    processAllVideosInDirectory(videoInputDirectory, landmarkOutputDirectory)

ai_model/pre_processing/processFineTune.py

@@ -0,0 +1,60 @@
+import os
+import pickle
+
+import mediapipe as mp
+import cv2
+import matplotlib.pyplot as plt
+
+mpHands = mp.solutions.hands
+mpDrawing = mp.solutions.drawing_utils
+mpDrawingStyles = mp.solutions.drawing_styles
+
+hands = mpHands.Hands(static_image_mode=True, min_detection_confidence=0.3)
+
+DATADIR = '../letters_fine_tune'
+
+data = []
+labels = []
+for dir in os.listdir(DATADIR):
+    for imgName in os.listdir(os.path.join(DATADIR, dir)):
+        dataAux = []
+
+        x_ = []
+        y_ = []
+
+        imgPath = os.path.join(DATADIR, dir, imgName)
+        print("Reading:", imgPath)
+
+        img = cv2.imread(imgPath)
+        if img is None:
+            print("Failed to load:", imgPath)
+            continue  
+
+        imgRGB = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+
+        results = hands.process(imgRGB)
+        if results.multi_hand_landmarks:
+            for handLandmarks in results.multi_hand_landmarks:
+                
+                for i in range(len(handLandmarks.landmark)):
+                    x = handLandmarks.landmark[i].x
+                    y = handLandmarks.landmark[i].y
+
+                    x_.append(x)
+                    y_.append(y)
+
+                for i in range(len(handLandmarks.landmark)):
+                    x = handLandmarks.landmark[i].x
+                    y = handLandmarks.landmark[i].y
+                    dataAux.append(x - min(x_))
+                    dataAux.append(y - min(y_))
+
+            data.append(dataAux)
+            labels.append(dir.strip().upper())
+
+print(labels)
+
+file = open('../processed_data/fineTuneData.pickle', 'wb')
+pickle.dump({'data': data, 'labels': labels}, file)
+file.close()
+

ai_model/pre_processing/processTestData.py

@@ -0,0 +1,75 @@
+import pickle
+import mediapipe as mp
+import cv2
+import matplotlib.pyplot as plt
+import os
+
+mpHands = mp.solutions.hands
+mpDrawing = mp.solutions.drawing_utils
+mpDrawingStyles = mp.solutions.drawing_styles
+
+# def brighten_image(img, factor=1.5):
+#     return cv2.convertScaleAbs(img, alpha=factor, beta=20)
+
+hands = mpHands.Hands(static_image_mode=True, min_detection_confidence=0.3)
+
+DATADIR = '../letters/asl_alphabet_test'
+
+data = []
+labels = []
+for imgName in os.listdir(DATADIR):
+    dataAux = []
+
+    x_ = []
+    y_ = []
+
+    imgPath = os.path.join(DATADIR,imgName)
+    print("Reading:", imgPath)
+
+    img = cv2.imread(imgPath)
+    if img is None:
+        print("Failed to load:", imgPath)
+        continue  
+
+    img = cv2.imread(imgPath)
+    imgRGB = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+
+    results = hands.process(imgRGB)
+    if results.multi_hand_landmarks:
+        for handLandmarks in results.multi_hand_landmarks:
+            
+            mpDrawing.draw_landmarks(
+                img,  
+                handLandmarks,
+                mpHands.HAND_CONNECTIONS,
+                mpDrawingStyles.get_default_hand_landmarks_style(),
+                mpDrawingStyles.get_default_hand_connections_style()
+            )
+
+            for i in range(len(handLandmarks.landmark)):
+                x = handLandmarks.landmark[i].x
+                y = handLandmarks.landmark[i].y
+
+                x_.append(x)
+                y_.append(y)
+
+            for i in range(len(handLandmarks.landmark)):
+                x = handLandmarks.landmark[i].x
+                y = handLandmarks.landmark[i].y
+                dataAux.append(x - min(x_))
+                dataAux.append(y - min(y_))
+
+        ind = imgName.find("test")
+
+        data.append(dataAux)
+        labels.append(imgName[0:ind-1].upper())
+
+        # cv2.imshow("Hand Landmarks", img)
+        # cv2.waitKey(0)
+        # cv2.destroyAllWindows() 
+    
+print(labels)
+
+file = open('../processed_data/testData.pickle', 'wb')
+pickle.dump({'data': data, 'labels': labels}, file)
+file.close()

ai_model/pre_processing/processTrainData.py

@@ -0,0 +1,64 @@
+import os
+import pickle
+
+import mediapipe as mp
+import cv2
+import matplotlib.pyplot as plt
+
+mpHands = mp.solutions.hands
+mpDrawing = mp.solutions.drawing_utils
+mpDrawingStyles = mp.solutions.drawing_styles
+
+hands = mpHands.Hands(static_image_mode=True, min_detection_confidence=0.3)
+
+DATADIR = '../numbers/nums_test'
+
+data = []
+labels = []
+for dir in os.listdir(DATADIR):
+    #if dir.strip().upper() == 'C': #Starting with C because dataset too big
+    for imgName in os.listdir(os.path.join(DATADIR, dir)):
+        dataAux = []
+
+        x_ = []
+        y_ = []
+
+        imgPath = os.path.join(DATADIR, dir, imgName)
+        print("Reading:", imgPath)
+
+        img = cv2.imread(imgPath)
+        if img is None:
+            print("Failed to load:", imgPath)
+            continue  
+
+        img = cv2.imread(imgPath) # this iwwill read image and save to img variable
+        imgRGB = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+
+        results = hands.process(imgRGB)
+        if results.multi_hand_landmarks:
+            for handLandmarks in results.multi_hand_landmarks:
+                
+                for i in range(len(handLandmarks.landmark)):
+                    x = handLandmarks.landmark[i].x
+                    y = handLandmarks.landmark[i].y
+
+                    x_.append(x)
+                    y_.append(y)
+
+                for i in range(len(handLandmarks.landmark)):
+                    x = handLandmarks.landmark[i].x
+                    y = handLandmarks.landmark[i].y
+                    dataAux.append(x - min(x_))
+                    dataAux.append(y - min(y_))
+
+            data.append(dataAux)
+            labels.append(dir.strip().upper())
+
+            # cv2.imshow("Hand Landmarks", img)
+            # cv2.waitKey(0)
+print(labels)
+
+file = open('../processed_data/numTestData.pickle', 'wb')
+pickle.dump({'data': data, 'labels': labels}, file)
+file.close()
+

ai_model/words/README copy.md

@@ -0,0 +1,75 @@
+MS-ASL: A Large-Scale Data Set and Benchmark for Understanding American Sign Language
+============================================================================================
+
+This package containing the `MS-ASL dataset`, as proposed in MS-ASL: A Large-Scale Data Set and Benchmark for Understanding American Sign Language paper.  
+For more information visit [the website](https://www.microsoft.com/en-us/research/project/ms-asl/).
+
+Contents
+-----------------
+The directory contains following files:
+
+ * `MSASL_train.json`: includes a json file of 16054 train samples.
+
+ * `MSASL_test.json`: includes a json file of 4172 test samples.
+
+ * `MSASL_val.json`: includes a json file of 5287 validation samples.
+
+ * `MSASL_classes.json`: includes a json file of 1000 glosses (words) representing classes in the classification task. The first word in the set (`hello`) is class 0 and the last world on the set (`challenge`) is class 999.
+
+ * `MSASL_synonym.json`: each row is a list of words that we considered synonym and assign a single class for all of them.
+
+ * `C-UDA-0.1_annotated_discussion.pdf`: the Computational Use of Data Agreement (C-UDA) lisense agreement.
+
+ * `README.md`: this file.
+
+
+Structure
+-----------------
+Here is the structure of each sample clip in Train, test or validation sets.
+
+
+{'url': a url link to the video
+ 'start_time': the starting point of the clip in the original video in seconds
+ 'end_time': the starting point of the clip in the original video in seconds
+ 'label': class (an integer between 0 to 1000)
+ 'signer_id': the id of the signer
+ 'box': the boudy bounding box of the signer such as [y0, x0, y1, x1] where (x0, y0) is up-left corner and (x1,y1) is bottom-right corner. All the values are normalized (between zero and one) according to width and height.
+ 'text': the gloss for this clip which match the 'label',
+ 'width': height for the original video
+ 'height': height for the original video
+ 'fps': frame per second for the original video
+}
+
+We are suggesting the width, height and fps because the labeling has been done based on them but since start and end are time base and box is normalize in theory this annotation works for arbitrary width, height and fps.
+
+
+Subsets
+---------------
+As mention in the paper this dataset includes 4 subsets: MS-ASL100, MS-ASL200, MS-ASL500 and MS-ASL1000.
+To obtain each of these subsets you need to filter the train, test and validation based on the 'label'. This means that for MS-ASL100 just keep the samples with 'label' less that 100.
+
+
+Credits
+---------------
+Licensed under the Computational Use of Data Agreement (C-UDA). Plaese refer to C-UDA-0.1_annotated_discussion.pdf for more information. 
+
+
+Citation
+--------------
+
+Please cite this in your publications if it help your research:
+
+    `@InProceedings{vaezijoze2019ms-asl,
+      author    = {Vaezi Joze, Hamid Reza and Koller, Oscar}
+      title     = {MS-ASL: A Large-Scale Data Set and Benchmark for Understanding American Sign Language},
+      booktitle = {The British Machine Vision Conference (BMVC)},
+      year      = {2019},
+      month = {September}
+    }`
+
+
+Contacts
+------------------
+- [Hamid Vaezi Joze](https://www.microsoft.com/en-us/research/people/hava/)
+- [Oscar Koller](https://www.microsoft.com/en-us/research/people/oskoller/)
+

ai_model/words/README.md

@@ -0,0 +1,33 @@
+TO set up the enviroment:
+python -m venv venv_signlang
+  your should see (venv-signlang)
+run: pip install numpy opencv-python mediapipe pandas
+next run:  python.exe
+and to check if its working :>>>
+>>> import numpy
+>>> import cv2
+>>> import mediapipe
+>>> import pandas
+>>> print("All libraries imported successfully!")
+you should see : All libraries imported successfully!
+
+ENVIROMENT SETUP COMPLETE!!!
+
+
+next add all files into root
+-venv_signlang
+-videos
+-missing.txt
+nslt_100.json
+nslt_300.json
+nslt_1000.json
+nslt_2000.json
+wlasl_class_list.txt
+WLASL_v0.3.json
+and add 2 new files:
+data_explorer.py
+WLASL_parser.py
+
+THis outlines how the code works:
+1) python wlasl_parser.py is run which aims to identify the missing videos in the WLASL dataset. THese are outlined in missing.txt and will affect the accurcay of the model if not done.
+2

ai_model/words/WLASL_parser.py

@@ -0,0 +1,132 @@
+import os
+import json
+import pandas as pd
+import cv2 
+
+# --- CONFIGURATION FOR TARGET WORD LIST ---
+# This is your CUMULATIVE 20-word subset (lowercase recommended for WLASL compatibility)
+TARGET_GLOSS_LIST = [
+    # Your original 10 words:
+    "apple", "black", "blue", "brown", "can", "cat", "chair", "child", "cold", "come",
+    # Your NEW 10 words:
+    "computer", "cow", "cry", "cup", "deaf", "dog", "drink", "drive", "eat", "egg"
+]
+TARGET_GLOSS_SET = set(g.lower() for g in TARGET_GLOSS_LIST) # Convert to set of lowercase glosses
+
+# --- GLOSSES TO EXCLUDE (ALPHABETS) ---
+ALPHABET_GLOSSES = [
+    "A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", 
+    "N", "O", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z"
+]
+GLOSSES_TO_EXCLUDE_SET = set(g.lower() for g in ALPHABET_GLOSSES) # Convert to lowercase set for comparison
+# --- END CONFIGURATION ---
+
+def parse_wlasl_dataset(base_data_path):
+    videos_dir = os.path.join(base_data_path, 'videos')
+    wlasl_json_path = os.path.join(base_data_path, 'WLASL_v0.3.json')
+    missing_txt_path = os.path.join(base_data_path, 'missing.txt')
+    class_list_path = os.path.join(base_data_path, 'wlasl_class_list.txt') # For original class list
+
+    print(f"Parsing WLASL dataset from: {base_data_path}")
+
+    if not os.path.exists(wlasl_json_path):
+        print(f"Error: WLASL_v0.3.json not found at {wlasl_json_path}")
+        return None
+
+    with open(wlasl_json_path, 'r') as f:
+        wlasl_data = json.load(f)
+    print(f"Loaded WLASL_v0.3.json with {len(wlasl_data)} entries (glosses).")
+
+    missing_video_ids = set()
+    if os.path.exists(missing_txt_path):
+        with open(missing_txt_path, 'r') as f:
+            for line in f:
+                missing_video_ids.add(line.strip())
+        print(f"Loaded missing.txt with {len(missing_video_ids)} missing video IDs.")
+    else:
+        print(f"Warning: missing.txt not found at {missing_txt_path}. Proceeding without it.")
+
+    gloss_id_map_orig = {} 
+    if os.path.exists(class_list_path):
+        with open(class_list_path, 'r') as f:
+            for line in f:
+                parts = line.strip().split('\t')
+                if len(parts) == 2:
+                    gloss_id_map_orig[int(parts[0])] = parts[1]
+            
+    data_records = []
+    available_videos_count = 0
+    total_instances = 0
+
+    for gloss_entry in wlasl_data:
+        gloss = gloss_entry['gloss']
+        for instance in gloss_entry['instances']:
+            total_instances += 1
+            video_id = instance['video_id']
+
+            if video_id in missing_video_ids:
+                continue
+
+            # --- FILTERING: INCLUDE IN TARGET SET AND NOT IN EXCLUDE SET ---
+            gloss_lower = gloss.lower()
+            if gloss_lower not in TARGET_GLOSS_SET:
+                continue # Skip if not in our desired target list
+            if gloss_lower in GLOSSES_TO_EXCLUDE_SET:
+                # print(f"Skipping excluded alphabet gloss: {gloss}") # Uncomment to see what's skipped
+                continue # Skip if it's an alphabet gloss
+
+            video_filename = f"{video_id}.mp4"
+            video_full_path = os.path.join(videos_dir, video_filename)
+
+            if os.path.exists(video_full_path):
+                available_videos_count += 1
+                data_records.append({
+                    'gloss': gloss,
+                    'gloss_id': gloss_id_map_orig.get(gloss, -1), # Use original ID for now, re-map later
+                    'video_id': video_id,
+                    'video_path': video_full_path,
+                    'signer_id': instance.get('signer_id'),
+                    'split': instance.get('split'),
+                    'frame_start': instance.get('frame_start'),
+                    'frame_end': instance.get('frame_end'),
+                    'fps': instance.get('fps'),
+                    'bbox': instance.get('bbox')
+                })
+            else:
+                pass 
+            
+    df = pd.DataFrame(data_records)
+    print(f"\n--- Parsing Summary ---")
+    print(f"Total instances in WLASL_v0.3.json: {total_instances}")
+    print(f"Instances skipped due to 'missing.txt' and not found on disk: {total_instances - available_videos_count}")
+    print(f"Total *available* and valid video instances found (before target list filter): {available_videos_count}")
+    
+    return df
+
+if __name__ == "__main__":
+    wlasl_dataset_path = "." 
+    wlasl_df = parse_wlasl_dataset(wlasl_dataset_path)
+
+    if wlasl_df is not None:
+        df_filtered_by_glosses = wlasl_df[wlasl_df['gloss'].str.lower().isin(TARGET_GLOSS_SET)].copy()
+        
+        # --- NEW EXCLUSION: Remove alphabet glosses from the final filtered DataFrame ---
+        df_filtered_by_glosses = df_filtered_by_glosses[~df_filtered_by_glosses['gloss'].str.lower().isin(GLOSSES_TO_EXCLUDE_SET)].copy()
+
+        # Re-map gloss_ids to be sequential (0 to N-1) for the new, smaller set of classes
+        unique_filtered_glosses = df_filtered_by_glosses['gloss'].unique()
+        new_gloss_to_id = {gloss: i for i, gloss in enumerate(unique_filtered_glosses)}
+        df_filtered_by_glosses['gloss_id'] = df_filtered_by_glosses['gloss'].map(new_gloss_to_id)
+
+        print(f"\n--- Fixed Word List Subset Summary ({len(TARGET_GLOSS_SET)} words requested, {df_filtered_by_glosses['gloss'].nunique()} words after filtering) ---")
+        print(f"Total instances in fixed list subset: {len(df_filtered_by_glosses)}")
+        print(f"Unique glosses in fixed list: {df_filtered_by_glosses['gloss'].nunique()} (Expected: {len(TARGET_GLOSS_SET) - len(GLOSSES_TO_EXCLUDE_SET.intersection(TARGET_GLOSS_SET))})")
+        print(f"Distribution of splits:\n{df_filtered_by_glosses['split'].value_counts()}")
+        print("First 5 entries of Fixed List DataFrame:")
+        print(df_filtered_by_glosses.head())
+
+        output_csv_name = 'wlasl_20_words_processed.csv' # <-- CHANGED TO 20 WORDS
+        df_filtered_by_glosses.to_csv(output_csv_name, index=False)
+        print(f"\nFixed list processed data saved to '{output_csv_name}'")
+    else:
+        print("\nError: Could not parse base WLASL dataset. Cannot filter by fixed list.")

ai_model/words/WLASL_v0.3.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f8510e2de97930e7bd46f1536aba9541fa13b60d6bdf48643413b1944b2eb80
+size 12322133

ai_model/words/analyze_durations.py

@@ -0,0 +1,59 @@
+# analyze_durations.py
+import pandas as pd
+import os
+import cv2 
+
+# IMPORTANT: This path must point to the CSV created in the previous step (wlasl_20_words_processed.csv)
+PROCESSED_DATA_CSV = 'wlasl_20_words_processed.csv' # <-- CHANGED
+# Path to your main WLASL videos directory (e.g., if videos are in ./videos/)
+WLASL_VIDEOS_DIR = './videos' 
+
+
+if os.path.exists(PROCESSED_DATA_CSV):
+    df_subset = pd.read_csv(PROCESSED_DATA_CSV)
+
+    actual_durations = []
+
+    print("Calculating actual video durations (this might take a moment)...")
+    for index, row in df_subset.iterrows():
+        video_path = row['video_path']
+        frame_start = row['frame_start']
+        frame_end = row['frame_end']
+
+        duration = 0
+        if frame_end != -1 and frame_end >= frame_start:
+            duration = frame_end - frame_start + 1
+        else:
+            full_video_path_on_disk = video_path 
+
+            cap = cv2.VideoCapture(full_video_path_on_disk)
+            if cap.isOpened():
+                total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+                if total_frames > 0 and frame_start > 0:
+                    duration = total_frames - frame_start + 1
+                    if duration < 0: duration = 0 
+                cap.release()
+            else:
+                print(f"Warning: Could not open video file {full_video_path_on_disk} to get full duration. Assigning 0.")
+                duration = 0 
+
+        actual_durations.append(duration)
+
+    df_subset['duration_frames'] = actual_durations
+
+    df_subset = df_subset[df_subset['duration_frames'] > 0]
+
+
+    print(f"--- WLASL Fixed List Video Duration Analysis (20 Words - CORRECTED) ---")
+    print(f"Total instances analyzed: {len(df_subset)}")
+    print(f"Min duration (frames): {df_subset['duration_frames'].min()}")
+    print(f"Max duration (frames): {df_subset['duration_frames'].max()}")
+    print(f"Mean duration (frames): {df_subset['duration_frames'].mean():.2f}")
+    print(f"Median duration (frames): {df_subset['duration_frames'].median()}")
+    print(f"75th percentile duration (frames): {df_subset['duration_frames'].quantile(0.75)}")
+    print(f"90th percentile duration (frames): {df_subset['duration_frames'].quantile(0.90)}")
+
+    print(f"\nMean FPS: {df_subset['fps'].mean():.2f}")
+    print(f"Most common FPS: {df_subset['fps'].mode()[0] if not df_subset['fps'].mode().empty else 'N/A'}")
+else:
+    print(f"Error: {PROCESSED_DATA_CSV} not found. Please run wlasl_parser.py for your fixed list first.")

ai_model/words/combine_datasets.py

@@ -0,0 +1,92 @@
+import pandas as pd
+import numpy as np
+import os
+import re # For regular expressions to parse filenames
+
+# --- Configuration ---
+WLASL_PROCESSED_CSV = 'wlasl_10_words_processed.csv'
+MY_RECORDED_SIGNS_DIR = 'my_recorded_signs'
+COMBINED_PROCESSED_CSV = 'wlasl_10_words_combined_processed.csv'
+
+print(f"--- Combining WLASL and Personal Data ---")
+
+# Load WLASL data
+if not os.path.exists(WLASL_PROCESSED_CSV):
+    print(f"Error: WLASL processed CSV not found at {WLASL_PROCESSED_CSV}.")
+    print("Please ensure wlasl_parser.py has been run for your 10-word list.")
+    exit()
+
+df_wlasl = pd.read_csv(WLASL_PROCESSED_CSV)
+print(f"Loaded {len(df_wlasl)} instances from WLASL: {WLASL_PROCESSED_CSV}")
+
+# --- Process Personal Recordings ---
+personal_records = []
+if os.path.exists(MY_RECORDED_SIGNS_DIR):
+    print(f"Scanning for personal recordings in: {MY_RECORDED_SIGNS_DIR}")
+    for filename in os.listdir(MY_RECORDED_SIGNS_DIR):
+        if filename.endswith('.npy'):
+            # Example filename: 'apple_1678901234.npy'
+            match = re.match(r'(.+?)_(\d+)\.npy', filename)
+            if match:
+                gloss = match.group(1).lower() # Extract gloss, ensure lowercase
+                video_id = match.group(2) # Use timestamp as a unique ID
+
+                full_npy_path = os.path.join(MY_RECORDED_SIGNS_DIR, filename)
+
+                # We need to verify if the original video_id exists for context, 
+                # but for personal recordings, the NPY is the "video" representation.
+                # We'll treat the NPY path as the 'video_path' for these.
+
+                # Get original gloss_id mapping from WLASL data to ensure consistency
+                # This assumes all personal glosses are already in the 10-word list
+                original_gloss_to_id = {g:i for i, g in enumerate(df_wlasl['gloss'].unique())}
+
+                if gloss in original_gloss_to_id:
+                    personal_records.append({
+                        'gloss': gloss,
+                        'gloss_id': original_gloss_to_id[gloss], # Use WLASL's mapping
+                        'video_id': f'personal_{video_id}', # Prefix to differentiate from WLASL
+                        'video_path': full_npy_path, # Path to the NPY file itself
+                        'signer_id': 'personal',
+                        'split': 'train', # New personal data always goes to train
+                        'frame_start': 1, # Dummy values as it's already extracted
+                        'frame_end': -1,  # Dummy values
+                        'fps': 25,        # Assume 25 FPS for consistency
+                        'bbox': [0,0,0,0] # Dummy bbox
+                    })
+                else:
+                    print(f"Warning: Personal recording '{filename}' has gloss '{gloss}' which is not in the 10-word WLASL list. Skipping.")
+            else:
+                print(f"Warning: Could not parse filename for '{filename}'. Skipping.")
+else:
+    print(f"Warning: Personal recordings directory '{MY_RECORDED_SIGNS_DIR}' not found. No personal data will be added.")
+
+df_personal = pd.DataFrame(personal_records)
+print(f"Found {len(df_personal)} personal recordings.")
+
+# Combine DataFrames
+# We need to ensure columns are compatible for concatenation
+# The 'video_path' for WLASL points to MP4, for personal it points to NPY.
+# landmark_extractor and data_preprocessor will need to handle this.
+df_combined = pd.concat([df_wlasl, df_personal], ignore_index=True)
+
+# Re-map gloss_ids to be sequential (0 to N-1) for the entire combined set
+# This is CRITICAL because the unique glosses list might change if warnings were ignored
+final_unique_glosses = df_combined['gloss'].unique()
+final_gloss_to_id = {g: i for i, g in enumerate(final_unique_glosses)}
+df_combined['gloss_id'] = df_combined['gloss'].map(final_gloss_to_id)
+
+
+print(f"\n--- Combined Dataset Summary ---")
+print(f"Total instances in combined dataset: {len(df_combined)}")
+print(f"Unique glosses in combined dataset: {df_combined['gloss'].nunique()} (Expected: {len(df_wlasl['gloss'].unique())})")
+print(f"Distribution of splits:\n{df_combined['split'].value_counts()}")
+print(f"First 5 entries of Combined DataFrame:\n{df_combined.head()}")
+print(f"Last 5 entries of Combined DataFrame (should include personal data):\n{df_combined.tail()}")
+
+df_combined.to_csv(COMBINED_PROCESSED_CSV, index=False)
+print(f"\nCombined dataset metadata saved to '{COMBINED_PROCESSED_CSV}'")
+
+# --- Special Note for landmark_extractor.py ---
+print("\nIMPORTANT: Next, you MUST modify landmark_extractor.py to handle .npy video_paths for personal data.")
+print("          Look for changes related to 'full_video_path_on_disk' in the extract_landmarks_from_video function.")

ai_model/words/create_personal_dataset_csv.py

@@ -0,0 +1,241 @@
+import pandas as pd
+import numpy as np
+import os
+import re # For regular expressions to parse filenames
+from sklearn.model_selection import train_test_split # For splitting data
+
+# --- Configuration ---
+MY_RECORDED_SIGNS_DIR = 'my_recorded_signs'
+PERSONAL_PROCESSED_CSV = 'wlasl_125_words_personal_processed.csv' # Changed to 125 words
+
+# --- Define your expected 125-word list (for validation, ensure only these glosses are included) ---
+# This list MUST match the glosses of the .npy files you have recorded
+EXPECTED_GLOSSES = [
+    "again",
+    "ambulance",
+    "and",
+    "angry",
+    "animal",
+    "apple",
+    "aunt",
+    "autum",
+    "bird",
+    "black",
+    "blue",
+    "boy",
+    "brother",
+    "brown",
+    "can",
+    "candy",
+    "car",
+    "cat",
+    "cereal",
+    "chair",
+    "child",
+    "cold",
+    "come",
+    "computer",
+    "cool",
+    "cow",
+    "cry",
+    "cup",
+    "deaf",
+    "dog",
+    "drink",
+    "drive",
+    "drive-there",
+    "eat",
+    "egg",
+    "father",
+    "feel",
+    "fish",
+    "freeze",
+    "friday",
+    "full",
+    "future",
+    "girl",
+    "give",
+    "go",
+    "gold",
+    "grandma",
+    "grandpa",
+    "green",
+    "grey",
+    "happy",
+    "hate",
+    "hello",
+    "horse",
+    "hot",
+    "hungry",
+    "hurt",
+    "juice",
+    "light",
+    "like",
+    "love",
+    "meet",
+    "milk",
+    "monday",
+    "mother",
+    "my",
+    "name",
+    "nice",
+    "now",
+    "o'clock",
+    "orange",
+    "parents",
+    "pet",
+    "pink",
+    "pizza",
+    "popcorn",
+    "purple",
+    "rain",
+    "red",
+    "sad",
+    "saturday",
+    "shower",
+    "siblings",
+    "silver",
+    "sister",
+    "sit",
+    "sleep",
+    "snow",
+    "sorry",
+    "soup",
+    "spring",
+    "stand",
+    "stay",
+    "summer",
+    "sun",
+    "sunday",
+    "sunrise",
+    "table",
+    "talk",
+    "tell",
+    "this",
+    "thursday",
+    "today",
+    "tomorrow",
+    "tuesday",
+    "uncle",
+    "understand",
+    "walk",
+    "warm",
+    "watch",
+    "water",
+    "weather",
+    "wednesday",
+    "what",
+    "white",
+    "who",
+    "why",
+    "wind",
+    "window",
+    "winter",
+    "year",
+    "yellow",
+    "yesterday",
+    "you",
+    "your",
+]
+EXPECTED_GLOSSES_SET = set(g.lower() for g in EXPECTED_GLOSSES)
+
+
+print(f"--- Generating CSV from Personal Recordings Only ---")
+
+personal_records = []
+if os.path.exists(MY_RECORDED_SIGNS_DIR):
+    print(f"Scanning for personal recordings in: {MY_RECORDED_SIGNS_DIR}")
+    for filename in os.listdir(MY_RECORDED_SIGNS_DIR):
+        if filename.lower().endswith('.npy'):
+            # Filename format: 'gloss_timestamp.npy' or similar
+            match = re.match(r'(.+?)_(\d+)\.npy', filename, re.IGNORECASE)
+            if match:
+                gloss = match.group(1).lower() # Extract gloss, ensure lowercase
+                video_id_from_timestamp = match.group(2) # Use timestamp as a unique ID
+
+                full_npy_path = os.path.join(MY_RECORDED_SIGNS_DIR, filename)
+
+                if gloss in EXPECTED_GLOSSES_SET:
+                    personal_records.append({
+                        'gloss': gloss,
+                        'video_id': f'personal_{video_id_from_timestamp}',
+                        'video_path': full_npy_path, # Path to the NPY file itself
+                        'signer_id': 'personal_user',
+                        'frame_start': 1,
+                        'frame_end': -1,
+                        'fps': 25,
+                        'bbox': [0,0,0,0],
+                        'split': 'temp' # Temporary split, will be re-assigned
+                    })
+                else:
+                    print(f"Warning: Personal recording '{filename}' has gloss '{gloss}' which is not in the EXPECTED_GLOSSES list. Skipping.")
+            else:
+                print(f"Warning: Could not parse filename for '{filename}'. Skipping.")
+else:
+    print(f"Error: Personal recordings directory '{MY_RECORDED_SIGNS_DIR}' not found. Exiting.")
+    exit()
+
+df_personal_raw = pd.DataFrame(personal_records)
+print(f"Found {len(df_personal_raw)} personal recordings for the {len(EXPECTED_GLOSSES)} expected glosses.")
+
+if df_personal_raw.empty:
+    print("No valid personal recordings found matching the expected glosses. Cannot create CSV.")
+    exit()
+
+# --- Assign Train/Val/Test Splits to the Personal Data ---
+# Group by gloss first to ensure all instances of a gloss are considered for splitting
+df_personal_final = pd.DataFrame()
+for gloss in EXPECTED_GLOSSES_SET:
+    gloss_df = df_personal_raw[df_personal_raw['gloss'] == gloss].copy()
+
+    if len(gloss_df) == 0:
+        print(f"Warning: No personal recordings found for expected gloss '{gloss}'. This gloss will be missing from the dataset.")
+        continue # Skip to next gloss
+
+    # Determine optimal split ratios based on the quantity of data you have recorded
+    # For a total dataset of a few hundred instances, 70/15/15 is a decent starting point.
+    # If you recorded many videos per word (e.g., >30 per word): train 80%, val 10%, test 10% might be better.
+    train_ratio = 0.70
+    val_ratio = 0.15
+    test_ratio = 0.15
+
+    # Ensure enough samples for all splits (min 3 for train_test_split stratify to work well)
+    if len(gloss_df) < 3:
+        print(f"Warning: Gloss '{gloss}' has only {len(gloss_df)} instances. Adding all to train. (Cannot split)")
+        gloss_df['split'] = 'train'
+        df_personal_final = pd.concat([df_personal_final, gloss_df])
+        continue
+
+    # Stratify by gloss (though not strictly necessary as we are looping by gloss)
+    # The 'stratify' argument in train_test_split ensures that the proportion of classes
+    # is roughly the same in the training, validation, and test sets.
+
+    # Split into train/test first
+    train_val_df, test_df = train_test_split(
+        gloss_df, test_size=test_ratio, random_state=42, stratify=gloss_df['gloss'] if len(gloss_df['gloss'].unique()) > 1 else None # Stratify only if multiple classes present, which is true here.
+    )
+    # Split train_val into train/val
+    # Adjust val_size ratio for the remaining train_val_df
+    remaining_val_ratio = val_ratio / (train_ratio + val_ratio)
+    train_df, val_df = train_test_split(
+        train_val_df, test_size=remaining_val_ratio, random_state=42, stratify=train_val_df['gloss'] if len(train_val_df['gloss'].unique()) > 1 else None
+    )
+
+    train_df['split'] = 'train'
+    val_df['split'] = 'val'
+    test_df['split'] = 'test'
+
+    df_personal_final = pd.concat([df_personal_final, train_df, val_df, test_df])
+
+# Re-map gloss_ids to be sequential (0 to N-1) for the final set of classes
+final_unique_glosses_in_dataset = df_personal_final['gloss'].unique()
+final_gloss_to_id = {g: i for i, g in enumerate(final_unique_glosses_in_dataset)}
+df_personal_final['gloss_id'] = df_personal_final['gloss'].map(final_gloss_to_id)
+
+print(f"\n--- Personal Dataset Summary ---")
+print(f"Total instances in final dataset: {len(df_personal_final)}")
+print(f"Unique glosses: {df_personal_final['gloss'].nunique()} (Out of {len(EXPECTED_GLOSSES)} requested)")
+print(f"Distribution of splits:\n{df_personal_final['split'].value_counts()}")
+
+df_personal_final.to_csv(PERSONAL_PROCESSED_CSV, index=False)
+print(f"\nPersonal dataset metadata saved to '{PERSONAL_PROCESSED_CSV}'")

ai_model/words/data_preprocessor.py

@@ -0,0 +1,232 @@
+import os
+import pandas as pd
+import numpy as np
+from sklearn.model_selection import train_test_split 
+from tensorflow.keras.utils import to_categorical 
+from tqdm import tqdm
+import random 
+
+# --- Configuration (match with previous steps) ---
+# IMPORTANT: This should now point to the CSV generated by create_personal_dataset_csv.py (your 125 words)
+PROCESSED_DATA_CSV = 'wlasl_125_words_personal_processed.csv'
+EXTRACTED_LANDMARKS_DIR = 'extracted_landmarks'
+PROCESSED_SEQUENCES_DIR = 'processed_sequences'
+
+SEQUENCE_LENGTH = 90 # Confirmed optimal from duration analysis
+EXPECTED_COORDS_PER_FRAME = 1662 
+
+NUM_AUGMENTATIONS_PER_TRAIN_VIDEO = 5 
+AUG_MAX_ROTATION_DEG = 10 
+AUG_MAX_SCALE_FACTOR = 0.1 
+AUG_MAX_JITTER_AMOUNT = 0.003 
+
+MAX_VIDEOS_FOR_TEST = None 
+
+
+def normalize_landmarks(landmarks_sequence):
+    """
+    Normalizes a sequence of landmarks (frames, coords) to be translation and scale invariant.
+    """
+    if landmarks_sequence.size == 0:
+        return landmarks_sequence.astype(np.float32)
+    if landmarks_sequence.ndim == 1:
+        input_is_single_frame = True
+        landmarks_sequence_2d = np.expand_dims(landmarks_sequence, axis=0)
+    else:
+        input_is_single_frame = False
+        landmarks_sequence_2d = landmarks_sequence
+    NUM_POSE_COORDS_SINGLE = 33 * 4
+    NUM_HAND_COORDS_SINGLE = 21 * 3
+    NUM_FACE_COORDS_SINGLE = 468 * 3
+    normalized_sequences = []
+    for frame_landmarks in landmarks_sequence_2d:
+        if np.all(frame_landmarks == 0):
+            normalized_sequences.append(np.zeros(EXPECTED_COORDS_PER_FRAME, dtype=np.float32))
+            continue 
+        pose_coords_flat = frame_landmarks[0 : NUM_POSE_COORDS_SINGLE]
+        left_hand_coords_flat = frame_landmarks[NUM_POSE_COORDS_SINGLE : NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE]
+        right_hand_coords_flat = frame_landmarks[NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE : NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE]
+        face_coords_flat = frame_landmarks[NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE : ]
+        all_parts_data = [
+            (pose_coords_flat, 4, [0.0] * NUM_POSE_COORDS_SINGLE),
+            (left_hand_coords_flat, 3, [0.0] * NUM_HAND_COORDS_SINGLE),
+            (right_hand_coords_flat, 3, [0.0] * NUM_HAND_COORDS_SINGLE),
+            (face_coords_flat, 3, [0.0] * NUM_FACE_COORDS_SINGLE)
+        ]
+        normalized_frame_parts = []
+        for flat_lms, coords_per_lm, original_padded_list_template in all_parts_data:
+            if np.all(flat_lms == 0):
+                normalized_frame_parts.append(np.array(original_padded_list_template, dtype=np.float32))
+                continue
+            lms_array = flat_lms.reshape(-1, coords_per_lm)
+            coords_for_mean = lms_array[:, :3] if coords_per_lm == 4 else lms_array
+            if np.all(coords_for_mean == 0):
+                normalized_frame_parts.append(np.array(original_padded_list_template, dtype=np.float32))
+                continue
+            mean_coords = np.mean(coords_for_mean, axis=0)
+            translated_lms = lms_array.copy()
+            translated_lms[:, :3] -= mean_coords
+            scale_factor = np.max(np.linalg.norm(translated_lms[:, :3], axis=1))
+            if scale_factor > 1e-6:
+                translated_lms[:, :3] /= scale_factor
+            normalized_frame_parts.append(translated_lms.flatten())
+        combined_normalized_frame = np.concatenate(normalized_frame_parts).astype(np.float32)
+        if len(combined_normalized_frame) < EXPECTED_COORDS_PER_FRAME:
+            combined_normalized_frame = np.pad(combined_normalized_frame, (0, EXPECTED_COORDS_PER_FRAME - len(combined_normalized_frame)), 'constant', constant_values=0.0)
+        elif len(combined_normalized_frame) > EXPECTED_COORDS_PER_FRAME:
+            combined_normalized_frame = combined_normalized_frame[:EXPECTED_COORDS_PER_FRAME]
+        normalized_sequences.append(combined_normalized_frame)
+    result_array = np.array(normalized_sequences, dtype=np.float32)
+    if input_is_single_frame:
+        return result_array[0] 
+    else:
+        return result_array 
+
+
+def augment_sequence(sequence, max_rotation_deg, max_scale_factor, max_jitter_amount):
+    """
+    Applies random geometric augmentations to a landmark sequence.
+    """
+    augmented_sequence = sequence.copy()
+
+    NUM_POSE_COORDS_SINGLE = 33 * 4
+    NUM_HAND_COORDS_SINGLE = 21 * 3
+    NUM_FACE_COORDS_SINGLE = 468 * 3
+
+    angle_rad = np.deg2rad(np.random.uniform(-max_rotation_deg, max_rotation_deg))
+    cos_val = np.cos(angle_rad)
+    sin_val = np.sin(angle_rad)
+    scale = np.random.uniform(1.0 - max_scale_factor, 1.0 + max_scale_factor)
+
+    temp_sequence = []
+    for frame_lms in augmented_sequence:
+        processed_parts = []
+        current_idx = 0
+
+        pose_lms = frame_lms[current_idx : current_idx + NUM_POSE_COORDS_SINGLE].reshape(-1, 4)
+        current_idx += NUM_POSE_COORDS_SINGLE
+        if pose_lms.size > 0 and not np.all(pose_lms[:,:3] == 0):
+            rotated_xy_pose = pose_lms[:, :2].dot(np.array([[cos_val, -sin_val], [sin_val, cos_val]]))
+            pose_lms[:, :2] = rotated_xy_pose
+            pose_lms[:, :3] *= scale
+        processed_parts.append(pose_lms.flatten())
+
+        left_hand_lms = frame_lms[current_idx : current_idx + NUM_HAND_COORDS_SINGLE].reshape(-1, 3)
+        current_idx += NUM_HAND_COORDS_SINGLE
+        if left_hand_lms.size > 0 and not np.all(left_hand_lms == 0):
+            rotated_xy_lh = left_hand_lms[:, :2].dot(np.array([[cos_val, -sin_val], [sin_val, cos_val]]))
+            left_hand_lms[:, :2] = rotated_xy_lh
+            left_hand_lms *= scale
+        processed_parts.append(left_hand_lms.flatten())
+
+        right_hand_lms = frame_lms[current_idx : current_idx + NUM_HAND_COORDS_SINGLE].reshape(-1, 3)
+        current_idx += NUM_HAND_COORDS_SINGLE
+        if right_hand_lms.size > 0 and not np.all(right_hand_lms == 0):
+            rotated_xy_rh = right_hand_lms[:, :2].dot(np.array([[cos_val, -sin_val], [sin_val, cos_val]]))
+            right_hand_lms[:, :2] = rotated_xy_rh
+            right_hand_lms *= scale
+        processed_parts.append(right_hand_lms.flatten())
+
+        face_lms = frame_lms[current_idx : current_idx + NUM_FACE_COORDS_SINGLE].reshape(-1, 3)
+        if face_lms.size > 0 and not np.all(face_lms == 0):
+            rotated_xy_face = face_lms[:, :2].dot(np.array([[cos_val, -sin_val], [sin_val, cos_val]]))
+            face_lms[:, :2] = rotated_xy_face
+            face_lms *= scale
+        processed_parts.append(face_lms.flatten())
+        
+        temp_sequence.append(np.concatenate(processed_parts))
+    
+    augmented_sequence = np.array(temp_sequence, dtype=np.float32)
+    jitter = np.random.normal(loc=0, scale=max_jitter_amount, size=augmented_sequence.shape)
+    augmented_sequence += jitter
+
+    return augmented_sequence
+
+
+def _process_and_pad_sequence(raw_landmarks, output_dir, video_id, gloss, split):
+    """Helper to normalize, pad/truncate and save a single sequence."""
+    normalized_landmarks = normalize_landmarks(raw_landmarks)
+    if normalized_landmarks is None or normalized_landmarks.size == 0:
+        return None
+    if normalized_landmarks.shape[0] < SEQUENCE_LENGTH:
+        padding = np.zeros((SEQUENCE_LENGTH - normalized_landmarks.shape[0], EXPECTED_COORDS_PER_FRAME), dtype=np.float32)
+        processed_sequence = np.vstack((normalized_landmarks, padding))
+    elif normalized_landmarks.shape[0] > SEQUENCE_LENGTH:
+        processed_sequence = normalized_landmarks[:SEQUENCE_LENGTH, :]
+    else:
+        processed_sequence = normalized_landmarks
+    if processed_sequence.shape != (SEQUENCE_LENGTH, EXPECTED_COORDS_PER_FRAME):
+        print(f"Error: Processed sequence for {video_id} has incorrect shape: {processed_sequence.shape}. Expected: ({SEQUENCE_LENGTH}, {EXPECTED_COORDS_PER_FRAME}). Skipping.")
+        return None
+    split_output_dir = os.path.join(output_dir, split if pd.notna(split) else 'unknown_split')
+    os.makedirs(split_output_dir, exist_ok=True)
+    output_filepath = os.path.join(split_output_dir, f"{video_id}.npy")
+    np.save(output_filepath, processed_sequence)
+    return {'video_id': video_id, 'gloss': gloss, 'split': split, 'processed_path': output_filepath}
+
+
+if __name__ == "__main__":
+    os.makedirs(PROCESSED_SEQUENCES_DIR, exist_ok=True)
+    if not os.path.exists(PROCESSED_DATA_CSV):
+        print(f"Error: Processed data CSV not found at {PROCESSED_DATA_CSV}.")
+        exit()
+    df_base = pd.read_csv(PROCESSED_DATA_CSV)
+    print(f"Loaded {len(df_base)} video entries from {PROCESSED_DATA_CSV}")
+    unique_glosses = df_base['gloss'].unique()
+    gloss_to_id = {gloss: i for i, gloss in enumerate(unique_glosses)}
+    id_to_gloss = {i: gloss for i, gloss in enumerate(unique_glosses)}
+    print(f"Created mapping for {len(unique_glosses)} unique glosses.")
+    processed_data_records = []
+    MAX_VIDEOS_FOR_TEST = None 
+    df_to_process = df_base.head(MAX_VIDEOS_FOR_TEST) if MAX_VIDEOS_FOR_TEST is not None else df_base 
+    print(f"\nProcessing {len(df_to_process)} videos for feature engineering, sequence preparation, and augmentation...")
+    for index, row in tqdm(df_to_process.iterrows(), total=len(df_to_process), desc="Processing Sequences"):
+        video_id = row['video_id']
+        gloss = row['gloss']
+        split = row['split']
+        raw_landmarks_path = os.path.join(EXTRACTED_LANDMARKS_DIR, f"{video_id}.npy")
+        if not os.path.exists(raw_landmarks_path):
+            print(f"Warning: Raw landmark file not found for {video_id}. Skipping.")
+            continue
+        raw_landmarks = np.load(raw_landmarks_path)
+        original_record = _process_and_pad_sequence(raw_landmarks, PROCESSED_SEQUENCES_DIR, video_id, gloss, split)
+        if original_record:
+            original_record['gloss_id'] = gloss_to_id[original_record['gloss']]
+            processed_data_records.append(original_record)
+        if split == 'train':
+            for aug_idx in range(NUM_AUGMENTATIONS_PER_TRAIN_VIDEO):
+                augmented_raw_landmarks = augment_sequence(
+                    raw_landmarks.copy(), 
+                    AUG_MAX_ROTATION_DEG, 
+                    AUG_MAX_SCALE_FACTOR, 
+                    AUG_MAX_JITTER_AMOUNT
+                )
+                aug_video_id = f"{video_id}_aug{aug_idx}"
+                aug_record = _process_and_pad_sequence(augmented_raw_landmarks, PROCESSED_SEQUENCES_DIR, aug_video_id, gloss, split)
+                if aug_record:
+                    aug_record['gloss_id'] = gloss_to_id[aug_record['gloss']]
+                    processed_data_records.append(aug_record)
+    final_processed_df = pd.DataFrame(processed_data_records)
+    print(f"\n--- Feature Engineering and Data Preparation Summary ---")
+    print(f"Total processed sequences saved: {len(final_processed_df)}")
+    print(f"Processed sequences saved to: {PROCESSED_SEQUENCES_DIR}")
+    print(f"Final DataFrame for training/validation/test (including augmentations):\n{final_processed_df['split'].value_counts()}")
+    final_processed_df.to_csv('wlasl_125_words_personal_final_processed_data_augmented_seq90.csv', index=False)
+    print("\nFinal processed data metadata saved to 'wlasl_125_words_personal_final_processed_data_augmented_seq90.csv'")
+    if not final_processed_df.empty:
+        original_entries = final_processed_df[~final_processed_df['video_id'].astype(str).str.contains('_aug', na=False)]
+        augmented_entries = final_processed_df[final_processed_df['video_id'].astype(str).str.contains('_aug', na=False)]
+        original_example_entry = original_entries.iloc[0] if not original_entries.empty else None
+        augmented_example_entry = augmented_entries.iloc[0] if not augmented_entries.empty else None
+        if original_example_entry is not None:
+            loaded_original = np.load(original_example_entry['processed_path'])
+            print(f"\nSuccessfully loaded example ORIGINAL sequence: {original_example_entry['processed_path']}")
+            print(f"Shape: {loaded_original.shape}. Expected: ({SEQUENCE_LENGTH}, {EXPECTED_COORDS_PER_FRAME})")
+            print(f"First few values of first frame:\n{loaded_original[0, :10]}")
+        if augmented_example_entry is not None:
+            loaded_augmented = np.load(augmented_example_entry['processed_path'])
+            print(f"\nSuccessfully loaded example AUGMENTED sequence: {augmented_example_entry['processed_path']}")
+            print(f"Shape: {loaded_augmented.shape}. Expected: ({SEQUENCE_LENGTH}, {EXPECTED_COORDS_PER_FRAME})")
+            print(f"First few values of first frame:\n{loaded_augmented[0, :10]}")
+    else:
+        print("\nNo processed sequences to test loading.")

ai_model/words/landmark_extractor.py

@@ -0,0 +1,183 @@
+import cv2
+import mediapipe as mp
+import pandas as pd
+import numpy as np
+import os
+from tqdm import tqdm # For progress bars
+
+# --- Configuration ---
+# Path to your personal 125-words CSV (Generated by create_personal_dataset_csv.py)
+PROCESSED_DATA_CSV = 'wlasl_125_words_personal_processed.csv'
+# Directory to save extracted landmark data
+OUTPUT_LANDMARKS_DIR = 'extracted_landmarks'
+# Whether to visualize the landmarks during extraction (set to False for faster processing)
+VISUALIZE_LANDMARKS = False
+# Max videos to process for quick test (set to None to process all from CSV)
+MAX_VIDEOS_FOR_TEST = None
+
+
+# --- Define the expected number of coordinates for each type of landmark ---
+NUM_POSE_COORDS = 33 * 4   # 33 landmarks * (x, y, z, visibility)
+NUM_HAND_COORDS = 21 * 3   # 21 landmarks * (x, y, z)
+NUM_FACE_COORDS = 468 * 3 # 468 landmarks * (x, y, z)
+
+
+def extract_landmarks_from_video(video_path, mp_holistic_instance):
+    is_npy_file = video_path.lower().endswith('.npy')
+
+    if is_npy_file:
+        try:
+            landmarks_array = np.load(video_path)
+            if landmarks_array.ndim == 2 and landmarks_array.shape[1] == (NUM_POSE_COORDS + 2*NUM_HAND_COORDS + NUM_FACE_COORDS):
+                return landmarks_array
+            else:
+                print(f"Warning: NPY file {video_path} has unexpected shape {landmarks_array.shape}. Skipping.")
+                return None
+        except Exception as e:
+            print(f"Error loading NPY file {video_path}: {e}. Skipping.")
+            return None
+
+    cap = cv2.VideoCapture(video_path)
+    if not cap.isOpened():
+        print(f"Error: Could not open video {video_path}")
+        return None
+
+    frame_landmarks_list = []
+
+    while cap.isOpened():
+        ret, frame = cap.read()
+        if not ret:
+            break
+
+        image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+        image.flags.writeable = False
+        results = mp_holistic_instance.process(image)
+        if VISUALIZE_LANDMARKS:
+            image.flags.writeable = True
+            frame_copy = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+            mp.solutions.drawing_utils.draw_landmarks(frame_copy, results.pose_landmarks, mp.solutions.holistic.POSE_CONNECTIONS)
+            mp.solutions.drawing_utils.draw_landmarks(frame_copy, results.left_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+            mp.solutions.drawing_utils.draw_landmarks(frame_copy, results.right_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+            mp.solutions.drawing_utils.draw_landmarks(frame_copy, results.face_landmarks, mp.solutions.holistic.FACEMESH_CONTOURS)
+            cv2.imshow('MediaPipe Landmarks', frame_copy)
+            if cv2.waitKey(1) & 0xFF == 27:
+                break
+
+        current_frame_raw_landmarks_flat = np.zeros(NUM_POSE_COORDS + NUM_HAND_COORDS + NUM_HAND_COORDS + NUM_FACE_COORDS, dtype=np.float32)
+        current_idx = 0
+
+        if results.pose_landmarks:
+            pose_lms_flat = []
+            for landmark in results.pose_landmarks.landmark:
+                pose_lms_flat.extend([landmark.x, landmark.y, landmark.z, landmark.visibility])
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(pose_lms_flat)] = pose_lms_flat
+        current_idx += NUM_POSE_COORDS
+
+        if results.left_hand_landmarks:
+            lh_lms_flat = []
+            for landmark in results.left_hand_landmarks.landmark:
+                lh_lms_flat.extend([landmark.x, landmark.y, landmark.z])
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(lh_lms_flat)] = lh_lms_flat
+        current_idx += NUM_HAND_COORDS
+
+        if results.right_hand_landmarks:
+            rh_lms_flat = []
+            for landmark in results.right_hand_landmarks.landmark:
+                rh_lms_flat.extend([landmark.x, landmark.y, landmark.z])
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(rh_lms_flat)] = rh_lms_flat
+        current_idx += NUM_HAND_COORDS
+
+        if results.face_landmarks:
+            face_lms_flat = []
+            for landmark in results.face_landmarks.landmark:
+                face_lms_flat.extend([landmark.x, landmark.y, landmark.z])
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(face_lms_flat)] = face_lms_flat
+
+        frame_landmarks_list.append(current_frame_raw_landmarks_flat)
+
+    cap.release()
+    if VISUALIZE_LANDMARKS:
+        cv2.destroyAllWindows()
+
+    if frame_landmarks_list:
+        return np.array(frame_landmarks_list, dtype=np.float32)
+    else:
+        return None
+
+
+if __name__ == "__main__":
+    os.makedirs(OUTPUT_LANDMARKS_DIR, exist_ok=True)
+
+    if not os.path.exists(PROCESSED_DATA_CSV):
+        print(f"Error: Processed data CSV not found at {PROCESSED_DATA_CSV}. Please run create_personal_dataset_csv.py first.")
+        exit()
+
+    df_base = pd.read_csv(PROCESSED_DATA_CSV)
+    print(f"Loaded {len(df_base)} video entries from {PROCESSED_DATA_CSV}")
+
+    mp_holistic = mp.solutions.holistic.Holistic(
+        static_image_mode=False,
+        model_complexity=1,
+        min_detection_confidence=0.5,
+        min_tracking_confidence=0.5
+    )
+
+    processed_count = 0
+    skipped_count = 0
+
+    for index, row in tqdm(df_base.iterrows(), total=len(df_base), desc="Extracting Landmarks"):
+        if MAX_VIDEOS_FOR_TEST is not None and processed_count >= MAX_VIDEOS_FOR_TEST:
+            print(f"Stopping after {MAX_VIDEOS_FOR_TEST} videos for testing.")
+            break
+
+        video_path = row['video_path']
+        video_id = row['video_id']
+        gloss = row['gloss']
+
+        output_npy_filename = f"{video_id}.npy"
+        output_npy_path = os.path.join(OUTPUT_LANDMARKS_DIR, output_npy_filename)
+
+        if os.path.exists(output_npy_path):
+            processed_count += 1
+            continue
+
+        landmarks_data = extract_landmarks_from_video(video_path, mp_holistic)
+
+        if landmarks_data is not None and landmarks_data.shape[0] > 0:
+            np.save(output_npy_path, landmarks_data)
+            processed_count += 1
+        else:
+            print(f"Warning: No valid landmarks extracted for {video_path}. Skipping.")
+            skipped_count += 1
+
+    mp_holistic.close()
+    print(f"\n--- Landmark Extraction Summary ---")
+    print(f"Total videos processed: {processed_count}")
+    print(f"Total videos skipped (e.g., no landmarks, already processed): {skipped_count}")
+    print(f"Landmark data saved to: {OUTPUT_LANDMARKS_DIR}")
+
+    # --- Test a saved landmark file ---
+    if processed_count > 0:
+        example_video_id = None
+        for _, row in df_base.iterrows():
+            temp_output_npy_path = os.path.join(OUTPUT_LANDMARKS_DIR, f"{row['video_id']}.npy")
+            if os.path.exists(temp_output_npy_path):
+                example_video_id = row['video_id']
+                break
+
+        if example_video_id:
+            example_npy_path = os.path.join(OUTPUT_LANDMARKS_DIR, f"{example_video_id}.npy")
+            loaded_landmarks = np.load(example_npy_path)
+            print(f"\nSuccessfully loaded example landmark file: {example_npy_path}")
+            print(f"Shape of loaded landmarks (frames, total_coords): {loaded_landmarks.shape}")
+            print(f"First few values of the first frame's landmarks:\n{loaded_landmarks[0, :10]}")
+
+            expected_total_coords = NUM_POSE_COORDS + NUM_HAND_COORDS + NUM_HAND_COORDS + NUM_FACE_COORDS
+            print(f"Expected total coordinates per frame: {expected_total_coords}")
+            if loaded_landmarks.shape[1] != expected_total_coords:
+                print(f"WARNING: Loaded landmark dimension {loaded_landmarks.shape[1]} does not match expected {expected_total_coords}.")
+
+        else:
+            print("\nNo videos were processed to test landmark saving, or no .npy files were found in the output directory.")
+    else:
+        print("\nNo videos were processed to test landmark saving.")

ai_model/words/list_dataset_videos.py

@@ -0,0 +1,57 @@
+import pandas as pd
+import os
+
+# --- Configuration ---
+PROCESSED_DATA_CSV = 'wlasl_20_words_processed.csv' 
+
+# Base directory where your script is running from
+BASE_DIR = os.getcwd() 
+
+# --- NEW: Display Limit per Sign ---
+MAX_VIDEOS_PER_SIGN_TO_DISPLAY = 5 
+
+
+print(f"--- Listing Video Files and Sign Names for Dataset: {PROCESSED_DATA_CSV} (20 Words) ---") 
+
+if not os.path.exists(PROCESSED_DATA_CSV):
+    print(f"Error: Processed data CSV not found at {PROCESSED_DATA_CSV}.")
+    print("Please ensure wlasl_parser.py has been run for your 20-word list.") 
+    exit()
+
+try:
+    df_videos = pd.read_csv(PROCESSED_DATA_CSV)
+    print(f"Loaded {len(df_videos)} video entries.")
+
+    print(f"\nExact video paths and their corresponding sign names (glosses) (showing max {MAX_VIDEOS_PER_SIGN_TO_DISPLAY} per sign):")
+    
+    videos_displayed_per_sign = {} # To track how many videos we've shown for each sign
+    total_found_on_disk = 0
+
+    # Sort DataFrame by gloss to group signs together in output
+    df_videos_sorted = df_videos.sort_values(by='gloss').reset_index(drop=True)
+
+    for index, row in df_videos_sorted.iterrows(): # Iterate through the sorted DataFrame
+        gloss = row['gloss']
+        relative_video_path = row['video_path']
+        full_video_path = os.path.join(BASE_DIR, relative_video_path)
+        
+        # Check if we've already displayed the max for this sign
+        if videos_displayed_per_sign.get(gloss, 0) >= MAX_VIDEOS_PER_SIGN_TO_DISPLAY:
+            continue # Skip to the next video if limit reached for this gloss
+        
+        if os.path.exists(full_video_path):
+            print(f"  Sign: {gloss.upper():<10} | Path: {full_video_path}")
+            videos_displayed_per_sign[gloss] = videos_displayed_per_sign.get(gloss, 0) + 1
+            total_found_on_disk += 1
+        else:
+            print(f"  [NOT FOUND ON DISK] Sign: {gloss.upper():<10} | Path: {full_video_path}")
+            # We still count this towards the limit for the gloss, even if not found
+            videos_displayed_per_sign[gloss] = videos_displayed_per_sign.get(gloss, 0) + 1 
+            
+    print(f"\nTotal video files found on disk and displayed: {total_found_on_disk}")
+    # Note: The total_found_on_disk here reflects only those *displayed*, not all available if limits applied.
+    print(f"Total available instances in CSV: {len(df_videos)}")
+
+
+except Exception as e:
+    print(f"An error occurred: {e}")

ai_model/words/list_glosses.py

@@ -0,0 +1,62 @@
+import os
+import re
+from collections import defaultdict
+
+# --- Configuration ---
+# The directory where your .npy recordings are stored.
+RECORDINGS_DIR = 'my_recorded_signs'
+
+def get_unique_glosses_from_recordings(directory):
+    """
+    Scans a directory for .npy files and extracts the unique glosses
+    (words) from their filenames.
+    
+    Args:
+        directory (str): The path to the directory containing recordings.
+        
+    Returns:
+        A defaultdict with glosses as keys and a count of recordings as values.
+    """
+    gloss_counts = defaultdict(int)
+    
+    if not os.path.isdir(directory):
+        print(f"Error: Directory not found at {directory}. Please check the path.")
+        return gloss_counts
+    
+    for filename in os.listdir(directory):
+        # We need a regex pattern to handle different timestamps and file extensions
+        # Pattern: (word)_(timestamp).npy
+        match = re.match(r'(.+?)_\d+\.npy', filename, re.IGNORECASE)
+        
+        if match:
+            # Extract the gloss (the first group in the regex) and convert to lowercase
+            gloss = match.group(1).lower()
+            gloss_counts[gloss] += 1
+            
+    return gloss_counts
+
+if __name__ == "__main__":
+    print(f"--- Scanning for Unique Signs in '{RECORDINGS_DIR}' ---")
+    
+    unique_glosses_with_counts = get_unique_glosses_from_recordings(RECORDINGS_DIR)
+    
+    if not unique_glosses_with_counts:
+        print("No valid recordings found. Make sure your .npy files are in the correct directory.")
+    else:
+        # Sort the glosses alphabetically for a clean, repeatable output
+        sorted_glosses = sorted(unique_glosses_with_counts.keys())
+        
+        print(f"\nFound {len(sorted_glosses)} unique glosses:")
+        print("---------------------------------")
+        for gloss in sorted_glosses:
+            count = unique_glosses_with_counts[gloss]
+            print(f"- {gloss:<15}: {count} recordings")
+        print("---------------------------------")
+        
+        # You can now easily copy this list into your create_personal_dataset_csv.py script.
+        # Here is the list format for easy copying:
+        print("\nReady to copy to your EXPECTED_GLOSSES list:")
+        print("EXPECTED_GLOSSES = [")
+        for i, gloss in enumerate(sorted_glosses):
+            print(f"    \"{gloss}\",")
+        print("]")

ai_model/words/missing.txt

@@ -0,0 +1,9103 @@
+65225
+68011
+68208
+68012
+70212
+70266
+07085
+07086
+07087
+07088
+07089
+07090
+07091
+07092
+07093
+07094
+07095
+07096
+07097
+07098
+07071
+07072
+07073
+67424
+07075
+07076
+07077
+07078
+07079
+07080
+07081
+07082
+07083
+07084
+70173
+68538
+68042
+68660
+68041
+17725
+17726
+17727
+17728
+17729
+17730
+17731
+17732
+17714
+17715
+17716
+17717
+17718
+67594
+17719
+12306
+68028
+69054
+12329
+12330
+12331
+12332
+12333
+12335
+12336
+12337
+67519
+12321
+12322
+12323
+12324
+05724
+70348
+68007
+05744
+05746
+05747
+05748
+05735
+05736
+05737
+09847
+70230
+68580
+70263
+68019
+09865
+09866
+09867
+09853
+09855
+09856
+67483
+09857
+09858
+09859
+09860
+09861
+09862
+09863
+24857
+68292
+24965
+24969
+24970
+24971
+24972
+24948
+67715
+24950
+24951
+11305
+68870
+68024
+11327
+11328
+11329
+11312
+11314
+11315
+11316
+11317
+11318
+11319
+11320
+11321
+11322
+11323
+11324
+11325
+11326
+63219
+68890
+68183
+63239
+63240
+63241
+63233
+63234
+66780
+67066
+63238
+08909
+68018
+70326
+68790
+08928
+08922
+08923
+67468
+67470
+08926
+08927
+70332
+68592
+13637
+13638
+13639
+67535
+13643
+13644
+13645
+14855
+70015
+68586
+68033
+14899
+14900
+14901
+14889
+14891
+14895
+14896
+14898
+65717
+70234
+68048
+21885
+21886
+21887
+21888
+21889
+21875
+21876
+67660
+67659
+21884
+27194
+27219
+27220
+27210
+27211
+27212
+67757
+27218
+38482
+68110
+68406
+38539
+38540
+38541
+38542
+38531
+67934
+38536
+38538
+57919
+57950
+57952
+67299
+57944
+57945
+62152
+68852
+68918
+68177
+62173
+62174
+62165
+62166
+67036
+62171
+62172
+64201
+68416
+68190
+70306
+68992
+64223
+64215
+64216
+67012
+64217
+64221
+64222
+64275
+68192
+64290
+64294
+64295
+64296
+64297
+64298
+64299
+64289
+01912
+68001
+68720
+02002
+01989
+01990
+01995
+67356
+01996
+01997
+01998
+01999
+02000
+06455
+68230
+70244
+06475
+06480
+67414
+06481
+06482
+06483
+06484
+06485
+70271
+13214
+13205
+13206
+13207
+21933
+70361
+68050
+68674
+68688
+21946
+67663
+21952
+21953
+21954
+21955
+21956
+28074
+70270
+28113
+28114
+28118
+28119
+28120
+28121
+28122
+28123
+28124
+68093
+70299
+33271
+33277
+33278
+33279
+33280
+33281
+33282
+33283
+33284
+34822
+70308
+34828
+34829
+67873
+34833
+34834
+34835
+34836
+34837
+34838
+36927
+36945
+36934
+36938
+67908
+36939
+36940
+36941
+36942
+36944
+38982
+70345
+68694
+68770
+68114
+39001
+39002
+39003
+67937
+39004
+39005
+38996
+40114
+70310
+70249
+68122
+40120
+40122
+40123
+67949
+40124
+40126
+40128
+56552
+70323
+68972
+56573
+56574
+56576
+56577
+56578
+56560
+56564
+67275
+56566
+56567
+56568
+56569
+56571
+69092
+57645
+57646
+67292
+57636
+57637
+57642
+57643
+62944
+68330
+62982
+62984
+62971
+67057
+62986
+62973
+67058
+63662
+68187
+68764
+63670
+63671
+67078
+63675
+63676
+63677
+63678
+05628
+05643
+05635
+67400
+05639
+05641
+05642
+06822
+68274
+68010
+06836
+06837
+06838
+06840
+06841
+67420
+06842
+06843
+06844
+07383
+07403
+07395
+07396
+67431
+07401
+07402
+08950
+08951
+08952
+08953
+08939
+67465
+08945
+08946
+08947
+08948
+08949
+17076
+68632
+68035
+17088
+17092
+67578
+17093
+17095
+17096
+20976
+69016
+20984
+20985
+67642
+20988
+20989
+20990
+20991
+22109
+22129
+22118
+22119
+67667
+22124
+22125
+22126
+22127
+22128
+25318
+70176
+68528
+25337
+25338
+25327
+25328
+67722
+25332
+25333
+26688
+68396
+68068
+26736
+26716
+67745
+26721
+26722
+26723
+26724
+26726
+70016
+68350
+68070
+26977
+26978
+67750
+26981
+26982
+26983
+26984
+26985
+26986
+31746
+31765
+31766
+31754
+31757
+31758
+67817
+31759
+31762
+31763
+31764
+32146
+68084
+68085
+68904
+32161
+32162
+67827
+32164
+32165
+32166
+32319
+32335
+32336
+32327
+32328
+67829
+32329
+32334
+34685
+68794
+34739
+34741
+67872
+34743
+34744
+34745
+51054
+68928
+51077
+67188
+51069
+51070
+51071
+51072
+55356
+68428
+68162
+68624
+55367
+55370
+55371
+55372
+55373
+55374
+56835
+56851
+56845
+67278
+67279
+56849
+56850
+63191
+68622
+68182
+70245
+63206
+63209
+63210
+63211
+63212
+63213
+64082
+68189
+70296
+70132
+64089
+64092
+64093
+67009
+67010
+64094
+64095
+64096
+00618
+00630
+67342
+00635
+00636
+00638
+02997
+68003
+68576
+70309
+03004
+03006
+03007
+67367
+06326
+70107
+06336
+06338
+06339
+67411
+06340
+06341
+06342
+09945
+70379
+09968
+09969
+65332
+09961
+09962
+11752
+68202
+68027
+11771
+11774
+11775
+11776
+67514
+11777
+11778
+11779
+13309
+13330
+13331
+67531
+67530
+13335
+13336
+13681
+13700
+13701
+13704
+67536
+13705
+13706
+13707
+13708
+13709
+14621
+68032
+70152
+68988
+69048
+14626
+14628
+14629
+14630
+67550
+14631
+14632
+14669
+14677
+14678
+14679
+14682
+14683
+14684
+17007
+68842
+70049
+17018
+17022
+67579
+17023
+17024
+17025
+18316
+68520
+68044
+69086
+18326
+18328
+18331
+18332
+67610
+67611
+18333
+18334
+19255
+68344
+68046
+70051
+19262
+19263
+19265
+19266
+67624
+19267
+19268
+68053
+68692
+68054
+70376
+22956
+22959
+22961
+22962
+22963
+22964
+22965
+22966
+24655
+24656
+24657
+24658
+24659
+24642
+24643
+24645
+24652
+68086
+69068
+32251
+32252
+67826
+32258
+32259
+35506
+68099
+68974
+35514
+35515
+67886
+35519
+35520
+35521
+35522
+42827
+68132
+68398
+70246
+42835
+42838
+67990
+42840
+42841
+42842
+42953
+42974
+42976
+42963
+42964
+42971
+42972
+68288
+68133
+43172
+43175
+67081
+43176
+43177
+43178
+49577
+68218
+68510
+70207
+68145
+49601
+49603
+49604
+67170
+49605
+50036
+50042
+50043
+67177
+50047
+50048
+50049
+50050
+50051
+51206
+70355
+51228
+51229
+51230
+51234
+58488
+68294
+70356
+68171
+58504
+58505
+58506
+67309
+58507
+58500
+58501
+62241
+68178
+68540
+62249
+62253
+67038
+62254
+62256
+62257
+62258
+63769
+68924
+63804
+63805
+63794
+67003
+63801
+63802
+63803
+01382
+01390
+67350
+01392
+01393
+01394
+01395
+01396
+01397
+05227
+68612
+05236
+05237
+67392
+05241
+06355
+68338
+70357
+70359
+06366
+06367
+06368
+67413
+06369
+06370
+06361
+06362
+07957
+70242
+69028
+07964
+07967
+07968
+07969
+67446
+07971
+07972
+08421
+68368
+68016
+08427
+08428
+08432
+08433
+67451
+08434
+08435
+08436
+10146
+10161
+10162
+10164
+67489
+67490
+10156
+10160
+10888
+68374
+10896
+10899
+10900
+10901
+67501
+10902
+10903
+13154
+70030
+68029
+69004
+13159
+13162
+67527
+13164
+13165
+13166
+15031
+70119
+15036
+15039
+15040
+67555
+67556
+15041
+15042
+70029
+23772
+23773
+23777
+67697
+23778
+23779
+23780
+28187
+70295
+68954
+28206
+28207
+28208
+67776
+28212
+28213
+30830
+68320
+68079
+30847
+30848
+30837
+30838
+67802
+30841
+30842
+30843
+68636
+70325
+68089
+32951
+67840
+32956
+32957
+68810
+35457
+35460
+35461
+67890
+35462
+35463
+35464
+35465
+35466
+37879
+70237
+68544
+37885
+37887
+37888
+37889
+37890
+67925
+37891
+37892
+37893
+40816
+68364
+68125
+40838
+40839
+40842
+40843
+67958
+40844
+40845
+40846
+41008
+68446
+70211
+68127
+41031
+41033
+41034
+67961
+41035
+41036
+45252
+70378
+45266
+45268
+45269
+45270
+67111
+45271
+45272
+45432
+68137
+68726
+70247
+45439
+45440
+45441
+67114
+45442
+45437
+68142
+48122
+48123
+67143
+48111
+48112
+67142
+48117
+48120
+68856
+49179
+49180
+67159
+67161
+49184
+49185
+49186
+49187
+53258
+68336
+70335
+53272
+53276
+53277
+67227
+53278
+57273
+57279
+57280
+57281
+67287
+57288
+57289
+57290
+58359
+70026
+58364
+58367
+58368
+67306
+58369
+61804
+68248
+61811
+61813
+61814
+61815
+61816
+61817
+61818
+62077
+62104
+62105
+67034
+62109
+62110
+62111
+62112
+62479
+68179
+70069
+68722
+62504
+62507
+62508
+67044
+62509
+63325
+70333
+68766
+63330
+63331
+63333
+63334
+67068
+63335
+63336
+64260
+68191
+68964
+70240
+64267
+64268
+64269
+67013
+64270
+64265
+04614
+68956
+04621
+04622
+04626
+04627
+04628
+04629
+04630
+68626
+05092
+05093
+67389
+05099
+05101
+07929
+68594
+07939
+07941
+67445
+07942
+09431
+09530
+09531
+09534
+67477
+09535
+09536
+09537
+10189
+67491
+10193
+10194
+10195
+10196
+10965
+68376
+68021
+68900
+10974
+10975
+67502
+10971
+14172
+14178
+14179
+67545
+14185
+14186
+14187
+16190
+68034
+68666
+16196
+16197
+16199
+16200
+16201
+16202
+17317
+70264
+68038
+17329
+17331
+17332
+17333
+67583
+17334
+17335
+25674
+70241
+69066
+25691
+25692
+25693
+25694
+67729
+25688
+25690
+26122
+68424
+68064
+67737
+26167
+26169
+26170
+26172
+26757
+68069
+70221
+68984
+69088
+26775
+67746
+26776
+26777
+26778
+26770
+26832
+26838
+26841
+67747
+26842
+26843
+26844
+26845
+29628
+29650
+29651
+29652
+29653
+29654
+29659
+29660
+31841
+70320
+31846
+31849
+31850
+67821
+31851
+31852
+31853
+31854
+31855
+32373
+32393
+32395
+32397
+67831
+32392
+32598
+68088
+70205
+32609
+32610
+32613
+67835
+32614
+32615
+37120
+68104
+68832
+37147
+37149
+37150
+37151
+37128
+37130
+37134
+67914
+37135
+45835
+45841
+67119
+45845
+45846
+45847
+45848
+45849
+45850
+46260
+68400
+68138
+70274
+69034
+68139
+46273
+46275
+67125
+46270
+46271
+46712
+68664
+70239
+46738
+46739
+46740
+67129
+46741
+46735
+46736
+48507
+48515
+67149
+48518
+48519
+48520
+68762
+48816
+48817
+48818
+48819
+67153
+48824
+48804
+48805
+48814
+70231
+51349
+51350
+51351
+51352
+67195
+51357
+51358
+51494
+68152
+68776
+51516
+51505
+51506
+67201
+51514
+51515
+52834
+70122
+52864
+52865
+52866
+67221
+52861
+56652
+68228
+56697
+56703
+56695
+56696
+57034
+68412
+68168
+70034
+57083
+57084
+57085
+67280
+57086
+57081
+57037
+68166
+57043
+57046
+57047
+57048
+67282
+57049
+57050
+62728
+68740
+70304
+62744
+62749
+67052
+62750
+62751
+63279
+70213
+63286
+67067
+63290
+63291
+63574
+68572
+68714
+68186
+63594
+63595
+67076
+63591
+63593
+64049
+70261
+64062
+64064
+67008
+64066
+64067
+64303
+70347
+64310
+64312
+67014
+64313
+64314
+01457
+68276
+70219
+01467
+01468
+01469
+01470
+01465
+04694
+04716
+67381
+04720
+04721
+04722
+04833
+04853
+67385
+04859
+04860
+04861
+04862
+05296
+05301
+05302
+05304
+05305
+67394
+05307
+05308
+05309
+06550
+06557
+67415
+06560
+06561
+06562
+08478
+68566
+08489
+08490
+08491
+67453
+08487
+08695
+67460
+08697
+08698
+11552
+70298
+70033
+68025
+69024
+11565
+67510
+11562
+11564
+11630
+67513
+11636
+11637
+11704
+68026
+70206
+11714
+11715
+11716
+11717
+67512
+11712
+13258
+68756
+13270
+13271
+13272
+67528
+13276
+13277
+14450
+68652
+14455
+14456
+14459
+65432
+67549
+67548
+14460
+14748
+70334
+68858
+14754
+14758
+14759
+20064
+20073
+20075
+20076
+21052
+21076
+67643
+21079
+21080
+22071
+70083
+22088
+22089
+67666
+22096
+22097
+23567
+68214
+70342
+68952
+23575
+23576
+67691
+23577
+23573
+25037
+70259
+25074
+25075
+67723
+25071
+26506
+70284
+26528
+26533
+67743
+26534
+26535
+26536
+27762
+68514
+70085
+27772
+27773
+27774
+67765
+27769
+27771
+31896
+31902
+31904
+31905
+67823
+31906
+31907
+31908
+32298
+70297
+32305
+32307
+32308
+32309
+67828
+32310
+32673
+32675
+32663
+67836
+32664
+33450
+33478
+33481
+67852
+33482
+33484
+33485
+34009
+34010
+67861
+34013
+34014
+34015
+34016
+37575
+68480
+68106
+37587
+37588
+37589
+67922
+37583
+37584
+37586
+39604
+39629
+39633
+69040
+39630
+39631
+67944
+39634
+39628
+42196
+68646
+68129
+42228
+42234
+67981
+42236
+42237
+42238
+43519
+70198
+43526
+43527
+43531
+43532
+43533
+43534
+44677
+44685
+44688
+44689
+67107
+67105
+44690
+47174
+68141
+68864
+70375
+47180
+67134
+47184
+47185
+47186
+50846
+50853
+50854
+50857
+50858
+50859
+50860
+50861
+50852
+53290
+53297
+53298
+53302
+53303
+54548
+68914
+54559
+54560
+54564
+54565
+54566
+55768
+70022
+55774
+55775
+55778
+55779
+55780
+55781
+57519
+57532
+57533
+57536
+67290
+57537
+57539
+57540
+58588
+68488
+68172
+58595
+58596
+58597
+67310
+58598
+58599
+59203
+59213
+59218
+59479
+68478
+68173
+59501
+59504
+67323
+59505
+63415
+68184
+70265
+63423
+63424
+63425
+63426
+67069
+63420
+63421
+64351
+64384
+64386
+64387
+64388
+67015
+64389
+64390
+00412
+00419
+00422
+00423
+67339
+67341
+00424
+00425
+03117
+03123
+03126
+03127
+03128
+03129
+03130
+03434
+70124
+03440
+03442
+03443
+67371
+03444
+04790
+04800
+04804
+67384
+04805
+04895
+04901
+04902
+04904
+67387
+04905
+68990
+70254
+05474
+05475
+67397
+05477
+05595
+05603
+67399
+05607
+05608
+07436
+69002
+07459
+07460
+07461
+67433
+07457
+08363
+08371
+08374
+08375
+08376
+67450
+08377
+08378
+09178
+70200
+09184
+09187
+67475
+09183
+09719
+09724
+09727
+09728
+67481
+09729
+09730
+09731
+10099
+68020
+69056
+10108
+67487
+10114
+68800
+10465
+10469
+10470
+67495
+10471
+10472
+10473
+10696
+68554
+10712
+10713
+10714
+10706
+10707
+11197
+11201
+11202
+11204
+11205
+11206
+67504
+11209
+11210
+11212
+68654
+11255
+11256
+11257
+11259
+11264
+11265
+13133
+13139
+13141
+13146
+13147
+13542
+13547
+13548
+67534
+13552
+13553
+70149
+13809
+13810
+13811
+13801
+13802
+14780
+14797
+14800
+14801
+67553
+14802
+16581
+16588
+16590
+16594
+16595
+16596
+16597
+68746
+68040
+70250
+17661
+67593
+17664
+17658
+17751
+68476
+70262
+17763
+17764
+17765
+17770
+17771
+17825
+17826
+17829
+67595
+17830
+17831
+21248
+21250
+21239
+67646
+21246
+21425
+21438
+21439
+21440
+21441
+67653
+21435
+21437
+22798
+70129
+22809
+67673
+22814
+22815
+23945
+69060
+23954
+67700
+23955
+23949
+24023
+68242
+68798
+68056
+24032
+67702
+24028
+24030
+24031
+24590
+68840
+24612
+24613
+67709
+24610
+25240
+25249
+67721
+25251
+25252
+26938
+26947
+26948
+26953
+26954
+26955
+27254
+68072
+68610
+27269
+27271
+27272
+67758
+27267
+27917
+27927
+67769
+27931
+27932
+27933
+28157
+70373
+28166
+28167
+28168
+67777
+28163
+28165
+28461
+68772
+28470
+28471
+28472
+67781
+28467
+28468
+30158
+67794
+30163
+30164
+30165
+31147
+69006
+31153
+31154
+31159
+31161
+31163
+68090
+33224
+33210
+33211
+33212
+33220
+33221
+33537
+70019
+68574
+33543
+33544
+67853
+67851
+33540
+33541
+34558
+68226
+70082
+34582
+67871
+34585
+70217
+35296
+35297
+67880
+67881
+35301
+35302
+36826
+70222
+36832
+67906
+36837
+36838
+36839
+36912
+68268
+68102
+70052
+36920
+36921
+36917
+36919
+37356
+68105
+69044
+37371
+37372
+67917
+37368
+38103
+38128
+38129
+38131
+67929
+38132
+38585
+68452
+68920
+68111
+38593
+38589
+38591
+38592
+39454
+68826
+39461
+39465
+67943
+39467
+70172
+40174
+40175
+67950
+67951
+68126
+40993
+40994
+67959
+40995
+40989
+40990
+41306
+41318
+41320
+66266
+41323
+41324
+68380
+70151
+41455
+67965
+41456
+41451
+41819
+70210
+70267
+41828
+41829
+41830
+41831
+67971
+41832
+41825
+43037
+70177
+43063
+43064
+43065
+43066
+43059
+43060
+43213
+70220
+68670
+43225
+67083
+43221
+43223
+43224
+44080
+68706
+68738
+44089
+44090
+44084
+44364
+44372
+44375
+44376
+67097
+44377
+47655
+70170
+68996
+47663
+47664
+67139
+47665
+47660
+48038
+68898
+48047
+48049
+48052
+48053
+48045
+48898
+48905
+67154
+48909
+48910
+48911
+48912
+49127
+49136
+49137
+49138
+67157
+49139
+49133
+49244
+49249
+49253
+67162
+49254
+68284
+68466
+70209
+68154
+51630
+51633
+51772
+70351
+68548
+51779
+51780
+67204
+51776
+52550
+52558
+52559
+67214
+52562
+52564
+52565
+53190
+68159
+68978
+53215
+67226
+53211
+53213
+53214
+53453
+70188
+53489
+53490
+53491
+67230
+53492
+53487
+55332
+55344
+55345
+55346
+67256
+55347
+55342
+57035
+70204
+68167
+57065
+57066
+57067
+67281
+57062
+57781
+57786
+57787
+57792
+57793
+58782
+58789
+58790
+58792
+58793
+67314
+58794
+59298
+68496
+59311
+67319
+59317
+59318
+60345
+70256
+60351
+67334
+60355
+60356
+62267
+62278
+62280
+62281
+62282
+67039
+62283
+62277
+63079
+68181
+68998
+63088
+63089
+63090
+67061
+63085
+64423
+64432
+64435
+64436
+64437
+64438
+64439
+64440
+64431
+02227
+70037
+02234
+02235
+02236
+02237
+02232
+02581
+02590
+02591
+67363
+02588
+03266
+03276
+03280
+03281
+04484
+04512
+04513
+67377
+04510
+70164
+04596
+04597
+67379
+04602
+04603
+04595
+04764
+70031
+04773
+04774
+67383
+67382
+04771
+05270
+05281
+05282
+05283
+05284
+67393
+05279
+05800
+05805
+05806
+05809
+05810
+67404
+05813
+05814
+05815
+07803
+68828
+07814
+07809
+07812
+07813
+09462
+09463
+09464
+09470
+09472
+09773
+09782
+67480
+09783
+09784
+09785
+09914
+09923
+09925
+10260
+70064
+10273
+10274
+67492
+10270
+10272
+13467
+13474
+13475
+13476
+67532
+13477
+13471
+13472
+13850
+70363
+13861
+13862
+67539
+13856
+13857
+15317
+15324
+15331
+15368
+67559
+15372
+15373
+16437
+70153
+16442
+67574
+16447
+16448
+16449
+16961
+70341
+68426
+16971
+65500
+67577
+16973
+16975
+68039
+68782
+17601
+17602
+67588
+17604
+17605
+17597
+18270
+18295
+18296
+18297
+67607
+18293
+18301
+70079
+68043
+68966
+18313
+67609
+18309
+18312
+18485
+70065
+18489
+18491
+67612
+18496
+18497
+19402
+19413
+19417
+21202
+21220
+21210
+67647
+21217
+21218
+21219
+21272
+21277
+67649
+21284
+21285
+22537
+22554
+22555
+22556
+22557
+67672
+22552
+23579
+70257
+23586
+23587
+23588
+67692
+23584
+24718
+68334
+24729
+67710
+24731
+26250
+26256
+26258
+26260
+26261
+26559
+68065
+68352
+26576
+67744
+26577
+26571
+26572
+27013
+27047
+27048
+67752
+27053
+27055
+28134
+70303
+28141
+28142
+28146
+28148
+28149
+28306
+28313
+67778
+28316
+28317
+28318
+70236
+68616
+28433
+28427
+28435
+28428
+28432
+29134
+29142
+29146
+29140
+29141
+30227
+30237
+30238
+67793
+30240
+30241
+68080
+69000
+31323
+67810
+31317
+31318
+31321
+31322
+31641
+31654
+31656
+31657
+31658
+67815
+31659
+70158
+32454
+32455
+32456
+32457
+32451
+32453
+35117
+35118
+35122
+67875
+35123
+35125
+35353
+70066
+35367
+35368
+67882
+35365
+36044
+68562
+68100
+36052
+36053
+67893
+36054
+36055
+36049
+36645
+70075
+68628
+36653
+36657
+67901
+36658
+36659
+36732
+68608
+70305
+36740
+67904
+36737
+36739
+36774
+36785
+36787
+36788
+67903
+36789
+36790
+68328
+37091
+67913
+37100
+37773
+70180
+37791
+37792
+37795
+37983
+70337
+69008
+37991
+37992
+67926
+37988
+37990
+38281
+70021
+70050
+69036
+38288
+38289
+38292
+38286
+38314
+70338
+68584
+38322
+38323
+38319
+38321
+38614
+70280
+68934
+38628
+38629
+38630
+38631
+38625
+38654
+70187
+38688
+38689
+38690
+67933
+38686
+38765
+68774
+38864
+38865
+38866
+38867
+38862
+39087
+39094
+39095
+39096
+67940
+39097
+39098
+39092
+39402
+39449
+39450
+39451
+39445
+39446
+39448
+39582
+68342
+68118
+39594
+39595
+39590
+39593
+41510
+67968
+41512
+67966
+41513
+41514
+41515
+41578
+68736
+41597
+41602
+41604
+41606
+41607
+42242
+42250
+42254
+42255
+43922
+70061
+43936
+43937
+67093
+43934
+48965
+48977
+48980
+67155
+48981
+48982
+49327
+49333
+49336
+67164
+49337
+49331
+49332
+49367
+49376
+49377
+49381
+49382
+49375
+49652
+49660
+49661
+67173
+49662
+49657
+50035
+50057
+50062
+67176
+50064
+50065
+68150
+70315
+51134
+51137
+67191
+51138
+51139
+51132
+68508
+68600
+68151
+67190
+67192
+51179
+51182
+51183
+51702
+51711
+67202
+51707
+51882
+68680
+51888
+67206
+51892
+51893
+51894
+52314
+68157
+68912
+67212
+52319
+52321
+52322
+52323
+53363
+70225
+53372
+53373
+53374
+67228
+53368
+55035
+68564
+55047
+55051
+55052
+55046
+56315
+56316
+56310
+56311
+56800
+68165
+68868
+56828
+56829
+67277
+56827
+57298
+57305
+57308
+57309
+57310
+57312
+57430
+57443
+57444
+57445
+57446
+57440
+57653
+68796
+57663
+57664
+57665
+57666
+57658
+57954
+57980
+57981
+67300
+57976
+57978
+57979
+58286
+68560
+58299
+58301
+58302
+67305
+58297
+58676
+70346
+58685
+58686
+58680
+59238
+70179
+59250
+59254
+67317
+59256
+59258
+59259
+60543
+60553
+70215
+67018
+60554
+60555
+60550
+61031
+61060
+61061
+61063
+61065
+67022
+61066
+61067
+61068
+61913
+61919
+61922
+67031
+61923
+61977
+61985
+67033
+61988
+61989
+61991
+61998
+61999
+67032
+62001
+62002
+62706
+62714
+62715
+67051
+62716
+62711
+62901
+70186
+62913
+62914
+67055
+62911
+62921
+62928
+62929
+67056
+62926
+63062
+68372
+63071
+63075
+67064
+63076
+63102
+68382
+63111
+63112
+67062
+63107
+63109
+63110
+68185
+63475
+63476
+63477
+67072
+63472
+63474
+63750
+63758
+63759
+63761
+63762
+63757
+01429
+70279
+68422
+01441
+01437
+01440
+01482
+01491
+01493
+01494
+67352
+01488
+02102
+02111
+67357
+02107
+03053
+70155
+67368
+04183
+04191
+04195
+04196
+04341
+04350
+04351
+65115
+04346
+04349
+04866
+04874
+67386
+04871
+05193
+70041
+05200
+05201
+05202
+67391
+05197
+05488
+67396
+05494
+05495
+05496
+05497
+05852
+05859
+05860
+05861
+67405
+05857
+06062
+70328
+06074
+06068
+06070
+06071
+06072
+06073
+06643
+06656
+06657
+67416
+06653
+07199
+07200
+67427
+07201
+07202
+07194
+07369
+70318
+07379
+07380
+67430
+07381
+07374
+07413
+70238
+67434
+07419
+07422
+07423
+07469
+07476
+07477
+07478
+67436
+07474
+07593
+70063
+07603
+67439
+07599
+07602
+08586
+70228
+69064
+67458
+08591
+70268
+69078
+09240
+67473
+09245
+09246
+10395
+70060
+10399
+10406
+10407
+10408
+10410
+10411
+10435
+10451
+10452
+10448
+10624
+10628
+10630
+10639
+10640
+10768
+68658
+10776
+65358
+67499
+10773
+10775
+13169
+13180
+13181
+67526
+13178
+14276
+14294
+14295
+14296
+67546
+14297
+14290
+14471
+14472
+14473
+14474
+14484
+15092
+15102
+15097
+15098
+15153
+15164
+15165
+67557
+15161
+15169
+15179
+67558
+15174
+15175
+15178
+15527
+15535
+15536
+67561
+15532
+16402
+67572
+16408
+16410
+16889
+16896
+16898
+67576
+17422
+70045
+17432
+17434
+17435
+17919
+17938
+17934
+17937
+18145
+18160
+67603
+18157
+18159
+18190
+18202
+18203
+67604
+18198
+18200
+18201
+18218
+18226
+18227
+67605
+18222
+18223
+18224
+18769
+18776
+18777
+18778
+67617
+18774
+19230
+68512
+68045
+67625
+19234
+19235
+19237
+19238
+19556
+19564
+19565
+19566
+19562
+20362
+20371
+67634
+20372
+20368
+20434
+70218
+20442
+20443
+20444
+20439
+21184
+21186
+67645
+21177
+21689
+21709
+21699
+67655
+21707
+21708
+68896
+70134
+21860
+67658
+21856
+22131
+22138
+22139
+22140
+67668
+22136
+22137
+22491
+68792
+70043
+22498
+22500
+67671
+22501
+22642
+22629
+22633
+22634
+67674
+22712
+70203
+22722
+22719
+23637
+70020
+67694
+23641
+23643
+23644
+24405
+70055
+24445
+67707
+24441
+24444
+26491
+70133
+26502
+67742
+26498
+26500
+26501
+27171
+70017
+68236
+27180
+67755
+27181
+27182
+27183
+27175
+27580
+27597
+27598
+27599
+27600
+27601
+27594
+68404
+70150
+28040
+28042
+67773
+28043
+28647
+28653
+28656
+67783
+28657
+29064
+67786
+29070
+29072
+29073
+30467
+30473
+30475
+30476
+65960
+30477
+30478
+30927
+30938
+30942
+67803
+30944
+30945
+30973
+30982
+67804
+30986
+30987
+31723
+31736
+31738
+31739
+67816
+31734
+31777
+31779
+31780
+67818
+31782
+31783
+32200
+32210
+32211
+32212
+32213
+32208
+32281
+32287
+32288
+32289
+32290
+32291
+32286
+32806
+32816
+32818
+32819
+67838
+32962
+32969
+32970
+32971
+67842
+32967
+34955
+68326
+34963
+34964
+34960
+34962
+35524
+68696
+35532
+35533
+67887
+35529
+35531
+70302
+36250
+67896
+36246
+36259
+36267
+36268
+67897
+36265
+70113
+36339
+36340
+36344
+36348
+36677
+36684
+36685
+36686
+67902
+36682
+36856
+70278
+69022
+36864
+36865
+67907
+36861
+36991
+69096
+36999
+37000
+67909
+36997
+37849
+37857
+67924
+37863
+38081
+70038
+38091
+67927
+38088
+38090
+38176
+68532
+70324
+67931
+38180
+38182
+38183
+38184
+38185
+38329
+70277
+68716
+38343
+38344
+67932
+38339
+38342
+39846
+39855
+67946
+39856
+39852
+41920
+68346
+70144
+41930
+67974
+41926
+41929
+41937
+70068
+41944
+67975
+41947
+41948
+42419
+68704
+68130
+42431
+42433
+42426
+42427
+42429
+42649
+68131
+42659
+67982
+42655
+42657
+42658
+43335
+68134
+43346
+43347
+43348
+43343
+43460
+43448
+43449
+67086
+43678
+43685
+43686
+43687
+67089
+43688
+43683
+43828
+43837
+67092
+43838
+43831
+45754
+45764
+45765
+45766
+67118
+45761
+45965
+70121
+46000
+67121
+45997
+45999
+46287
+46294
+46302
+47485
+47495
+47497
+47498
+47499
+67136
+49475
+49485
+49486
+67167
+49481
+49627
+70143
+49639
+67171
+49636
+49638
+49689
+49696
+50407
+50413
+50416
+50418
+50419
+50412
+50834
+69058
+50842
+50843
+50844
+50840
+51901
+68570
+51924
+51925
+67207
+51926
+51920
+52502
+52513
+52514
+67213
+52511
+52707
+52718
+52719
+67217
+52715
+52717
+52913
+52924
+52925
+67223
+52922
+52923
+53023
+68432
+70071
+68158
+67224
+53027
+53029
+53693
+53705
+53708
+54413
+54445
+54446
+67241
+54441
+54442
+54819
+70300
+54827
+54824
+54826
+55298
+70096
+55306
+55307
+55303
+55305
+55312
+70362
+55320
+67255
+55317
+55319
+55861
+55862
+55863
+67262
+55858
+56080
+56087
+56088
+56090
+56091
+56092
+57815
+70223
+57831
+57833
+57834
+67293
+57830
+58443
+70108
+67308
+58449
+67311
+58659
+58660
+58731
+58738
+58739
+58740
+67313
+58741
+58735
+58737
+58804
+70349
+58812
+58813
+58809
+58811
+59143
+70191
+59154
+59152
+59366
+59372
+59374
+59375
+59376
+59540
+59553
+59554
+67322
+59549
+59550
+59552
+59760
+59770
+59771
+59772
+59773
+59766
+59767
+60450
+70331
+68498
+67335
+60461
+60463
+60464
+61141
+68640
+61149
+67023
+61145
+61147
+61148
+62637
+67049
+62647
+62648
+62649
+62729
+70377
+62737
+62735
+00588
+00598
+00601
+00602
+01194
+01211
+01212
+01215
+01216
+01411
+01426
+01427
+67351
+01423
+01551
+01562
+01555
+02041
+02051
+02047
+02050
+02348
+68002
+67360
+02352
+02354
+02355
+02539
+02547
+67362
+02543
+02546
+02707
+70275
+02714
+02716
+02717
+02818
+70095
+02827
+02829
+02830
+02831
+02832
+68812
+03243
+03244
+03245
+03239
+03469
+70142
+03485
+67372
+03483
+03625
+68266
+68004
+68408
+68005
+03637
+03640
+03641
+04165
+68250
+04172
+04174
+04175
+04422
+04432
+04433
+04428
+04436
+04444
+04445
+67376
+04442
+04678
+04686
+67378
+04683
+04685
+05111
+05115
+05116
+05117
+05651
+05658
+05659
+05660
+67401
+05655
+05699
+05713
+67402
+05711
+05921
+05930
+67407
+05927
+05929
+06176
+06195
+06196
+06425
+06437
+06438
+06433
+07258
+68013
+68494
+07273
+07275
+07276
+08608
+08620
+67457
+08615
+08669
+08679
+08675
+08840
+08848
+67466
+08845
+08847
+09312
+09314
+67474
+09570
+09579
+67479
+09575
+09577
+09578
+09891
+09899
+67485
+09895
+09897
+09898
+10134
+10142
+10143
+10139
+10140
+10577
+10586
+67497
+10582
+10584
+10585
+11030
+68410
+11038
+11039
+11041
+11356
+67506
+67505
+11363
+12111
+12121
+12117
+12119
+12120
+12222
+12229
+12230
+67518
+12227
+12870
+12878
+65399
+67524
+12875
+12967
+12975
+12976
+67525
+12972
+12974
+13311
+70181
+13314
+13315
+13317
+13318
+13319
+13349
+13357
+13353
+14374
+14382
+14383
+14384
+14377
+14381
+14829
+68614
+14843
+14844
+14838
+14840
+14841
+15421
+15429
+67560
+15427
+15594
+15609
+15604
+15605
+15608
+15729
+15730
+67565
+15731
+15726
+15769
+70229
+15777
+15778
+15779
+15780
+15932
+15940
+15937
+15939
+16123
+16132
+16133
+16128
+16129
+16130
+17122
+17151
+17152
+67580
+17150
+18066
+18067
+18060
+18062
+18063
+18064
+18065
+18271
+18277
+18278
+18282
+18283
+18521
+18529
+18530
+67614
+18531
+18785
+70195
+68910
+18792
+18793
+18799
+18876
+18877
+67618
+18878
+18882
+18883
+19103
+19117
+19118
+19112
+19113
+19116
+19325
+70146
+19334
+19329
+19331
+19332
+19333
+19679
+19693
+19694
+67631
+19690
+19752
+19767
+19768
+19769
+19763
+20137
+20144
+20148
+20374
+20385
+67635
+20382
+20384
+20693
+20703
+20708
+20709
+20792
+70100
+20796
+20798
+20799
+20800
+21393
+68582
+21410
+21414
+21415
+21416
+21548
+21558
+21559
+67652
+21561
+21562
+21555
+22269
+22279
+67669
+22275
+22277
+22278
+22739
+68052
+70058
+22747
+22749
+22836
+70273
+67675
+22995
+22997
+22998
+22999
+23047
+23053
+23054
+23439
+23450
+23446
+23447
+23449
+23541
+70027
+23546
+23550
+67690
+23551
+70163
+23672
+67695
+23669
+24580
+24588
+67708
+24585
+24587
+24909
+24921
+67716
+24918
+24920
+25370
+25379
+25376
+25404
+70330
+67724
+25410
+25412
+25628
+25634
+25635
+25640
+25649
+25650
+67728
+25646
+67731
+67732
+25921
+25923
+25924
+25925
+25975
+25983
+67734
+25980
+25981
+26312
+26316
+26318
+26321
+67741
+26323
+27150
+27158
+67754
+27156
+27387
+27409
+67759
+27410
+27414
+27415
+27568
+27574
+27576
+67761
+28375
+28383
+28384
+67779
+28380
+28382
+28438
+70147
+28443
+28444
+28450
+28452
+28453
+30383
+30391
+67796
+30387
+30389
+30390
+30457
+30458
+30461
+30463
+30464
+32509
+32517
+67833
+32513
+32515
+32516
+33048
+68206
+70226
+67843
+33054
+33056
+33057
+33122
+67844
+33124
+33125
+33126
+33638
+33647
+67855
+33644
+33645
+34118
+34125
+34133
+34134
+34136
+34210
+34212
+34213
+67866
+34215
+34385
+34393
+34394
+67867
+34389
+34391
+34392
+70103
+67883
+35342
+35343
+35345
+35944
+35949
+35951
+35953
+67892
+37464
+67920
+37472
+37474
+37475
+37476
+37945
+37963
+37964
+37965
+37960
+39031
+68117
+68824
+39035
+39037
+39038
+39039
+70214
+39745
+39746
+39749
+40463
+68362
+40471
+41661
+70311
+67970
+41665
+41667
+41668
+42695
+67984
+42705
+42707
+42708
+43570
+43578
+43579
+43574
+43576
+43577
+43900
+70125
+43911
+43905
+44034
+44048
+44049
+67094
+44056
+66325
+44339
+44340
+44348
+44469
+44477
+67102
+44472
+44475
+44476
+44598
+44609
+44610
+67106
+44605
+44608
+45474
+45486
+67115
+45482
+45484
+45485
+45852
+45860
+45856
+45858
+45859
+46339
+70353
+67126
+46370
+46371
+46425
+46435
+67127
+46432
+46434
+47103
+47111
+67132
+47108
+47110
+47602
+47613
+47614
+47609
+47610
+48738
+70159
+48744
+48748
+48749
+48750
+49073
+70062
+67158
+49080
+49082
+49083
+49537
+49545
+49546
+67169
+49541
+49543
+49544
+50263
+68830
+50270
+50272
+50273
+50502
+50509
+50510
+50511
+67180
+50507
+51325
+51336
+51337
+51338
+67197
+51339
+51332
+51813
+51820
+51821
+51822
+51823
+51816
+52570
+52583
+67215
+52588
+52613
+52622
+67216
+52618
+52619
+53046
+53069
+53071
+53072
+53413
+53421
+53422
+53418
+53419
+53427
+70059
+67229
+53430
+53432
+53433
+53543
+53551
+53553
+53554
+53555
+53549
+54069
+54079
+54086
+54078
+54204
+54212
+67236
+54208
+54210
+54211
+54870
+54885
+67248
+54881
+54884
+54991
+55002
+54996
+54997
+55001
+55069
+67250
+55073
+55075
+55076
+55145
+70192
+67253
+55153
+55155
+55156
+55169
+55183
+67252
+55180
+55197
+55205
+67254
+55201
+55203
+55204
+55274
+55284
+55285
+55280
+55656
+55666
+55667
+55662
+55665
+55730
+67263
+55734
+55736
+55737
+56157
+56165
+56166
+67269
+56162
+56164
+56294
+56302
+56303
+56304
+56299
+56301
+57131
+57140
+67284
+57135
+57669
+57674
+57675
+67294
+57678
+57679
+57680
+57673
+57757
+57766
+57767
+57760
+57761
+57801
+57808
+57809
+57811
+57812
+57805
+58023
+67303
+58028
+58030
+58384
+58395
+58396
+58390
+58391
+58394
+58933
+70165
+58943
+58938
+58939
+58942
+59902
+70024
+59911
+59912
+59907
+59958
+67328
+59961
+59963
+59965
+60759
+60767
+67021
+60764
+60766
+60850
+60858
+60859
+60855
+60857
+62447
+62448
+62454
+62455
+62584
+70123
+67047
+62589
+62592
+62684
+68468
+67050
+62693
+62695
+63358
+70350
+70292
+63363
+63369
+63370
+63371
+63372
+63399
+70086
+63406
+63409
+63830
+63838
+63839
+67004
+63840
+63834
+63872
+70301
+63878
+63881
+67006
+63882
+63877
+63947
+68876
+63961
+64401
+68606
+70366
+67016
+64415
+64417
+64418
+00972
+00966
+01721
+01733
+67354
+01728
+02158
+02167
+02164
+02173
+02176
+02179
+02180
+02473
+02482
+02479
+02517
+02524
+67361
+02520
+02522
+02523
+02564
+02570
+67364
+02575
+02941
+02959
+67366
+02955
+03328
+67369
+03340
+03343
+03360
+67370
+03363
+03365
+03366
+03367
+03422
+03423
+03426
+03427
+03429
+03570
+67373
+03566
+03569
+03587
+03599
+03602
+67374
+04039
+04043
+04046
+04047
+04056
+04066
+04067
+04075
+04065
+04814
+04825
+04829
+04830
+05068
+05071
+05165
+70128
+67390
+05174
+05176
+05756
+05760
+05761
+05764
+05765
+05766
+05959
+67406
+05963
+05966
+06407
+06417
+06418
+06412
+06413
+06526
+06530
+06531
+06534
+06535
+06929
+67421
+06934
+06941
+67423
+06950
+06952
+06953
+07226
+67426
+07229
+07232
+07233
+07236
+67428
+07246
+07241
+07259
+07268
+07263
+07284
+67429
+07289
+07291
+70166
+07302
+07306
+07490
+67437
+07494
+07496
+07497
+08118
+08122
+08124
+08125
+08126
+08330
+68960
+08353
+08359
+08360
+08361
+08439
+67452
+08451
+08453
+08454
+08739
+08754
+08750
+08751
+08858
+08867
+08864
+08897
+67469
+08901
+08903
+08904
+09149
+09154
+09156
+09160
+09161
+09162
+09164
+10588
+10597
+10593
+11812
+11827
+67515
+11822
+11826
+11900
+11907
+11909
+11910
+70102
+12589
+12591
+12592
+13654
+13658
+13660
+13661
+13662
+13663
+15556
+67563
+15560
+15562
+15563
+15807
+15816
+15812
+15825
+15833
+15830
+15832
+16066
+67567
+16070
+16071
+16072
+17123
+17130
+17134
+17588
+67590
+17584
+17587
+17624
+17634
+17635
+67591
+17629
+17907
+67598
+17913
+17916
+17945
+67600
+17950
+17952
+17953
+18027
+70039
+18031
+18033
+18034
+18035
+18137
+67601
+18141
+18143
+18164
+18183
+18177
+18181
+18182
+19456
+67628
+19460
+19461
+19463
+19798
+65648
+19844
+19846
+19847
+19848
+20152
+70286
+20160
+20161
+20159
+68212
+68047
+67636
+20228
+20231
+20232
+20387
+20397
+20398
+20400
+20394
+20402
+20409
+20412
+20413
+20736
+20743
+20744
+20747
+20748
+21319
+21332
+21333
+21326
+21330
+21331
+21394
+21401
+21403
+21404
+21400
+22683
+22697
+22698
+22699
+22701
+22907
+22919
+22916
+22918
+23326
+23336
+23337
+67684
+23333
+23626
+67693
+23631
+23633
+23865
+23876
+23877
+23871
+23872
+23875
+24056
+70042
+65801
+67703
+24058
+24060
+24061
+24062
+24867
+24873
+24878
+24879
+25003
+25018
+67718
+25016
+25024
+68662
+70114
+70321
+25030
+25034
+25182
+67719
+25187
+25189
+25190
+25428
+70329
+67725
+25433
+25481
+70312
+67726
+25487
+25820
+25830
+25827
+25968
+67733
+25963
+25964
+26294
+67740
+26299
+26301
+26302
+26697
+68392
+68067
+26704
+26708
+26709
+26702
+26831
+26855
+26857
+26858
+26859
+27249
+27277
+27278
+27983
+70106
+28002
+67772
+27998
+28000
+28001
+28772
+28782
+67784
+28777
+28778
+28779
+29087
+29095
+67788
+29091
+29094
+29171
+29722
+29724
+29725
+29380
+29387
+29389
+29390
+29865
+70162
+29869
+29871
+29872
+30051
+67792
+30056
+30058
+30059
+30674
+67799
+30678
+30680
+30681
+32692
+67837
+32698
+32700
+32885
+32893
+67839
+32890
+32892
+33303
+33315
+33317
+33318
+33390
+70109
+67849
+33394
+33396
+33397
+33451
+33462
+33463
+33459
+34869
+34871
+34865
+34866
+35147
+35161
+35162
+67876
+68524
+70099
+67877
+35221
+35225
+35371
+35384
+67885
+67884
+35382
+68672
+36608
+36609
+36610
+37045
+37048
+37049
+67911
+37051
+37044
+37201
+67916
+37197
+37199
+37200
+37313
+37321
+37322
+37317
+37319
+37320
+37332
+68286
+67918
+37336
+37338
+37339
+37395
+70291
+37429
+37431
+37396
+68778
+70368
+37404
+37405
+67919
+37909
+37912
+37913
+37916
+38008
+38020
+38016
+38017
+38061
+38070
+67928
+38066
+38068
+38721
+70253
+38732
+38734
+38736
+39503
+70036
+39512
+39513
+39508
+39992
+39993
+39994
+39988
+39989
+40010
+40018
+40019
+40015
+40308
+67952
+40316
+40318
+40319
+40320
+41354
+41360
+41361
+41363
+42752
+67987
+42756
+42757
+44267
+44277
+44279
+44280
+44422
+70255
+67100
+44425
+44427
+44428
+44491
+67104
+44494
+44497
+44498
+44499
+44828
+44832
+44834
+44835
+44836
+44896
+70154
+67108
+44899
+44901
+44902
+45494
+45514
+45516
+45517
+45518
+45684
+67117
+45688
+45690
+45691
+45777
+45784
+45787
+45788
+45905
+67120
+45911
+45913
+45914
+46155
+67123
+46177
+46180
+46181
+46997
+47008
+47009
+47004
+47006
+47007
+47458
+47464
+47465
+47467
+47468
+70057
+67138
+47670
+47673
+47674
+47782
+47791
+47792
+47793
+48207
+48214
+48215
+48222
+48249
+67145
+48254
+48256
+48257
+48603
+48609
+48612
+48613
+50107
+68444
+67178
+50123
+50126
+50127
+50138
+70360
+50141
+50142
+50145
+50328
+50338
+50339
+50340
+50333
+50334
+50719
+68420
+68146
+67182
+50722
+50725
+50726
+50932
+67185
+50937
+50939
+50940
+50941
+51019
+51028
+51026
+51094
+51112
+51272
+70294
+51291
+51293
+51294
+51695
+51689
+51691
+51692
+51693
+51694
+52168
+52176
+52177
+52178
+52172
+52197
+68482
+52204
+67211
+52202
+52236
+52237
+52233
+53435
+53446
+53450
+53451
+53640
+53645
+53647
+53899
+53904
+53905
+53909
+54255
+67237
+54259
+54261
+54262
+54263
+54360
+68734
+68160
+54380
+54383
+54384
+54586
+67245
+54598
+54599
+54915
+67249
+54926
+54929
+54930
+54931
+70371
+55424
+67258
+67257
+55429
+55430
+55745
+55754
+55748
+55749
+55752
+55753
+55837
+55843
+55844
+55845
+55847
+55848
+55840
+56191
+56201
+67270
+56197
+57260
+57268
+57269
+57264
+57267
+58629
+58726
+58728
+58729
+58705
+58708
+58711
+58712
+58713
+59404
+67320
+59408
+59410
+59604
+59605
+59607
+59608
+59599
+60043
+67330
+60047
+60049
+60050
+60051
+67019
+60578
+60580
+60887
+60911
+60913
+60914
+60947
+60953
+61737
+67029
+61741
+61744
+61965
+68378
+61970
+61972
+62285
+67041
+62292
+62293
+62295
+62296
+62301
+67040
+62304
+62306
+62307
+62308
+62329
+62360
+62363
+62370
+62371
+63447
+70070
+67071
+63451
+63453
+63454
+63621
+63626
+63637
+63638
+63639
+63720
+67080
+63725
+63727
+63728
+64543
+68193
+68862
+67017
+64546
+64548
+64550
+00829
+00840
+00841
+00838
+00897
+67347
+00902
+00903
+01572
+70117
+01578
+01580
+01581
+01796
+67355
+01803
+01805
+01806
+01964
+01959
+02089
+02094
+02096
+02097
+02098
+02618
+67365
+02621
+02623
+02624
+02871
+02876
+02878
+02879
+04008
+67375
+04014
+04016
+04110
+04121
+04122
+04117
+04154
+04158
+04160
+04161
+04288
+04293
+04295
+04386
+04396
+04391
+04395
+05554
+05561
+05564
+05614
+70148
+05618
+05620
+05621
+05910
+05919
+05914
+05918
+06014
+06024
+06026
+06027
+06028
+06039
+06048
+06050
+06051
+06157
+67408
+06162
+06164
+06312
+67410
+06316
+06318
+70343
+07444
+07446
+07447
+07450
+07608
+70098
+07639
+67440
+07635
+07747
+67444
+07752
+07754
+07755
+07756
+08002
+08007
+67447
+70196
+68590
+67449
+08133
+08410
+70105
+08418
+08419
+08415
+08782
+08790
+67463
+08785
+08789
+08799
+68017
+68884
+08821
+08822
+67471
+09089
+09094
+09202
+09211
+09212
+09317
+09327
+09322
+09632
+09635
+09638
+09639
+09805
+67482
+09809
+09811
+09834
+09839
+09841
+09933
+09940
+09941
+67484
+09938
+10014
+10022
+67488
+10088
+10092
+10094
+10095
+11006
+11010
+11012
+11013
+11014
+11059
+67503
+11063
+11065
+11066
+11142
+11150
+11151
+11153
+11146
+11936
+11950
+11951
+11946
+11947
+12058
+12065
+12067
+12068
+12188
+12194
+12196
+12197
+12290
+12299
+12295
+12297
+13015
+13021
+13027
+13248
+13252
+13407
+13423
+13418
+13455
+13459
+13461
+13462
+13463
+13743
+67537
+13747
+13750
+14311
+14315
+14317
+14318
+14365
+14369
+14371
+14372
+68204
+14570
+14571
+15546
+15554
+15555
+67562
+15550
+15705
+15710
+15712
+16076
+67568
+16084
+16086
+16095
+16101
+16103
+16427
+70118
+16432
+16434
+16435
+16803
+16807
+16810
+16922
+16934
+16935
+16928
+17161
+17163
+67581
+17164
+17165
+17374
+17380
+17382
+17383
+17388
+17399
+67586
+17393
+17981
+70370
+67602
+17989
+17993
+17994
+18239
+67606
+18244
+18247
+18253
+67608
+18261
+18257
+18259
+18572
+18574
+18576
+18577
+18887
+67619
+18901
+18902
+18970
+67621
+18982
+18984
+19086
+19091
+19093
+19164
+19168
+19170
+19171
+19304
+67626
+19308
+19310
+19311
+19703
+19712
+67630
+19708
+19710
+19966
+19972
+19973
+20007
+70115
+20011
+20013
+20020
+20025
+20026
+20028
+20029
+20327
+20338
+20340
+20341
+21022
+21031
+21028
+21302
+68536
+21310
+67650
+21306
+21828
+67657
+21834
+23273
+67682
+23278
+68055
+68822
+67688
+23483
+23486
+23708
+67696
+23706
+23807
+23809
+23810
+24165
+24175
+24169
+24170
+24376
+24386
+67706
+24383
+24829
+24830
+24831
+67711
+24832
+24826
+24827
+24828
+24883
+67714
+24891
+24894
+25041
+25053
+25054
+25047
+25048
+65862
+67730
+25849
+25851
+26001
+26005
+26007
+26008
+26203
+26222
+26223
+26217
+26221
+26797
+26804
+26805
+27312
+27317
+27318
+27444
+27455
+27447
+27788
+68073
+67766
+27800
+27795
+27881
+27882
+67768
+27877
+28581
+28587
+28589
+28734
+28742
+28747
+28748
+29324
+70077
+29331
+29332
+29334
+29399
+29408
+29409
+29410
+29405
+29406
+29631
+29635
+29639
+30252
+68077
+68962
+30267
+30269
+30270
+30311
+30321
+30323
+30325
+30326
+30509
+67797
+30512
+30513
+30515
+30706
+30715
+67800
+30717
+30718
+31047
+31053
+31055
+31056
+31481
+31494
+31495
+31488
+31553
+67812
+31558
+31560
+32121
+32122
+32123
+32119
+32473
+67832
+32477
+32479
+32480
+32540
+32543
+32545
+32858
+68556
+32865
+32866
+33343
+33344
+67848
+33340
+33677
+33703
+67854
+33699
+34086
+34094
+67863
+34093
+34109
+34113
+34115
+34116
+70028
+34272
+34275
+34270
+34352
+67865
+34356
+34358
+34365
+34379
+34380
+34381
+34372
+35603
+35608
+35610
+36228
+68101
+36239
+36233
+36236
+36238
+70023
+36637
+67900
+36634
+37013
+37025
+67910
+37023
+37024
+37061
+67912
+37064
+37066
+37067
+37343
+37351
+37352
+37353
+37348
+37610
+67923
+37615
+37836
+37840
+37843
+37844
+38222
+38239
+38240
+38241
+38235
+39343
+39351
+39353
+39354
+39959
+39961
+39954
+40629
+40641
+40649
+67953
+40653
+40655
+40656
+40767
+40768
+40762
+40763
+40766
+40797
+67956
+40804
+40808
+40809
+41062
+67963
+41067
+41069
+41163
+67964
+41167
+41168
+41169
+41170
+41171
+42003
+42006
+67979
+42081
+42082
+42089
+42090
+42091
+42398
+42404
+42405
+42409
+42469
+42473
+42474
+42476
+42477
+42983
+70190
+42990
+42992
+42993
+43552
+43556
+43558
+43559
+43725
+43728
+43731
+43732
+44119
+44134
+44135
+44129
+44199
+44204
+44205
+44207
+44319
+44328
+44556
+67103
+44569
+44571
+44557
+44562
+44564
+44565
+44618
+44627
+44632
+44732
+44736
+44740
+44859
+44864
+44866
+45059
+45063
+45065
+45066
+45097
+45102
+45103
+45105
+45106
+45108
+45112
+45115
+45116
+45169
+45176
+45178
+45179
+45362
+67113
+45366
+45368
+45570
+67116
+45573
+45576
+45711
+45717
+45720
+45721
+45722
+45739
+45744
+45748
+45966
+45970
+45972
+45973
+45974
+46053
+67122
+46056
+46057
+46059
+46956
+46959
+46964
+47236
+47241
+47243
+47857
+67140
+47860
+47863
+47898
+47907
+47911
+48019
+48027
+48030
+48031
+48487
+70197
+48497
+48498
+48492
+48495
+48537
+67150
+48541
+48543
+48544
+67152
+48713
+48717
+48718
+49427
+49431
+49434
+49435
+49643
+67172
+49647
+49649
+49650
+49726
+49735
+49731
+49733
+49803
+49810
+67174
+49808
+50010
+50011
+50012
+50021
+50316
+50320
+50322
+50323
+50427
+50441
+50442
+50438
+50546
+50552
+50554
+51400
+67199
+51406
+51407
+51474
+67198
+51479
+51481
+51530
+51547
+51549
+51550
+51551
+51932
+51936
+51938
+51939
+52287
+52292
+52294
+52295
+52937
+67222
+52942
+52944
+52945
+53021
+67225
+53017
+53018
+53152
+53156
+53158
+53159
+53239
+70035
+53244
+53245
+53246
+53824
+67233
+53828
+53830
+53880
+53882
+54168
+54178
+54179
+54174
+54177
+54304
+54305
+67238
+54299
+54419
+54430
+54432
+54454
+68161
+69042
+67242
+54460
+54462
+54963
+54964
+54958
+55101
+55121
+55124
+55287
+55296
+55293
+55669
+55674
+55676
+55677
+55810
+67261
+55813
+55807
+55970
+55980
+55982
+55983
+55993
+70120
+55998
+56000
+56001
+56002
+56025
+56037
+67265
+56033
+56036
+56099
+67267
+56109
+56110
+67271
+56250
+56252
+56253
+56340
+70178
+69010
+56345
+56346
+56347
+56460
+56468
+56469
+56466
+56969
+56974
+56976
+56977
+57089
+57098
+57094
+57097
+57235
+67286
+57240
+57242
+57408
+67289
+57414
+57416
+57724
+57733
+57727
+57731
+57732
+57769
+68550
+57773
+57775
+57776
+57837
+57846
+67295
+57841
+57921
+57924
+57926
+57927
+58000
+58015
+58007
+58131
+58134
+58135
+58138
+67304
+58184
+58187
+68169
+58432
+58437
+58945
+58950
+58952
+59044
+70327
+59054
+59056
+59057
+59102
+67318
+59109
+59847
+67327
+59850
+59853
+67331
+60222
+60223
+60226
+67332
+60306
+60307
+60310
+60311
+60378
+67333
+60382
+60384
+61364
+67025
+61366
+61367
+61369
+61878
+61885
+61887
+61888
+62422
+67043
+62425
+62426
+62428
+62583
+62777
+62779
+62607
+67048
+62616
+62617
+62753
+68604
+62755
+62758
+62759
+62760
+62826
+62836
+62832
+62834
+62835
+63494
+63503
+63500
+63501
+63605
+63611
+63613
+63650
+67077
+63655
+63658
+67005
+63852
+63854
+63855
+64041
+67007
+64045
+64047
+64444
+64450
+64455
+64456
+64449
+00429
+70167
+00432
+00434
+00662
+67343
+00665
+00667
+01149
+68958
+01161
+01242
+01251
+01247
+01248
+01249
+02143
+67359
+02146
+02147
+02150
+02607
+02612
+02613
+02615
+02696
+02705
+02700
+02702
+03100
+03104
+03106
+03107
+70258
+03452
+03455
+03514
+03518
+03520
+03521
+04579
+04587
+04589
+04583
+05667
+05686
+05687
+70116
+67403
+05795
+05797
+05869
+05876
+05880
+06346
+06378
+67412
+06622
+06627
+06630
+06723
+67418
+06735
+06737
+07169
+07170
+07171
+07172
+67425
+07164
+07574
+67438
+07577
+07579
+67442
+07679
+07681
+07788
+07789
+07783
+08169
+08171
+08172
+08173
+08441
+67454
+08444
+08446
+09129
+09487
+09491
+09492
+09494
+09517
+67478
+09520
+09522
+09738
+09739
+09741
+10305
+10312
+10314
+10335
+10345
+67493
+10344
+10514
+10524
+10525
+67496
+10519
+10655
+10650
+10685
+10690
+67500
+10788
+10790
+10791
+11087
+11090
+11095
+11376
+67507
+11381
+11383
+11440
+11442
+11443
+70251
+11472
+11474
+11475
+11497
+11502
+67509
+11530
+11532
+11533
+12011
+12015
+12017
+12019
+12071
+12075
+12077
+12078
+13040
+13045
+13047
+13048
+13513
+13533
+13535
+14253
+14257
+14263
+14265
+14266
+14579
+14581
+14582
+14583
+15209
+70161
+15216
+15218
+15292
+15300
+15295
+15473
+15478
+15480
+15576
+15585
+67564
+15586
+15581
+16155
+16167
+67571
+16162
+16165
+16225
+16239
+16236
+16237
+16326
+16329
+16332
+67570
+16350
+16353
+16356
+16357
+16382
+16385
+16387
+16388
+16389
+16529
+16565
+16568
+68356
+68037
+17303
+17895
+67597
+17898
+17899
+18108
+18117
+18111
+18115
+18116
+18418
+18421
+18719
+67615
+18722
+18725
+19044
+19056
+19053
+19055
+19218
+19227
+19228
+19225
+19611
+19618
+19620
+19621
+20197
+67633
+20199
+20200
+20202
+20203
+20621
+20630
+67638
+20624
+20626
+20765
+20769
+20771
+20772
+68234
+20927
+20930
+20931
+21008
+21012
+21014
+21104
+21486
+21490
+21492
+68946
+22249
+22242
+22408
+22414
+22453
+22458
+22460
+22479
+22488
+22484
+22486
+22487
+23002
+70319
+23011
+23012
+23008
+67683
+23201
+23204
+23205
+23799
+23855
+23838
+23843
+23845
+23846
+23900
+70044
+23906
+23909
+67699
+23910
+23986
+67701
+23989
+23991
+23992
+24252
+24256
+24257
+24514
+24517
+24520
+24679
+24684
+24686
+24744
+24745
+25166
+67720
+25170
+25172
+25568
+70248
+25575
+25939
+25943
+25944
+25947
+26079
+26084
+26085
+26087
+26088
+26906
+67749
+26918
+26958
+67751
+26961
+26964
+27085
+27086
+67753
+27081
+27093
+27098
+27100
+27101
+27185
+67756
+27189
+27191
+68860
+27378
+27380
+27381
+27432
+67763
+27437
+27438
+27440
+67762
+27636
+27638
+27639
+27664
+27670
+27672
+27673
+27821
+67767
+27825
+27828
+28248
+28262
+28265
+28266
+28386
+28396
+67780
+28391
+28875
+28885
+28879
+28880
+29617
+29618
+29620
+68308
+30177
+30180
+30256
+67795
+30259
+30261
+30564
+30567
+30749
+30752
+67801
+30754
+31092
+67806
+31097
+31098
+31251
+31259
+67808
+31262
+31571
+67813
+31567
+31975
+31980
+31983
+32552
+32556
+32558
+66028
+32726
+32730
+32732
+32822
+32826
+32829
+32869
+32879
+32882
+32883
+33373
+33378
+33765
+67856
+33769
+33807
+33809
+33810
+33811
+34396
+67868
+34399
+34401
+34508
+70208
+34513
+34515
+68097
+34546
+34548
+34549
+34698
+34708
+34709
+34704
+35207
+35213
+35215
+35267
+67879
+35547
+35549
+35550
+35886
+35891
+35892
+35962
+35970
+67895
+36120
+36122
+36123
+36413
+67898
+36418
+36420
+36421
+36712
+36716
+36718
+36719
+37266
+37276
+37277
+67921
+37486
+37487
+38828
+67938
+38843
+39359
+39363
+39365
+39366
+39377
+39380
+39381
+39383
+39384
+39424
+39429
+39431
+39672
+70169
+39842
+39844
+39873
+39876
+39880
+40049
+70183
+40053
+40055
+70053
+40401
+40406
+40407
+40576
+40580
+40582
+40583
+41477
+41484
+41486
+41970
+67977
+41978
+41980
+41981
+42551
+42556
+42558
+42564
+67980
+42568
+42570
+42577
+42591
+42596
+42587
+42662
+67983
+42674
+42676
+43110
+43119
+43120
+43118
+43375
+43383
+43386
+43695
+67091
+43699
+43701
+43702
+43808
+70365
+43811
+43813
+44232
+68388
+44236
+44239
+44440
+44444
+44446
+44447
+44952
+44954
+44955
+45323
+45326
+45328
+45329
+45447
+45452
+45454
+45869
+45871
+45872
+46326
+70354
+46335
+46332
+46334
+46587
+46591
+46594
+46595
+46862
+46866
+46869
+47037
+67130
+47041
+47043
+47295
+47303
+47304
+47306
+47329
+70126
+47334
+47336
+47348
+47366
+47360
+47361
+47363
+47364
+47371
+47386
+47449
+47452
+47454
+47455
+47456
+47638
+67137
+47729
+47735
+47837
+47845
+47849
+47842
+48164
+67144
+48170
+48172
+48304
+67146
+48375
+48377
+48403
+66412
+48407
+48409
+48556
+67151
+48559
+48561
+48562
+48885
+48892
+48895
+48896
+49148
+49151
+49153
+49154
+49155
+49317
+49321
+49322
+49323
+49935
+49937
+49938
+49939
+50243
+50247
+50248
+50250
+50397
+50400
+50401
+50403
+50528
+50533
+50535
+50536
+50619
+50621
+50622
+50825
+67184
+50829
+50831
+50918
+50950
+50953
+50956
+50959
+50961
+50962
+70072
+51200
+51201
+51202
+51203
+51204
+51657
+51665
+67200
+51661
+51737
+51732
+51733
+51787
+67205
+51805
+51808
+51956
+51958
+51959
+52081
+52084
+52085
+52088
+52102
+52106
+52109
+52137
+52145
+52146
+52149
+52150
+70289
+52347
+52350
+52351
+52440
+52431
+52432
+52602
+52605
+52607
+52608
+52763
+67219
+52766
+52768
+52769
+52792
+67220
+52801
+53664
+53669
+53673
+54350
+67239
+54351
+54352
+54607
+54611
+54615
+67251
+55542
+67259
+55538
+55552
+55556
+55558
+55559
+55633
+55639
+55642
+55643
+56046
+56050
+56053
+56288
+67272
+56283
+56285
+56286
+67273
+56389
+56391
+56392
+56393
+56422
+56428
+56430
+56781
+56785
+56786
+56870
+56877
+56879
+56880
+57203
+67285
+57207
+57209
+57210
+57482
+57491
+57488
+57611
+57614
+57615
+57616
+57873
+57882
+57877
+57881
+58273
+58277
+58279
+58280
+58574
+58576
+58577
+58578
+58581
+58619
+58622
+58624
+58625
+58886
+58912
+58914
+58893
+67315
+58898
+58900
+59028
+59035
+59036
+59038
+59070
+59081
+59075
+59225
+59234
+59236
+59230
+59722
+67325
+59726
+59728
+59734
+59742
+59743
+59740
+59783
+67326
+59794
+59796
+59827
+59832
+59835
+60500
+70168
+60530
+60525
+61439
+67026
+61445
+61447
+61448
+62187
+62198
+62200
+62202
+62515
+62520
+62522
+62523
+62569
+67046
+62573
+62576
+62817
+62820
+62822
+62823
+63116
+63123
+63125
+63308
+63313
+63315
+63376
+63379
+63382
+63555
+67074
+63557
+63561
+63562
+63688
+70112
+63692
+63820
+63823
+63825
+63826
+63884
+63888
+63890
+64000
+64006
+64010
+64011
+64012
+64002
+00295
+01611
+01613
+01614
+02123
+00333
+00337
+00340
+00374
+00378
+00383
+67344
+00691
+00693
+67345
+00857
+00867
+00870
+00873
+00886
+68450
+00893
+01016
+01020
+67348
+01021
+01322
+01325
+01372
+01375
+01379
+01472
+01475
+01478
+01504
+01513
+01514
+01509
+01596
+01601
+01603
+01604
+01651
+01644
+01645
+01828
+01833
+01837
+01838
+01873
+01877
+01879
+01921
+69090
+01925
+02275
+67358
+02284
+02556
+02562
+02863
+02866
+02869
+02914
+02916
+02968
+02972
+02974
+03067
+03071
+03072
+03150
+03157
+03159
+03199
+03203
+03206
+03380
+03385
+03387
+03489
+03492
+03494
+03495
+03752
+03760
+03762
+03992
+04001
+04005
+04097
+04103
+04105
+04225
+04229
+04231
+04323
+04328
+04329
+04354
+04357
+04360
+04466
+06121
+04467
+06123
+04530
+04534
+04535
+04536
+05366
+05370
+05371
+05373
+05776
+05850
+06139
+06143
+06144
+06145
+68008
+68752
+69076
+06217
+06602
+06616
+06619
+07032
+67422
+07035
+07037
+07467
+67435
+07501
+07503
+07768
+07770
+07771
+07883
+07884
+07876
+67448
+08109
+08112
+08113
+08181
+08186
+08467
+08471
+08473
+08532
+67455
+08534
+08537
+70081
+67456
+08548
+08630
+08634
+08638
+08640
+08642
+08646
+08648
+67464
+08828
+08829
+08978
+65302
+67472
+08984
+09215
+70201
+09224
+09217
+09219
+09222
+09620
+09630
+09710
+09714
+10004
+10009
+10012
+10603
+10606
+10810
+10815
+10816
+11839
+11879
+12099
+12107
+12108
+12102
+12219
+12213
+12400
+67521
+12405
+12407
+12411
+12419
+12420
+12607
+12610
+12615
+12618
+12648
+12655
+12674
+12677
+12679
+12694
+12703
+12697
+12699
+12806
+12813
+12986
+12988
+12991
+12992
+12993
+13604
+70157
+13622
+13624
+13725
+13736
+14055
+67544
+14050
+14125
+14130
+14132
+14133
+14156
+14160
+14161
+14164
+14738
+14740
+14732
+15017
+15021
+15023
+15118
+15114
+15115
+15143
+15144
+15182
+15190
+15192
+15197
+15634
+15636
+15637
+15647
+70111
+15651
+15840
+15846
+15847
+15848
+15901
+15902
+15905
+15906
+16001
+16006
+16008
+16267
+16271
+16273
+16392
+70374
+16399
+16634
+16645
+16699
+16705
+16706
+16708
+16980
+16985
+16987
+16988
+17031
+17357
+17358
+17360
+67584
+17363
+17455
+17459
+17461
+17462
+18373
+18375
+18376
+18424
+18430
+18431
+18477
+18478
+18480
+18481
+18551
+18556
+19187
+19188
+19189
+19190
+67622
+19195
+19196
+19199
+67623
+19206
+19392
+67627
+19396
+19398
+19502
+19508
+19510
+19653
+19658
+19659
+19739
+67632
+19744
+19747
+19812
+19817
+70040
+69070
+20310
+20321
+20317
+20320
+20601
+20635
+67639
+20664
+20666
+20860
+20871
+21085
+21088
+21090
+21091
+21355
+67651
+21359
+21361
+21368
+21372
+21374
+21375
+21816
+21823
+21825
+22509
+22513
+22515
+22516
+22609
+22613
+22840
+67676
+22845
+22847
+67679
+22931
+23237
+23241
+23245
+23395
+23414
+23415
+23411
+23469
+23472
+24013
+24018
+24021
+67705
+24133
+24134
+24137
+24138
+24267
+65805
+24270
+24273
+24328
+24371
+24373
+24374
+24597
+24599
+24601
+24978
+24995
+24997
+67717
+68668
+25424
+25425
+25426
+25503
+67727
+25513
+25869
+68063
+25873
+25875
+26010
+67736
+26014
+26017
+26135
+69062
+67738
+26140
+26141
+26584
+68066
+26592
+27536
+27537
+27539
+27540
+27541
+27552
+27554
+27555
+27558
+27835
+27848
+27849
+28412
+28418
+28419
+28673
+28678
+28682
+29275
+29280
+29282
+29283
+29561
+29567
+29789
+29795
+29796
+29797
+29841
+29845
+29848
+29961
+29964
+29966
+29976
+29981
+29984
+30215
+30218
+30220
+30221
+30222
+30277
+30287
+30288
+30524
+30525
+30527
+30870
+30881
+30877
+30917
+30922
+30924
+31179
+31184
+31185
+31345
+31349
+31351
+31441
+67811
+31444
+31446
+31447
+31456
+31459
+31461
+31462
+31588
+31595
+31599
+31832
+67819
+31835
+31837
+31874
+31880
+31883
+31884
+31885
+32189
+67825
+33169
+33171
+33235
+33244
+33402
+33405
+33406
+33409
+33422
+67850
+33425
+33427
+33650
+33654
+33656
+33657
+34686
+34689
+34691
+35188
+35197
+35192
+35196
+35573
+35577
+35579
+35769
+35764
+35768
+35796
+35806
+67888
+35805
+35922
+35929
+35930
+36461
+36455
+36743
+36746
+36748
+36749
+70092
+37005
+37007
+37669
+37688
+37689
+38042
+38046
+38298
+38302
+38303
+38304
+38397
+38402
+38642
+38645
+38647
+38972
+38976
+38978
+39114
+67941
+39119
+39371
+39373
+39374
+39674
+39681
+39685
+40226
+40233
+40236
+40568
+40562
+40658
+67955
+40662
+40701
+42512
+42514
+42515
+41054
+67962
+41057
+41059
+41077
+41081
+41083
+41093
+41096
+41098
+41579
+68128
+68872
+41583
+41643
+67969
+41655
+41658
+41692
+41695
+41697
+42036
+42043
+67978
+42213
+42216
+42218
+42219
+42316
+42317
+42311
+42378
+70138
+42380
+42382
+42543
+42546
+42663
+42666
+42668
+42669
+42938
+67992
+42946
+43087
+43089
+43082
+43253
+43259
+43260
+43364
+43370
+43372
+43593
+43597
+43799
+43803
+43805
+43806
+43995
+43992
+44010
+44003
+44004
+44096
+44099
+44102
+44137
+44150
+44152
+44140
+67096
+44142
+44144
+44145
+44213
+68490
+44222
+44225
+44226
+44394
+44399
+44401
+44540
+44542
+44543
+44582
+44589
+44592
+44593
+44781
+44789
+44792
+44795
+67109
+44908
+44911
+44912
+45050
+45053
+45055
+45056
+45089
+45093
+45095
+45823
+45825
+46489
+46494
+46496
+46505
+46750
+46761
+46757
+46773
+46794
+46796
+46774
+46778
+46781
+47025
+47031
+47033
+47058
+67131
+47061
+47064
+47220
+67133
+47223
+47269
+47275
+47278
+47439
+47443
+47445
+47552
+47556
+47559
+47647
+47649
+47650
+47651
+48072
+48075
+48076
+48078
+48264
+48275
+48276
+48278
+48339
+48342
+48343
+48345
+48687
+48689
+48690
+49005
+49003
+48998
+49341
+49344
+49346
+49348
+49357
+49361
+49363
+49364
+49666
+49670
+49671
+49674
+50356
+50371
+50449
+50455
+50457
+50703
+70307
+50706
+50708
+50809
+50816
+50818
+50890
+50903
+67186
+50898
+51163
+51157
+51273
+67194
+51277
+51280
+51725
+67203
+51747
+51753
+51756
+51873
+51875
+52045
+52048
+52050
+52053
+67208
+52074
+52075
+52156
+52165
+52213
+52219
+52220
+52372
+52380
+52376
+52379
+52659
+52664
+52666
+52683
+52690
+52693
+52873
+52875
+52876
+52990
+52998
+52995
+53217
+70127
+53219
+53221
+53222
+53260
+53263
+53266
+53346
+53348
+53461
+53462
+68750
+53567
+53569
+53611
+53624
+53625
+53733
+53739
+53743
+53774
+53777
+54001
+54011
+67234
+54006
+54120
+54123
+54362
+54367
+54369
+54529
+54532
+54534
+54535
+54761
+54765
+54767
+54788
+54789
+54782
+55190
+55192
+55193
+55194
+56072
+67266
+56074
+56075
+56077
+56270
+56275
+56277
+56279
+56368
+56369
+56374
+56480
+67274
+56483
+56487
+56765
+56770
+68816
+68164
+56774
+57171
+57173
+57174
+70056
+57328
+57367
+57370
+57599
+57602
+57705
+69026
+67298
+57711
+57742
+57753
+57755
+57743
+57747
+57749
+57899
+57900
+57894
+58070
+58079
+58080
+58074
+58251
+58255
+58257
+58610
+58615
+58819
+58825
+58828
+58833
+58837
+58839
+58968
+67316
+58978
+59131
+59134
+59136
+59137
+59138
+59348
+59351
+59355
+59379
+59385
+59387
+59388
+60056
+60060
+60061
+68438
+60157
+60162
+60720
+60729
+61232
+61240
+61241
+61238
+61239
+61825
+61828
+61830
+62086
+70171
+67035
+62091
+62093
+62142
+62147
+62149
+62193
+62194
+67037
+62232
+62235
+62343
+62344
+62345
+62346
+62347
+62348
+62794
+62795
+62938
+62941
+62951
+67060
+62954
+62956
+63597
+63601
+63602
+63687
+67079
+63698
+63779
+63782
+63784
+63916
+63919
+63921
+63972
+63976
+00581
+00587
+67346
+00850
+00950
+01062
+01067
+01069
+01075
+01077
+01078
+01102
+01106
+01124
+01121
+01224
+01228
+01230
+01263
+01267
+01269
+67353
+01625
+01628
+02011
+02013
+02432
+02433
+02484
+02488
+02491
+02634
+02652
+02646
+02850
+02853
+02855
+03089
+03094
+03311
+03317
+03320
+03750
+03767
+03797
+03801
+03804
+04049
+04052
+04054
+04128
+04131
+04133
+04135
+04138
+04140
+04204
+04205
+04207
+04216
+04220
+04222
+04300
+04303
+04305
+04363
+04367
+04370
+04371
+04374
+04377
+04379
+68886
+67388
+04973
+04975
+05016
+05024
+05215
+05218
+05220
+70094
+67395
+05363
+05364
+05986
+05991
+05997
+06000
+06077
+06298
+67409
+06301
+06790
+06793
+06924
+06926
+06927
+06967
+06971
+06974
+07115
+07120
+07116
+07117
+07126
+70227
+07129
+07425
+07432
+07434
+07516
+07511
+07513
+07523
+07529
+07531
+07611
+70093
+67441
+07627
+07629
+07701
+07704
+07705
+07715
+07717
+07719
+07720
+07860
+07863
+07865
+08091
+08094
+08300
+08304
+08306
+08342
+08511
+08516
+08776
+08778
+08779
+08888
+08892
+08971
+09079
+09082
+09083
+09084
+09256
+09263
+09504
+09506
+10249
+10253
+10254
+10295
+10299
+10388
+67494
+10390
+10392
+10608
+10610
+10613
+10673
+10681
+10683
+11106
+11107
+11112
+11292
+11288
+11412
+11415
+11421
+11423
+11688
+11691
+11692
+11929
+11932
+11934
+11974
+11977
+11996
+11998
+11999
+12004
+12007
+12022
+12026
+12160
+12162
+12163
+12387
+12393
+12395
+12396
+67523
+12639
+12640
+12749
+12757
+12760
+12822
+12825
+12857
+12858
+12861
+12941
+12944
+12946
+13089
+13090
+13110
+13112
+13113
+13222
+13225
+13226
+13227
+13498
+13502
+13507
+67533
+13511
+13821
+13823
+67541
+13879
+13883
+67543
+14012
+14014
+14101
+14104
+14106
+14205
+14207
+14209
+14211
+14390
+67547
+14392
+14394
+14395
+14431
+14435
+14436
+15997
+14553
+15999
+14601
+14605
+14607
+14634
+67551
+14637
+14644
+14649
+14655
+14772
+14775
+14776
+67554
+14956
+14960
+14965
+14967
+14980
+14982
+14983
+15397
+15404
+15407
+15447
+15451
+15452
+15691
+15696
+15698
+15741
+15752
+15874
+70199
+15877
+15879
+15956
+67566
+15961
+15962
+67569
+16112
+16116
+16211
+16214
+16219
+16313
+16318
+16505
+16510
+16540
+16543
+16545
+70110
+16637
+16638
+16874
+16879
+16949
+16956
+16957
+17208
+17213
+17511
+17520
+17513
+17518
+17531
+17533
+17534
+17535
+17561
+17563
+17885
+17888
+17879
+17958
+17978
+18039
+18041
+18072
+18074
+18075
+18090
+18093
+18094
+18107
+18534
+18536
+18640
+18643
+18646
+18647
+18731
+18733
+18755
+18758
+18760
+18854
+18857
+18859
+18860
+18939
+18948
+18950
+19198
+19216
+19301
+19302
+67629
+19516
+19851
+19854
+19929
+19932
+19909
+19913
+19916
+20108
+20114
+20258
+20259
+20261
+20455
+20458
+20460
+20508
+20514
+20515
+20670
+21569
+21571
+20840
+20841
+20842
+20965
+20970
+20971
+20972
+68588
+21526
+21528
+21837
+21840
+21842
+68049
+68648
+21926
+21929
+22017
+22018
+67664
+22012
+21998
+22000
+22193
+22196
+22356
+22361
+22363
+22775
+22776
+22973
+22977
+23029
+23034
+23432
+23426
+23431
+23649
+23651
+23959
+24524
+24526
+24042
+24045
+24047
+24199
+67704
+24201
+24346
+24348
+24501
+24539
+24551
+24554
+25270
+25278
+25281
+25352
+25355
+25521
+25529
+68502
+68062
+25715
+25717
+25719
+26065
+26066
+26068
+26074
+27358
+27356
+65877
+26466
+26467
+26468
+68948
+26486
+26488
+26934
+26936
+27017
+27021
+27485
+27489
+27490
+27491
+27752
+27747
+27915
+27941
+27944
+28050
+28059
+28088
+28089
+28285
+28290
+28295
+28557
+28790
+28793
+28823
+28810
+28826
+28811
+28816
+67787
+29009
+29461
+29464
+29467
+29899
+29902
+29922
+29925
+70182
+30300
+30302
+30582
+30588
+30589
+67798
+30600
+30601
+30820
+31101
+31103
+31072
+31216
+31219
+31222
+31234
+31238
+31297
+31302
+31304
+31436
+31584
+31586
+67814
+31620
+31690
+31692
+32048
+32051
+32092
+32096
+32100
+67824
+32104
+32275
+32279
+32419
+32423
+32426
+32519
+32524
+32561
+32565
+32566
+32567
+32576
+32573
+32752
+32756
+32758
+32902
+32914
+32917
+32999
+33004
+33041
+33045
+33046
+33061
+33065
+33067
+33068
+33517
+68094
+68596
+33520
+33607
+33610
+33613
+33615
+33848
+67859
+33855
+33856
+33858
+33880
+68976
+67860
+33883
+33979
+33982
+33985
+34336
+35898
+35900
+35309
+35310
+35311
+35312
+35313
+35554
+35559
+35566
+35569
+35571
+35673
+35681
+35678
+35680
+67891
+35934
+35936
+35994
+35991
+35993
+36036
+36039
+36041
+36042
+36141
+36144
+36145
+36398
+36401
+36403
+36487
+70202
+67899
+36491
+36803
+36806
+36807
+36907
+36908
+36904
+36974
+37302
+37307
+37309
+37512
+38266
+38268
+37644
+37649
+37651
+37827
+37830
+37832
+37930
+37935
+37937
+37938
+38206
+38213
+67930
+38375
+38377
+38700
+38703
+38705
+38948
+68530
+67936
+38874
+39068
+39069
+39074
+39078
+39081
+39147
+39554
+39556
+39197
+39200
+39202
+39288
+39290
+39333
+39336
+39338
+39339
+39386
+39391
+39646
+39648
+39650
+39651
+40093
+40246
+40241
+40243
+40094
+67948
+40100
+40340
+40344
+40346
+40527
+40530
+40725
+40727
+40827
+40832
+41256
+41264
+41266
+41560
+41561
+41704
+41707
+41709
+41710
+41771
+41776
+41779
+41810
+41813
+41815
+41895
+41901
+67973
+41903
+41904
+41906
+42136
+42144
+42145
+42508
+42601
+42610
+42611
+42760
+42763
+42766
+43146
+43148
+43150
+43509
+43512
+67087
+43542
+43673
+43676
+43882
+43883
+43876
+43880
+43881
+43944
+43947
+44018
+44021
+44022
+44166
+44171
+44181
+44183
+44187
+44189
+44296
+44297
+44300
+44302
+67101
+44489
+44522
+44525
+44527
+44528
+44530
+44533
+44535
+44536
+44548
+44553
+44714
+44718
+44720
+44810
+44813
+44816
+44842
+44846
+44848
+44957
+44959
+44960
+44961
+44963
+44969
+44972
+44974
+45163
+67112
+45165
+45167
+45186
+45197
+45222
+45412
+45414
+45415
+45421
+45422
+45424
+45467
+45470
+45472
+45524
+45527
+45595
+45598
+45600
+45636
+45641
+45651
+45654
+45657
+45672
+45666
+45829
+47984
+47986
+45927
+45930
+45935
+46164
+46167
+46172
+46310
+46316
+46318
+46576
+46577
+46579
+46605
+46608
+46610
+46656
+46929
+46931
+46932
+47079
+47082
+47092
+47135
+47141
+47341
+47344
+47346
+47478
+47481
+47483
+47565
+47571
+47573
+47574
+47721
+47723
+47725
+47726
+47802
+47808
+47810
+48297
+48300
+48302
+48305
+48312
+70344
+48524
+48568
+48571
+48729
+48733
+48735
+48933
+51427
+51429
+49088
+49091
+49412
+67163
+49418
+49408
+67166
+49461
+49462
+49464
+49776
+49782
+49784
+49785
+49884
+49901
+49904
+50192
+50193
+50194
+50200
+50465
+50474
+50578
+50583
+50586
+51299
+51302
+51304
+51305
+51312
+67196
+51314
+51316
+51771
+51828
+51830
+51795
+51797
+51799
+51985
+52114
+68470
+67209
+52125
+52127
+52180
+52182
+52183
+67218
+52652
+52653
+52654
+52655
+53035
+53040
+53042
+53122
+53124
+53125
+53177
+53181
+53183
+53201
+53202
+53228
+53232
+53234
+53251
+53254
+53256
+53402
+53407
+53848
+53853
+53856
+53917
+53923
+53925
+66549
+54091
+54094
+54141
+54188
+54192
+54194
+54253
+54661
+54665
+54666
+54668
+54670
+54804
+54806
+54808
+54809
+54829
+67247
+54831
+54838
+54842
+54844
+55322
+55325
+55327
+55626
+70160
+55630
+55865
+55870
+55914
+55915
+55916
+56113
+56116
+56118
+56168
+56177
+56179
+56358
+56546
+58324
+58326
+56788
+56794
+56797
+57254
+57256
+57344
+57350
+57352
+57353
+57394
+57396
+57361
+57379
+57381
+57382
+57542
+57551
+57566
+57569
+57571
+57572
+67291
+57604
+57607
+57609
+57818
+57821
+57823
+57824
+58236
+58246
+58663
+67312
+58666
+58743
+58747
+58749
+58768
+58771
+58773
+59000
+59004
+59006
+59120
+59125
+59668
+59671
+59841
+59845
+59969
+59989
+59993
+68518
+60072
+67329
+60073
+60074
+60338
+60359
+60361
+60701
+60966
+60969
+60972
+60973
+61130
+61562
+61564
+61134
+61137
+61139
+61187
+61190
+61192
+61219
+61223
+61225
+61248
+61250
+61511
+61587
+61591
+61594
+61696
+61698
+61845
+61849
+61871
+61872
+61875
+61946
+67030
+62032
+63275
+63277
+62384
+70137
+62487
+62491
+62781
+70104
+67053
+62784
+63045
+63048
+63049
+63185
+67065
+63187
+63189

ai_model/words/realtime_translator.py

@@ -0,0 +1,319 @@
+import cv2
+import mediapipe as mp
+import numpy as np
+import pandas as pd
+import os
+from tensorflow.keras.models import load_model
+import time
+
+# --- Configuration (must match training parameters) ---
+MODEL_PATH = 'saved_models/best_sign_classifier_model_125_words_seq90.keras'
+PROCESSED_DATA_CSV = 'wlasl_125_words_personal_final_processed_data_augmented_seq90.csv'
+
+SEQUENCE_LENGTH = 90
+EXPECTED_COORDS_PER_FRAME = 1662
+
+# Define the expected number of coordinates for each type of landmark
+NUM_POSE_COORDS_SINGLE = 33 * 4
+NUM_HAND_COORDS_SINGLE = 21 * 3
+NUM_FACE_COORDS_SINGLE = 468 * 3
+
+# --- Real-time Translation Specific Parameters ---
+RECORDING_DURATION_SECONDS = 3.6
+CONFIDENCE_THRESHOLD = 0.50 # Minimum confidence for a prediction to be displayed
+
+# --- Automatic Detection & String Building Parameters ---
+MOTION_THRESHOLD = 0.05 # How much landmark movement is needed to trigger a recording
+COOLDOWN_DURATION = 2.0 # Seconds to wait after a prediction before looking for a new sign
+TIME_BETWEEN_WORDS = 1.0 # Minimum time between adding new, distinct words to the sentence
+SENTENCE_CLEAR_DURATION = 8.0 # Time of inactivity before the sentence is cleared
+
+# States for the application
+STATE_IDLE = "IDLE (Signing will auto-record)"
+STATE_RECORDING = "RECORDING..."
+STATE_PREDICTING = "PREDICTING..."
+STATE_RESULT = "" # No longer needed as a state string
+STATE_COOLDOWN = "COOLDOWN..."
+
+# --- Utility Functions (Copied/Adapted from data_preprocessor.py) ---
+def normalize_landmarks(landmarks_sequence):
+    """
+    Normalizes a sequence of landmarks (frames, coords) to be translation and scale invariant.
+    Handles single frames (ndim=1) or multiple frames (ndim=2).
+    """
+    if landmarks_sequence.size == 0:
+        return landmarks_sequence.astype(np.float32)
+
+    if landmarks_sequence.ndim == 1:
+        input_is_single_frame = True
+        landmarks_sequence_2d = np.expand_dims(landmarks_sequence, axis=0)
+    else:
+        input_is_single_frame = False
+        landmarks_sequence_2d = landmarks_sequence
+
+    normalized_sequences = []
+    for frame_landmarks in landmarks_sequence_2d:
+        if np.all(frame_landmarks == 0):
+            normalized_sequences.append(np.zeros(EXPECTED_COORDS_PER_FRAME, dtype=np.float32))
+            continue
+
+        pose_coords_flat = frame_landmarks[0 : NUM_POSE_COORDS_SINGLE]
+        left_hand_coords_flat = frame_landmarks[NUM_POSE_COORDS_SINGLE : NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE]
+        right_hand_coords_flat = frame_landmarks[NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE : NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE]
+        face_coords_flat = frame_landmarks[NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE : ]
+        
+        all_parts_data = [
+            (pose_coords_flat, 4, [0.0] * NUM_POSE_COORDS_SINGLE),
+            (left_hand_coords_flat, 3, [0.0] * NUM_HAND_COORDS_SINGLE),
+            (right_hand_coords_flat, 3, [0.0] * NUM_HAND_COORDS_SINGLE),
+            (face_coords_flat, 3, [0.0] * NUM_FACE_COORDS_SINGLE)
+        ]
+        
+        normalized_frame_parts = []
+        for flat_lms, coords_per_lm, original_padded_list_template in all_parts_data:
+            if np.all(flat_lms == 0):
+                normalized_frame_parts.append(np.array(original_padded_list_template, dtype=np.float32))
+                continue
+            
+            lms_array = flat_lms.reshape(-1, coords_per_lm)
+            coords_for_mean = lms_array[:, :3] if coords_per_lm == 4 else lms_array
+            
+            if np.all(coords_for_mean == 0):
+                 normalized_frame_parts.append(np.array(original_padded_list_template, dtype=np.float32))
+                 continue
+
+            mean_coords = np.mean(coords_for_mean, axis=0)
+            translated_lms = lms_array.copy()
+            translated_lms[:, :3] -= mean_coords
+
+            scale_factor = np.max(np.linalg.norm(translated_lms[:, :3], axis=1))
+            if scale_factor > 1e-6:
+                translated_lms[:, :3] /= scale_factor
+            
+            normalized_frame_parts.append(translated_lms.flatten())
+            
+        combined_normalized_frame = np.concatenate(normalized_frame_parts).astype(np.float32)
+        
+        if len(combined_normalized_frame) < EXPECTED_COORDS_PER_FRAME:
+            combined_normalized_frame = np.pad(combined_normalized_frame, (0, EXPECTED_COORDS_PER_FRAME - len(combined_normalized_frame)), 'constant', constant_values=0.0)
+        elif len(combined_normalized_frame) > EXPECTED_COORDS_PER_FRAME:
+            combined_normalized_frame = combined_normalized_frame[:EXPECTED_COORDS_PER_FRAME]
+
+        normalized_sequences.append(combined_normalized_frame)
+            
+    result_array = np.array(normalized_sequences, dtype=np.float32)
+    
+    if input_is_single_frame:
+        return result_array[0]
+    else:
+        return result_array
+
+def pad_or_truncate_sequence(sequence, target_length, feature_dimension):
+    """Pads or truncates a sequence to a target_length."""
+    if sequence.shape[0] < target_length:
+        padding = np.zeros((target_length - sequence.shape[0], feature_dimension), dtype=np.float32)
+        padded_sequence = np.vstack((sequence, padding))
+    elif sequence.shape[0] > target_length:
+        padded_sequence = sequence[:target_length, :]
+    else:
+        padded_sequence = sequence
+    return padded_sequence
+
+def calculate_motion(current_lms, previous_lms):
+    """Calculates the total change in landmark positions to detect motion."""
+    if previous_lms is None or np.all(previous_lms == 0):
+        return 0.0
+    
+    pose_lms_change = np.linalg.norm(current_lms[0:33*4:4] - previous_lms[0:33*4:4])
+    hands_lms_change = np.linalg.norm(current_lms[33*4:] - previous_lms[33*4:])
+    
+    return pose_lms_change + hands_lms_change
+
+# --- Main Real-time Translator Logic ---
+if __name__ == "__main__":
+    if not os.path.exists(MODEL_PATH):
+        print(f"Error: Model not found at {MODEL_PATH}. Please ensure your training completed and saved the model.")
+        exit()
+
+    print(f"Loading model from: {MODEL_PATH}")
+    model = load_model(MODEL_PATH)
+    model.summary()
+
+    if not os.path.exists(PROCESSED_DATA_CSV):
+        print(f"Error: Processed data CSV not found at {PROCESSED_DATA_CSV}. Cannot map gloss IDs to names.")
+        exit()
+
+    df_final = pd.read_csv(PROCESSED_DATA_CSV)
+    unique_glosses = df_final['gloss'].unique()
+    id_to_gloss = {i: gloss for i, gloss in enumerate(unique_glosses)}
+    ordered_gloss_names = [id_to_gloss[i] for i in range(len(id_to_gloss))]
+    print(f"Loaded {len(ordered_gloss_names)} gloss mappings.")
+
+    mp_holistic = mp.solutions.holistic.Holistic(
+        static_image_mode=False,
+        model_complexity=1,
+        min_detection_confidence=0.5,
+        min_tracking_confidence=0.5
+    )
+
+    cap = cv2.VideoCapture(0)
+    if not cap.isOpened():
+        print("Error: Could not open webcam. Make sure no other apps are using it.")
+        exit()
+
+    print(f"\nReal-time translation started.")
+    print(f"Signing motion will automatically trigger a {RECORDING_DURATION_SECONDS}-second recording.")
+    print("Press 'Q' to quit.")
+    print("--------------------------------------------------")
+    print(f"Console will show predictions for all {len(ordered_gloss_names)} words.")
+    print("--------------------------------------------------")
+
+    current_state = STATE_IDLE
+    recording_start_time = 0
+    cooldown_start_time = 0
+    recorded_raw_landmarks_buffer = []
+    previous_frame_raw_landmarks = None
+
+    display_status_text = STATE_IDLE
+    display_confidence_text = ""
+    current_sentence = []
+    last_word_added = ""
+    last_activity_time = time.time()
+    
+    fps_start_time = time.time()
+    fps_frame_count = 0
+
+    while cap.isOpened():
+        ret, frame = cap.read()
+        if not ret:
+            print("Failed to grab frame.")
+            break
+        
+        # --- CAMERA FLIP ADJUSTMENT ---
+        # This line will flip the image horizontally.
+        # This creates an un-mirrored view, as if someone is facing you.
+        # frame = cv2.flip(frame, 1)
+        # --- END CAMERA FLIP ADJUSTMENT ---
+
+        image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+        image.flags.writeable = False
+        results = mp_holistic.process(image)
+        image.flags.writeable = True
+        frame = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+
+        # Draw landmarks (unchanged)
+        mp.solutions.drawing_utils.draw_landmarks(frame, results.pose_landmarks, mp.solutions.holistic.POSE_CONNECTIONS)
+        mp.solutions.drawing_utils.draw_landmarks(frame, results.left_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+        mp.solutions.drawing_utils.draw_landmarks(frame, results.right_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+        mp.solutions.drawing_utils.draw_landmarks(frame, results.face_landmarks, mp.solutions.holistic.FACEMESH_CONTOURS)
+
+        current_frame_raw_landmarks_flat = np.zeros(EXPECTED_COORDS_PER_FRAME, dtype=np.float32)
+        current_idx = 0
+        if results.pose_landmarks:
+            pose_lms_flat = [coord for landmark in results.pose_landmarks.landmark for coord in [landmark.x, landmark.y, landmark.z, landmark.visibility]]
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(pose_lms_flat)] = pose_lms_flat
+        current_idx += NUM_POSE_COORDS_SINGLE
+
+        if results.left_hand_landmarks:
+            lh_lms_flat = [coord for landmark in results.left_hand_landmarks.landmark for coord in [landmark.x, landmark.y, landmark.z]]
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(lh_lms_flat)] = lh_lms_flat
+        current_idx += NUM_HAND_COORDS_SINGLE
+
+        if results.right_hand_landmarks:
+            rh_lms_flat = [coord for landmark in results.right_hand_landmarks.landmark for coord in [landmark.x, landmark.y, landmark.z]]
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(rh_lms_flat)] = rh_lms_flat
+        current_idx += NUM_HAND_COORDS_SINGLE
+
+        if results.face_landmarks:
+            face_lms_flat = [coord for landmark in results.face_landmarks.landmark for coord in [landmark.x, landmark.y, landmark.z]]
+            current_frame_raw_landmarks_flat[current_idx : current_idx + len(face_lms_flat)] = face_lms_flat
+        
+        # --- NEW LOGIC FOR AUTOMATIC RECORDING & SENTENCE BUILDING ---
+        if current_state == STATE_IDLE or current_state == STATE_COOLDOWN:
+            motion = calculate_motion(current_frame_raw_landmarks_flat, previous_frame_raw_landmarks)
+            previous_frame_raw_landmarks = current_frame_raw_landmarks_flat
+            
+            if current_state == STATE_IDLE and motion > MOTION_THRESHOLD:
+                current_state = STATE_RECORDING
+                recording_start_time = time.time()
+                recorded_raw_landmarks_buffer = [] # Reset buffer at start of new sign
+                print("Recording started automatically...")
+
+        if current_state == STATE_RECORDING:
+            recorded_raw_landmarks_buffer.append(current_frame_raw_landmarks_flat)
+            elapsed_time = time.time() - recording_start_time
+            display_status_text = f"{STATE_RECORDING} ({int(elapsed_time)}/{RECORDING_DURATION_SECONDS}s)"
+            if elapsed_time >= RECORDING_DURATION_SECONDS:
+                current_state = STATE_PREDICTING
+
+        if current_state == STATE_PREDICTING:
+            print("\n--- NEW PREDICTION ---")
+            if not recorded_raw_landmarks_buffer:
+                print("Warning: No frames recorded for prediction.")
+            else:
+                processed_sequence = normalize_landmarks(np.array(recorded_raw_landmarks_buffer))
+                final_input_sequence = pad_or_truncate_sequence(processed_sequence, SEQUENCE_LENGTH, EXPECTED_COORDS_PER_FRAME)
+                final_input_sequence = np.expand_dims(final_input_sequence, axis=0)
+                predictions = model.predict(final_input_sequence, verbose=0)
+                all_scores = predictions[0]
+                
+                print("Confidence scores for all classes:")
+                for i in np.argsort(all_scores)[::-1]:
+                    gloss_name = ordered_gloss_names[i]
+                    score = all_scores[i]
+                    print(f"   {gloss_name.upper():<10}: {score:.4f}")
+
+                predicted_class_id = np.argmax(predictions)
+                prediction_confidence = predictions[0, predicted_class_id]
+                predicted_gloss = id_to_gloss.get(predicted_class_id, "Unknown")
+                
+                # Logic to add a word to the sentence
+                if prediction_confidence >= CONFIDENCE_THRESHOLD and predicted_gloss != last_word_added and (time.time() - last_activity_time) > TIME_BETWEEN_WORDS:
+                    current_sentence.append(predicted_gloss)
+                    last_word_added = predicted_gloss
+                    last_activity_time = time.time() # This timer now tracks the last activity, not just prediction
+
+                display_confidence_text = f"Confidence: {prediction_confidence:.2f}"
+            
+            recorded_raw_landmarks_buffer = []
+            cooldown_start_time = time.time()
+            current_state = STATE_COOLDOWN
+
+        if current_state == STATE_COOLDOWN:
+            elapsed_cooldown = time.time() - cooldown_start_time
+            display_status_text = f"Resuming in {COOLDOWN_DURATION - elapsed_cooldown:.1f}s"
+            if elapsed_cooldown > COOLDOWN_DURATION:
+                current_state = STATE_IDLE
+                display_status_text = STATE_IDLE
+        
+        # Clear the sentence if there's been no activity for a while
+        if current_sentence and (time.time() - last_activity_time) > SENTENCE_CLEAR_DURATION:
+            current_sentence = []
+            last_word_added = ""
+            display_status_text = "Sentence cleared."
+
+        # --- END NEW LOGIC ---
+
+        # Text on screen (UPDATED to show sentence and status)
+        sentence_text = " ".join(current_sentence).upper()
+        cv2.putText(frame, display_status_text, (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (0, 255, 0), 2, cv2.LINE_AA)
+        cv2.putText(frame, sentence_text, (10, 70), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255), 2, cv2.LINE_AA)
+        cv2.putText(frame, display_confidence_text, (10, 110), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (0, 165, 255), 2, cv2.LINE_AA)
+
+        fps_frame_count += 1
+        if time.time() - fps_start_time >= 1:
+            fps = fps_frame_count / (time.time() - fps_start_time)
+            cv2.putText(frame, f"FPS: {fps:.1f}", (frame.shape[1] - 150, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 255), 2, cv2.LINE_AA)
+            fps_frame_count = 0
+            fps_start_time = time.time()
+
+        cv2.imshow('Sign Language Translator', frame)
+
+        key = cv2.waitKey(1) & 0xFF
+        if key == ord('q'):
+            break
+
+    cap.release()
+    cv2.destroyAllWindows()
+    mp_holistic.close()
+    print("\nReal-time translation stopped.")

ai_model/words/record_personal_data.py

@@ -0,0 +1,166 @@
+import cv2
+import mediapipe as mp
+import numpy as np
+import os
+import time
+
+# --- Configuration ---
+# Directory to save recorded raw landmark data
+OUTPUT_RECORDINGS_DIR = 'my_recorded_signs'
+
+# MediaPipe parameters (should match what you used for data extraction)
+MP_MODEL_COMPLEXITY = 1
+MP_DETECTION_CONFIDENCE = 0.5
+MP_TRACKING_CONFIDENCE = 0.5
+
+# Recording parameters
+DEFAULT_RECORDING_DURATION_SECONDS = 3.6 # Match SEQUENCE_LENGTH / FPS (90 frames / 25 FPS)
+
+# Expected number of coordinates per frame (from previous steps)
+EXPECTED_COORDS_PER_FRAME = 1662
+
+# Define the expected number of coordinates for each type of landmark
+NUM_POSE_COORDS_SINGLE = 33 * 4
+NUM_HAND_COORDS_SINGLE = 21 * 3
+NUM_FACE_COORDS_SINGLE = 468 * 3
+
+
+def extract_raw_landmarks(results):
+    """
+    Extracts raw landmark data from MediaPipe results into a flat numpy array.
+    This logic should match how you populate current_frame_raw_landmarks_flat in realtime_translator.py
+    """
+    current_frame_raw_landmarks_flat = np.zeros(EXPECTED_COORDS_PER_FRAME, dtype=np.float32)
+    current_idx = 0
+
+    if results.pose_landmarks:
+        pose_lms_flat = []
+        for landmark in results.pose_landmarks.landmark:
+            pose_lms_flat.extend([landmark.x, landmark.y, landmark.z, landmark.visibility])
+        current_frame_raw_landmarks_flat[current_idx : current_idx + len(pose_lms_flat)] = pose_lms_flat
+    current_idx += NUM_POSE_COORDS_SINGLE
+
+    if results.left_hand_landmarks:
+        lh_lms_flat = []
+        for landmark in results.left_hand_landmarks.landmark:
+            lh_lms_flat.extend([landmark.x, landmark.y, landmark.z])
+        current_frame_raw_landmarks_flat[current_idx : current_idx + len(lh_lms_flat)] = lh_lms_flat
+    current_idx += NUM_HAND_COORDS_SINGLE
+
+    if results.right_hand_landmarks:
+        rh_lms_flat = []
+        for landmark in results.right_hand_landmarks.landmark:
+            rh_lms_flat.extend([landmark.x, landmark.y, landmark.z])
+        current_frame_raw_landmarks_flat[current_idx : current_idx + len(rh_lms_flat)] = rh_lms_flat
+    current_idx += NUM_HAND_COORDS_SINGLE
+
+    if results.face_landmarks:
+        face_lms_flat = []
+        for landmark in results.face_landmarks.landmark:
+            face_lms_flat.extend([landmark.x, landmark.y, landmark.z])
+        current_frame_raw_landmarks_flat[current_idx : current_idx + len(face_lms_flat)] = face_lms_flat
+    
+    return current_frame_raw_landmarks_flat
+
+
+if __name__ == "__main__":
+    os.makedirs(OUTPUT_RECORDINGS_DIR, exist_ok=True)
+
+    mp_holistic = mp.solutions.holistic.Holistic(
+        static_image_mode=False,
+        model_complexity=MP_MODEL_COMPLEXITY,
+        min_detection_confidence=MP_DETECTION_CONFIDENCE,
+        min_tracking_confidence=MP_TRACKING_CONFIDENCE
+    )
+
+    cap = cv2.VideoCapture(0)
+    if not cap.isOpened():
+        print("Error: Could not open webcam.")
+        exit()
+
+    print(f"\n--- Personal Sign Recording Tool ---")
+    print(f"Record duration: {DEFAULT_RECORDING_DURATION_SECONDS} seconds per sign.")
+    print(f"Recorded landmark sequences will be saved to: {OUTPUT_RECORDINGS_DIR}")
+    print("\nInstructions:")
+    print("1. Enter the ASL gloss (word) for the sign you will perform.")
+    print("2. Press ENTER.")
+    print("3. You will see a 3-second countdown.")
+    print("4. Perform the sign clearly during the recording (green background recommended).")
+    print("5. The recording will stop automatically. The raw landmarks will be saved.")
+    print("   If you see 'No motion detected', try again with more prominent movement.")
+    print("6. Repeat or type 'q' and press ENTER to quit.")
+
+    while True:
+        gloss_input = input("\nEnter ASL gloss for sign (e.g., BLACK, APPLE), or 'q' to quit: ").strip().lower()
+        if gloss_input == 'q':
+            break
+        if not gloss_input:
+            print("Gloss cannot be empty. Please enter a valid gloss.")
+            continue
+
+        print(f"\n--- Preparing to record '{gloss_input.upper()}' ---")
+        print("Starting countdown...")
+        countdown_start = time.time()
+        
+        while time.time() - countdown_start < 3:
+            ret, frame = cap.read()
+            if not ret: break
+            frame = cv2.flip(frame, 1)
+            display_countdown = 3 - int(time.time() - countdown_start)
+            cv2.putText(frame, f"Recording in {display_countdown}...", (50, 50), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 255), 2, cv2.LINE_AA)
+            cv2.imshow('Recording Tool', frame)
+            cv2.waitKey(1)
+        
+        print(f"Recording '{gloss_input.upper()}' NOW!")
+        recorded_landmarks = []
+        record_start_time = time.time()
+        motion_detected = False
+
+        while time.time() - record_start_time < DEFAULT_RECORDING_DURATION_SECONDS:
+            ret, frame = cap.read()
+            if not ret: break
+            # frame = cv2.flip(frame, 1)
+            
+            image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+            image.flags.writeable = False
+            results = mp_holistic.process(image)
+            image.flags.writeable = True
+            frame = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+
+            # Draw landmarks for visualization
+            mp.solutions.drawing_utils.draw_landmarks(frame, results.pose_landmarks, mp.solutions.holistic.POSE_CONNECTIONS)
+            mp.solutions.drawing_utils.draw_landmarks(frame, results.left_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+            mp.solutions.drawing_utils.draw_landmarks(frame, results.right_hand_landmarks, mp.solutions.holistic.HAND_CONNECTIONS)
+            mp.solutions.drawing_utils.draw_landmarks(frame, results.face_landmarks, mp.solutions.holistic.FACEMESH_CONTOURS)
+
+            # Extract raw landmarks for saving
+            current_frame_lms = extract_raw_landmarks(results)
+            recorded_landmarks.append(current_frame_lms)
+
+            # Basic motion detection (if not all zeros)
+            if np.any(current_frame_lms != 0):
+                motion_detected = True
+
+            cv2.putText(frame, f"Recording {gloss_input.upper()} ({int(time.time() - record_start_time)}s)", (50, 50), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2, cv2.LINE_AA)
+            cv2.imshow('Recording Tool', frame)
+            cv2.waitKey(1)
+
+        print(f"Recording of '{gloss_input.upper()}' finished.")
+        
+        if not motion_detected:
+            print("Warning: No significant motion (landmarks != 0) detected during recording. Consider trying again.")
+        
+        if recorded_landmarks:
+            recorded_landmarks_array = np.array(recorded_landmarks, dtype=np.float32)
+            timestamp = int(time.time())
+            filename = f"{gloss_input}_{timestamp}.npy"
+            filepath = os.path.join(OUTPUT_RECORDINGS_DIR, filename)
+            np.save(filepath, recorded_landmarks_array)
+            print(f"Saved {recorded_landmarks_array.shape[0]} frames to {filepath}")
+        else:
+            print("No frames were recorded.")
+
+    cap.release()
+    cv2.destroyAllWindows()
+    mp_holistic.close()
+    print("\nRecording tool stopped.")

ai_model/words/saved_models/best_sign_classifier_model.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52de70ef35cccb8f991643336ff99cd3438552fa9250d60e983cc8578ed9a185
+size 13009869

ai_model/words/saved_models/best_sign_classifier_model_10_words_combined_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d13f78523998a1419136db97ae7698fa325ebccc7ace714fd7ac2ed4a4a8a1e
+size 12105586

ai_model/words/saved_models/best_sign_classifier_model_10_words_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b7e04e04caf184c67b97cf385274db5daa9954237deaad010a0bbf8ad34404e
+size 12105586

ai_model/words/saved_models/best_sign_classifier_model_125_words_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:754fc46f776cb54f5c38b829c743ecdaea52548f16b2af975e1e58cdecb97807
+size 70114011

ai_model/words/saved_models/best_sign_classifier_model_20_words_optimized_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ca916295d4efaca4758cf19d464a32db26dbc687096a87ecd9f7b49748b4a96
+size 25240130

ai_model/words/saved_models/best_sign_classifier_model_20_words_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52d794ce11e75337d368917db588c5a7c5982779aeddaa03e182bdd4e7a1640a
+size 19301354

ai_model/words/saved_models/best_sign_classifier_model_30_words_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eec69ef523d2b8a3a8798d0153db728a067b85f1b77ff9130b83f6edeaf66eff
+size 27506025

ai_model/words/saved_models/best_sign_classifier_model_40_words_seq90.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ba02a78a445689a18b06a613f011357a2dcc248fa579441de8bc26ef7b4fc01
+size 27529185

ai_model/words/saved_models/best_sign_classifier_model_improved.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:096f35162a09f28f99e298956519d4854fd6f43e5766f681bbe77d3915f4fa26
+size 5618837

ai_model/words/saved_models/best_sign_classifier_model_nslt300.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f655d7b2faf2c72ad2c8fa9f8c7b8d82996a08ac4cc07678a71abe1312cff8eb
+size 13009869

ai_model/words/saved_models/normalization_params.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50224cbc68fa85e27ac5d27f5d28f17cfc868587e576777ece244d849751e76e
+size 13796

ai_model/words/train_classifier.py

@@ -0,0 +1,127 @@
+import os
+import pandas as pd
+import numpy as np
+from sklearn.model_selection import train_test_split 
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Bidirectional, LSTM, Dense, Dropout, BatchNormalization 
+from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau
+from tensorflow.keras.optimizers import Adam
+from tensorflow.keras.utils import to_categorical
+from tensorflow.keras.regularizers import l2 
+from tqdm import tqdm 
+
+# --- Configuration (match with previous steps and new model params) ---
+# IMPORTANT: This now points to the CSV generated by data_preprocessor.py for your 125 words
+PROCESSED_DATA_CSV = 'wlasl_125_words_personal_final_processed_data_augmented_seq90.csv'
+PROCESSED_SEQUENCES_DIR = 'processed_sequences' 
+MODEL_SAVE_DIR = 'saved_models' 
+
+SEQUENCE_LENGTH = 90
+EXPECTED_COORDS_PER_FRAME = 1662 
+
+# --- Model Hyperparameters (ADJUSTED FOR 125 WORDS) ---
+# These parameters are increased significantly to handle the larger vocabulary
+LSTM_UNITS = 256 # Increased capacity from 128
+DENSE_UNITS = 512 # Increased capacity from 192
+DROPOUT_RATE = 0.6 # Increased for more aggressive regularization
+LEARNING_RATE = 0.0005 
+BATCH_SIZE = 32 
+EPOCHS = 150 # Increased max epochs to give the model more time to learn
+
+def load_data(df):
+    """Loads processed landmark sequences and their corresponding labels."""
+    X = [] 
+    y = [] 
+    for index, row in tqdm(df.iterrows(), total=len(df), desc="Loading processed data"):
+        seq_path = row['processed_path']
+        gloss_id = row['gloss_id']
+        try:
+            sequence = np.load(seq_path)
+            if sequence.shape == (SEQUENCE_LENGTH, EXPECTED_COORDS_PER_FRAME):
+                X.append(sequence)
+                y.append(gloss_id)
+            else:
+                print(f"Warning: Skipping {seq_path} due to incorrect shape: {sequence.shape}. Expected ({SEQUENCE_LENGTH}, {EXPECTED_COORDS_PER_FRAME})")
+        except Exception as e:
+            print(f"Error loading {seq_path}: {e}. Skipping.")
+    return np.array(X), np.array(y)
+
+def build_lstm_model(input_shape, num_classes, lstm_units, dense_units, dropout_rate, learning_rate):
+    model = Sequential([
+        Bidirectional(LSTM(lstm_units, return_sequences=True, activation='tanh'), input_shape=input_shape),
+        BatchNormalization(), 
+        Dropout(dropout_rate),
+        Bidirectional(LSTM(lstm_units, return_sequences=False, activation='tanh')),
+        BatchNormalization(), 
+        Dropout(dropout_rate),
+        Dense(dense_units, activation='relu', kernel_regularizer=l2(0.001)),
+        BatchNormalization(), 
+        Dropout(dropout_rate),
+        Dense(num_classes, activation='softmax', kernel_regularizer=l2(0.001))
+    ])
+    optimizer = Adam(learning_rate=learning_rate)
+    model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])
+    return model
+
+if __name__ == "__main__":
+    os.makedirs(MODEL_SAVE_DIR, exist_ok=True)
+    if not os.path.exists(PROCESSED_DATA_CSV):
+        print(f"Error: Processed data metadata CSV not found at {PROCESSED_DATA_CSV}.")
+        print("Please ensure data_preprocessor.py has been run to create this file.") 
+        exit()
+    df_final = pd.read_csv(PROCESSED_DATA_CSV)
+    print(f"Loaded {len(df_final)} entries from {PROCESSED_DATA_CSV}")
+    train_df = df_final[df_final['split'] == 'train']
+    val_df = df_final[df_final['split'] == 'val']
+    test_df = df_final[df_final['split'] == 'test']
+    print(f"Train samples (including augmented): {len(train_df)}")
+    print(f"Validation samples: {len(val_df)}")
+    print(f"Test samples: {len(test_df)}") 
+    X_train, y_train_ids = load_data(train_df)
+    X_val, y_val_ids = load_data(val_df)
+    X_test, y_test_ids = load_data(test_df)
+    print(f"\nLoaded data shapes:")
+    print(f"X_train: {X_train.shape}, y_train: {y_train_ids.shape}")
+    print(f"X_val: {X_val.shape}, y_val: {y_val_ids.shape}")
+    print(f"X_test: {X_test.shape}, y_test: {y_test_ids.shape}")
+    num_classes = df_final['gloss_id'].nunique()
+    print(f"Number of unique classes: {num_classes}") # Should be 125
+    y_train = to_categorical(y_train_ids, num_classes=num_classes)
+    y_val = to_categorical(y_val_ids, num_classes=num_classes)
+    y_test = to_categorical(y_test_ids, num_classes=num_classes) 
+    print(f"y_train (one-hot) shape: {y_train.shape}")
+    print(f"y_val (one-hot) shape: {y_val.shape}")
+    model = build_lstm_model(
+        input_shape=(SEQUENCE_LENGTH, EXPECTED_COORDS_PER_FRAME),
+        num_classes=num_classes,
+        lstm_units=LSTM_UNITS,
+        dense_units=DENSE_UNITS,
+        dropout_rate=DROPOUT_RATE,
+        learning_rate=LEARNING_RATE
+    )
+    model.summary()
+    early_stopping = EarlyStopping(monitor='val_accuracy', patience=40, restore_best_weights=True)
+    checkpoint_filepath = os.path.join(MODEL_SAVE_DIR, 'best_sign_classifier_model_125_words_seq90.keras')
+    model_checkpoint = ModelCheckpoint(checkpoint_filepath, monitor='val_accuracy', save_best_only=True, verbose=1)
+    reduce_lr = ReduceLROnPlateau(monitor='val_accuracy', factor=0.5, patience=20, min_lr=0.00001, verbose=1)
+    print(f"\nStarting model training with BATCH_SIZE={BATCH_SIZE} and EPOCHS={EPOCHS}...")
+    history = model.fit(
+        X_train, y_train,
+        epochs=EPOCHS,
+        batch_size=BATCH_SIZE,
+        validation_data=(X_val, y_val),
+        callbacks=[early_stopping, model_checkpoint, reduce_lr],
+        verbose=1
+    )
+    print("\n--- Training Finished ---")
+    if os.path.exists(checkpoint_filepath):
+        print(f"Loading best model from: {checkpoint_filepath}")
+        model.load_weights(checkpoint_filepath)
+    else:
+        print(f"Warning: Best model checkpoint not found at {checkpoint_filepath}. Using final trained weights (may not be the best).")
+    print("\nEvaluating model on the test set...")
+    loss, accuracy = model.evaluate(X_test, y_test, batch_size=BATCH_SIZE, verbose=1)
+    print(f"Test Loss: {loss:.4f}")
+    print(f"Test Accuracy: {accuracy:.4f}")
+    print("\nModel training and evaluation complete.")
+    print(f"Best model weights saved to: {checkpoint_filepath}")

ai_model/words/visualize_npy.py

@@ -0,0 +1,126 @@
+import cv2
+import numpy as np
+import mediapipe as mp
+import os
+import time
+
+# --- Configuration (should match your recording script) ---
+OUTPUT_RECORDINGS_DIR = 'my_recorded_signs'
+EXPECTED_COORDS_PER_FRAME = 1662
+NUM_POSE_COORDS_SINGLE = 33 * 4
+NUM_HAND_COORDS_SINGLE = 21 * 3
+NUM_FACE_COORDS_SINGLE = 468 * 3
+FRAME_DELAY_MS = 40  # Approximately 25 FPS (1000ms / 25)
+
+# --- Helper function to reconstruct landmarks ---
+def reconstruct_landmarks(flat_landmarks):
+    """
+    Reconstructs MediaPipe landmarks from a flat numpy array.
+    """
+    if np.all(flat_landmarks == 0):
+        return None, None, None, None
+
+    # Pose landmarks
+    pose_lms = mp.solutions.pose.PoseLandmark.list()
+    pose_lms_flat = flat_landmarks[0:NUM_POSE_COORDS_SINGLE]
+    if np.any(pose_lms_flat != 0):
+        pose_landmarks = mp.solutions.holistic.PoseLandmarks()
+        for i, lm in enumerate(pose_lms):
+            pose_landmarks.landmark[i].x = pose_lms_flat[i*4]
+            pose_landmarks.landmark[i].y = pose_lms_flat[i*4+1]
+            pose_landmarks.landmark[i].z = pose_lms_flat[i*4+2]
+            pose_landmarks.landmark[i].visibility = pose_lms_flat[i*4+3]
+    else:
+        pose_landmarks = None
+
+    # Left hand landmarks
+    lh_lms_flat = flat_landmarks[NUM_POSE_COORDS_SINGLE:NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE]
+    if np.any(lh_lms_flat != 0):
+        left_hand_landmarks = mp.solutions.holistic.HandLandmarks()
+        for i in range(21):
+            left_hand_landmarks.landmark[i].x = lh_lms_flat[i*3]
+            left_hand_landmarks.landmark[i].y = lh_lms_flat[i*3+1]
+            left_hand_landmarks.landmark[i].z = lh_lms_flat[i*3+2]
+    else:
+        left_hand_landmarks = None
+
+    # Right hand landmarks
+    rh_lms_flat = flat_landmarks[NUM_POSE_COORDS_SINGLE + NUM_HAND_COORDS_SINGLE:NUM_POSE_COORDS_SINGLE + 2*NUM_HAND_COORDS_SINGLE]
+    if np.any(rh_lms_flat != 0):
+        right_hand_landmarks = mp.solutions.holistic.HandLandmarks()
+        for i in range(21):
+            right_hand_landmarks.landmark[i].x = rh_lms_flat[i*3]
+            right_hand_landmarks.landmark[i].y = rh_lms_flat[i*3+1]
+            right_hand_landmarks.landmark[i].z = rh_lms_flat[i*3+2]
+    else:
+        right_hand_landmarks = None
+
+    # Face landmarks
+    face_lms_flat = flat_landmarks[NUM_POSE_COORDS_SINGLE + 2*NUM_HAND_COORDS_SINGLE:]
+    if np.any(face_lms_flat != 0):
+        face_landmarks = mp.solutions.holistic.FaceLandmarks()
+        for i in range(468):
+            face_landmarks.landmark[i].x = face_lms_flat[i*3]
+            face_landmarks.landmark[i].y = face_lms_flat[i*3+1]
+            face_landmarks.landmark[i].z = face_lms_flat[i*3+2]
+    else:
+        face_landmarks = None
+
+    return pose_landmarks, left_hand_landmarks, right_hand_landmarks, face_landmarks
+
+# --- Main Visualization Logic ---
+if __name__ == "__main__":
+    if not os.path.exists(OUTPUT_RECORDINGS_DIR):
+        print(f"Error: Directory '{OUTPUT_RECORDINGS_DIR}' not found.")
+        exit()
+
+    mp_drawing = mp.solutions.drawing_utils
+    files = [f for f in os.listdir(OUTPUT_RECORDINGS_DIR) if f.endswith('.npy')]
+
+    if not files:
+        print("No .npy files found in the directory.")
+        exit()
+
+    for file_name in files:
+        file_path = os.path.join(OUTPUT_RECORDINGS_DIR, file_name)
+        gloss = file_name.split('_')[0].upper()
+        
+        try:
+            landmarks_sequence = np.load(file_path)
+            print(f"\n--- Loading and visualizing: {file_name} (Gloss: {gloss}) ---")
+            print(f"Shape: {landmarks_sequence.shape}")
+
+            for i, frame_landmarks in enumerate(landmarks_sequence):
+                blank_frame = np.zeros((720, 1280, 3), dtype=np.uint8)
+                pose_lms, lh_lms, rh_lms, face_lms = reconstruct_landmarks(frame_landmarks)
+
+                if pose_lms:
+                    mp_drawing.draw_landmarks(blank_frame, pose_lms, mp.solutions.holistic.POSE_CONNECTIONS)
+                if lh_lms:
+                    mp_drawing.draw_landmarks(blank_frame, lh_lms, mp.solutions.holistic.HAND_CONNECTIONS)
+                if rh_lms:
+                    mp_drawing.draw_landmarks(blank_frame, rh_lms, mp.solutions.holistic.HAND_CONNECTIONS)
+                if face_lms:
+                    mp_drawing.draw_landmarks(blank_frame, face_lms, mp.solutions.holistic.FACEMESH_CONTOURS)
+
+                cv2.putText(blank_frame, f"File: {file_name}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 255, 255), 2, cv2.LINE_AA)
+                cv2.putText(blank_frame, f"Gloss: {gloss}", (10, 70), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2, cv2.LINE_AA)
+                cv2.putText(blank_frame, f"Frame: {i+1}/{len(landmarks_sequence)}", (10, 110), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (255, 255, 255), 1, cv2.LINE_AA)
+
+                cv2.imshow('NPY File Visualizer', blank_frame)
+                
+                # Wait for a brief period to simulate playback speed
+                key = cv2.waitKey(FRAME_DELAY_MS)
+                if key == ord('q'):
+                    break
+            
+            # Wait for user input before moving to the next file
+            print("Press any key to continue to the next file...")
+            cv2.waitKey(0)
+
+        except Exception as e:
+            print(f"Error loading or visualizing {file_name}: {e}")
+            continue
+    
+    cv2.destroyAllWindows()
+    print("\nVisualization complete.")

ai_model/words/wlasl_10_words_combined_processed.csv

@@ -0,0 +1,940 @@
+gloss,gloss_id,video_id,video_path,signer_id,split,frame_start,frame_end,fps,bbox
+chair,0,9848,.\videos\09848.mp4,0,train,1,-1,25,"[40, 6, 266, 240]"
+chair,0,9869,.\videos\09869.mp4,12,train,1,-1,25,"[183, 55, 560, 400]"
+chair,0,9849,.\videos\09849.mp4,38,train,1,-1,25,"[388, 56, 806, 720]"
+chair,0,9850,.\videos\09850.mp4,31,train,1,-1,25,"[667, 144, 1472, 1061]"
+chair,0,9851,.\videos\09851.mp4,59,val,1,-1,25,"[102, 22, 563, 480]"
+chair,0,65328,.\videos\65328.mp4,90,train,1,-1,25,"[169, 15, 476, 370]"
+chair,0,9854,.\videos\09854.mp4,11,val,1,-1,25,"[81, 15, 233, 192]"
+black,1,69236,.\videos\69236.mp4,118,val,1,-1,25,"[295, 38, 885, 720]"
+black,1,6472,.\videos\06472.mp4,36,train,1,-1,25,"[205, 17, 524, 414]"
+black,1,6473,.\videos\06473.mp4,4,train,1,-1,25,"[544, 60, 1618, 1080]"
+black,1,6474,.\videos\06474.mp4,59,train,1,-1,25,"[77, 18, 515, 480]"
+black,1,6476,.\videos\06476.mp4,11,val,1,-1,25,"[58, 15, 216, 192]"
+black,1,6477,.\videos\06477.mp4,11,train,1,-1,25,"[48, 11, 226, 192]"
+black,1,6478,.\videos\06478.mp4,11,train,1,-1,25,"[63, 8, 245, 192]"
+black,1,65200,.\videos\65200.mp4,90,train,1,-1,25,"[149, 24, 479, 370]"
+black,1,6471,.\videos\06471.mp4,56,train,1,-1,25,"[66, 13, 258, 240]"
+black,1,6486,.\videos\06486.mp4,12,train,1,-1,25,"[131, 52, 555, 400]"
+blue,2,69238,.\videos\69238.mp4,118,val,1,-1,25,"[257, 39, 881, 720]"
+blue,2,6835,.\videos\06835.mp4,59,train,1,-1,25,"[125, 26, 528, 480]"
+blue,2,6839,.\videos\06839.mp4,10,train,1,-1,25,"[90, 21, 213, 192]"
+blue,2,65216,.\videos\65216.mp4,90,train,1,-1,25,"[158, 24, 482, 370]"
+blue,2,6832,.\videos\06832.mp4,8,train,1,-1,25,"[41, 0, 232, 240]"
+blue,2,6845,.\videos\06845.mp4,41,test,1,-1,25,"[232, 39, 483, 400]"
+blue,2,6833,.\videos\06833.mp4,36,train,1,-1,25,"[225, 18, 519, 414]"
+blue,2,6834,.\videos\06834.mp4,46,train,1,-1,25,"[188, 0, 794, 720]"
+can,3,69257,.\videos\69257.mp4,118,test,1,-1,25,"[285, 31, 964, 720]"
+can,3,65294,.\videos\65294.mp4,90,val,1,-1,25,"[137, 15, 500, 370]"
+can,3,8955,.\videos\08955.mp4,12,train,1,-1,25,"[163, 54, 555, 400]"
+can,3,8935,.\videos\08935.mp4,38,train,1,-1,25,"[434, 54, 814, 720]"
+can,3,8936,.\videos\08936.mp4,70,train,1,-1,25,"[201, 5, 823, 480]"
+can,3,8937,.\videos\08937.mp4,59,train,1,-1,25,"[69, 17, 552, 480]"
+can,3,8938,.\videos\08938.mp4,59,train,1,-1,25,"[110, 29, 555, 480]"
+can,3,8942,.\videos\08942.mp4,11,train,1,-1,25,"[76, 18, 237, 192]"
+can,3,8944,.\videos\08944.mp4,11,train,1,-1,25,"[70, 18, 217, 192]"
+apple,4,69213,.\videos\69213.mp4,117,val,1,-1,25,"[304, 41, 916, 720]"
+apple,4,3005,.\videos\03005.mp4,11,train,1,-1,25,"[64, 15, 229, 192]"
+apple,4,3008,.\videos\03008.mp4,12,train,1,-1,25,"[158, 50, 541, 399]"
+apple,4,65085,.\videos\65085.mp4,89,train,1,-1,25,"[207, 0, 520, 370]"
+apple,4,65086,.\videos\65086.mp4,89,val,1,-1,25,"[206, 0, 535, 370]"
+apple,4,65084,.\videos\65084.mp4,88,train,1,-1,25,"[119, 4, 486, 370]"
+apple,4,2999,.\videos\02999.mp4,37,val,1,-1,25,"[44, 0, 232, 240]"
+apple,4,3000,.\videos\03000.mp4,28,train,1,-1,25,"[41, 13, 259, 240]"
+apple,4,3001,.\videos\03001.mp4,38,train,1,-1,25,"[350, 53, 815, 720]"
+apple,4,3002,.\videos\03002.mp4,36,train,1,-1,25,"[212, 17, 526, 414]"
+apple,4,3003,.\videos\03003.mp4,13,train,1,-1,25,"[24, 13, 526, 480]"
+brown,5,69252,.\videos\69252.mp4,118,test,1,-1,25,"[335, 37, 874, 720]"
+brown,5,7963,.\videos\07963.mp4,59,train,1,-1,25,"[102, 30, 523, 480]"
+brown,5,7966,.\videos\07966.mp4,11,train,1,-1,25,"[62, 15, 217, 192]"
+brown,5,65263,.\videos\65263.mp4,90,train,1,-1,25,"[158, 12, 462, 370]"
+brown,5,7960,.\videos\07960.mp4,2,train,1,-1,25,"[37, 6, 250, 240]"
+brown,5,7973,.\videos\07973.mp4,12,val,1,-1,25,"[182, 52, 542, 400]"
+brown,5,7961,.\videos\07961.mp4,36,train,1,-1,25,"[213, 11, 525, 414]"
+brown,5,7962,.\videos\07962.mp4,4,train,1,-1,25,"[381, 38, 1076, 720]"
+cat,6,69261,.\videos\69261.mp4,118,test,1,-1,25,"[357, 37, 874, 720]"
+cat,6,9528,.\videos\09528.mp4,32,train,1,-1,25,"[743, 79, 1579, 1066]"
+cat,6,9532,.\videos\09532.mp4,10,val,1,-1,25,"[84, 21, 216, 192]"
+cat,6,9533,.\videos\09533.mp4,10,train,1,-1,25,"[83, 20, 218, 192]"
+cat,6,65312,.\videos\65312.mp4,102,train,1,-1,25,"[156, 19, 486, 370]"
+cat,6,65313,.\videos\65313.mp4,101,train,1,-1,25,"[168, 18, 462, 370]"
+cat,6,9525,.\videos\09525.mp4,2,val,1,-1,25,"[57, 0, 271, 240]"
+cat,6,9538,.\videos\09538.mp4,41,train,1,-1,25,"[215, 42, 465, 400]"
+cat,6,9526,.\videos\09526.mp4,36,train,1,-1,25,"[210, 18, 523, 414]"
+cold,7,65377,.\videos\65377.mp4,99,test,1,-1,25,"[129, 16, 476, 370]"
+cold,7,11638,.\videos\11638.mp4,12,test,1,-1,25,"[180, 57, 551, 400]"
+cold,7,11622,.\videos\11622.mp4,36,train,1,-1,25,"[210, 16, 540, 414]"
+cold,7,11623,.\videos\11623.mp4,52,train,1,-1,25,"[380, 16, 825, 480]"
+cold,7,11624,.\videos\11624.mp4,6,train,1,-1,25,"[593, 70, 1468, 1080]"
+cold,7,11625,.\videos\11625.mp4,39,val,1,-1,25,"[624, 133, 1621, 1072]"
+cold,7,11627,.\videos\11627.mp4,4,train,1,-1,25,"[411, 31, 1128, 720]"
+cold,7,11628,.\videos\11628.mp4,59,val,1,-1,25,"[116, 29, 513, 480]"
+cold,7,11621,.\videos\11621.mp4,56,train,1,-1,25,"[59, 16, 263, 240]"
+cold,7,11633,.\videos\11633.mp4,10,train,1,-1,25,"[88, 22, 216, 192]"
+cold,7,11634,.\videos\11634.mp4,10,val,1,-1,25,"[65, 21, 233, 192]"
+cold,7,11635,.\videos\11635.mp4,11,train,1,-1,25,"[79, 16, 221, 192]"
+child,8,10450,.\videos\10450.mp4,11,train,1,-1,25,"[69, 17, 215, 192]"
+child,8,65351,.\videos\65351.mp4,90,train,1,-1,25,"[130, 14, 465, 370]"
+child,8,10443,.\videos\10443.mp4,2,train,1,-1,25,"[64, 0, 269, 240]"
+child,8,10453,.\videos\10453.mp4,12,train,1,-1,25,"[130, 54, 550, 400]"
+child,8,10444,.\videos\10444.mp4,32,train,1,-1,25,"[700, 66, 1636, 1073]"
+child,8,10445,.\videos\10445.mp4,4,train,1,-1,25,"[616, 61, 1702, 1079]"
+child,8,10446,.\videos\10446.mp4,32,test,1,-1,25,"[693, 70, 1631, 1073]"
+child,8,10447,.\videos\10447.mp4,59,train,1,-1,25,"[50, 24, 520, 480]"
+child,8,10449,.\videos\10449.mp4,10,val,1,-1,25,"[50, 17, 219, 192]"
+come,9,69275,.\videos\69275.mp4,115,test,1,-1,25,"[140, 37, 943, 720]"
+come,9,11884,.\videos\11884.mp4,12,train,1,-1,25,"[178, 59, 534, 400]"
+come,9,11876,.\videos\11876.mp4,56,train,1,-1,25,"[63, 13, 261, 240]"
+come,9,11877,.\videos\11877.mp4,31,train,1,-1,25,"[540, 64, 1480, 1080]"
+come,9,11878,.\videos\11878.mp4,59,val,1,-1,25,"[84, 27, 590, 480]"
+come,9,11882,.\videos\11882.mp4,11,train,1,-1,25,"[82, 14, 230, 192]"
+apple,4,personal_1753086535,my_recorded_signs\apple_1753086535.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086544,my_recorded_signs\apple_1753086544.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086553,my_recorded_signs\apple_1753086553.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086562,my_recorded_signs\apple_1753086562.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086570,my_recorded_signs\apple_1753086570.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086578,my_recorded_signs\apple_1753086578.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086587,my_recorded_signs\apple_1753086587.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086602,my_recorded_signs\apple_1753086602.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086624,my_recorded_signs\apple_1753086624.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086633,my_recorded_signs\apple_1753086633.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086643,my_recorded_signs\apple_1753086643.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086653,my_recorded_signs\apple_1753086653.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086662,my_recorded_signs\apple_1753086662.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086671,my_recorded_signs\apple_1753086671.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086683,my_recorded_signs\apple_1753086683.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086692,my_recorded_signs\apple_1753086692.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086702,my_recorded_signs\apple_1753086702.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086715,my_recorded_signs\apple_1753086715.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086726,my_recorded_signs\apple_1753086726.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086735,my_recorded_signs\apple_1753086735.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086744,my_recorded_signs\apple_1753086744.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086754,my_recorded_signs\apple_1753086754.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086763,my_recorded_signs\apple_1753086763.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086772,my_recorded_signs\apple_1753086772.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753086782,my_recorded_signs\apple_1753086782.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171696,my_recorded_signs\apple_1753171696.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171707,my_recorded_signs\apple_1753171707.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171715,my_recorded_signs\apple_1753171715.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171728,my_recorded_signs\apple_1753171728.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171738,my_recorded_signs\apple_1753171738.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171747,my_recorded_signs\apple_1753171747.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171756,my_recorded_signs\apple_1753171756.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171766,my_recorded_signs\apple_1753171766.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171778,my_recorded_signs\apple_1753171778.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171787,my_recorded_signs\apple_1753171787.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171797,my_recorded_signs\apple_1753171797.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171807,my_recorded_signs\apple_1753171807.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171817,my_recorded_signs\apple_1753171817.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171827,my_recorded_signs\apple_1753171827.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171838,my_recorded_signs\apple_1753171838.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171847,my_recorded_signs\apple_1753171847.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753171856,my_recorded_signs\apple_1753171856.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183653,my_recorded_signs\apple_1753183653.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183661,my_recorded_signs\apple_1753183661.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183670,my_recorded_signs\apple_1753183670.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183679,my_recorded_signs\apple_1753183679.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183688,my_recorded_signs\apple_1753183688.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183697,my_recorded_signs\apple_1753183697.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183708,my_recorded_signs\apple_1753183708.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183717,my_recorded_signs\apple_1753183717.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753183726,my_recorded_signs\apple_1753183726.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206542,my_recorded_signs\apple_1753206542.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206551,my_recorded_signs\apple_1753206551.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206559,my_recorded_signs\apple_1753206559.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206568,my_recorded_signs\apple_1753206568.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206576,my_recorded_signs\apple_1753206576.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206585,my_recorded_signs\apple_1753206585.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206615,my_recorded_signs\apple_1753206615.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206624,my_recorded_signs\apple_1753206624.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206633,my_recorded_signs\apple_1753206633.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206641,my_recorded_signs\apple_1753206641.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753206650,my_recorded_signs\apple_1753206650.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753274889,my_recorded_signs\apple_1753274889.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753274898,my_recorded_signs\apple_1753274898.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753274910,my_recorded_signs\apple_1753274910.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753274918,my_recorded_signs\apple_1753274918.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753274927,my_recorded_signs\apple_1753274927.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367091,my_recorded_signs\apple_1753367091.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367099,my_recorded_signs\apple_1753367099.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367108,my_recorded_signs\apple_1753367108.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367117,my_recorded_signs\apple_1753367117.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367126,my_recorded_signs\apple_1753367126.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367134,my_recorded_signs\apple_1753367134.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367143,my_recorded_signs\apple_1753367143.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367154,my_recorded_signs\apple_1753367154.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367163,my_recorded_signs\apple_1753367163.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+apple,4,personal_1753367174,my_recorded_signs\apple_1753367174.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085581,my_recorded_signs\black_1753085581.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085593,my_recorded_signs\black_1753085593.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085605,my_recorded_signs\black_1753085605.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085616,my_recorded_signs\black_1753085616.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085627,my_recorded_signs\black_1753085627.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085637,my_recorded_signs\black_1753085637.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085648,my_recorded_signs\black_1753085648.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085660,my_recorded_signs\black_1753085660.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085673,my_recorded_signs\black_1753085673.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085684,my_recorded_signs\black_1753085684.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085695,my_recorded_signs\black_1753085695.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085705,my_recorded_signs\black_1753085705.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085715,my_recorded_signs\black_1753085715.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085725,my_recorded_signs\black_1753085725.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085735,my_recorded_signs\black_1753085735.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085744,my_recorded_signs\black_1753085744.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085756,my_recorded_signs\black_1753085756.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085770,my_recorded_signs\black_1753085770.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085787,my_recorded_signs\black_1753085787.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753085797,my_recorded_signs\black_1753085797.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175295,my_recorded_signs\black_1753175295.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175308,my_recorded_signs\black_1753175308.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175317,my_recorded_signs\black_1753175317.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175331,my_recorded_signs\black_1753175331.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175339,my_recorded_signs\black_1753175339.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175348,my_recorded_signs\black_1753175348.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753175356,my_recorded_signs\black_1753175356.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183331,my_recorded_signs\black_1753183331.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183377,my_recorded_signs\black_1753183377.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183390,my_recorded_signs\black_1753183390.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183409,my_recorded_signs\black_1753183409.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183419,my_recorded_signs\black_1753183419.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183470,my_recorded_signs\black_1753183470.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183479,my_recorded_signs\black_1753183479.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183487,my_recorded_signs\black_1753183487.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183496,my_recorded_signs\black_1753183496.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753183511,my_recorded_signs\black_1753183511.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366647,my_recorded_signs\black_1753366647.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366656,my_recorded_signs\black_1753366656.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366664,my_recorded_signs\black_1753366664.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366672,my_recorded_signs\black_1753366672.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366681,my_recorded_signs\black_1753366681.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366699,my_recorded_signs\black_1753366699.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366707,my_recorded_signs\black_1753366707.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366716,my_recorded_signs\black_1753366716.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+black,1,personal_1753366725,my_recorded_signs\black_1753366725.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085844,my_recorded_signs\blue_1753085844.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085855,my_recorded_signs\blue_1753085855.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085863,my_recorded_signs\blue_1753085863.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085872,my_recorded_signs\blue_1753085872.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085882,my_recorded_signs\blue_1753085882.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085893,my_recorded_signs\blue_1753085893.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085912,my_recorded_signs\blue_1753085912.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085922,my_recorded_signs\blue_1753085922.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085957,my_recorded_signs\blue_1753085957.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085966,my_recorded_signs\blue_1753085966.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085976,my_recorded_signs\blue_1753085976.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085984,my_recorded_signs\blue_1753085984.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753085995,my_recorded_signs\blue_1753085995.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086005,my_recorded_signs\blue_1753086005.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086020,my_recorded_signs\blue_1753086020.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086030,my_recorded_signs\blue_1753086030.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086040,my_recorded_signs\blue_1753086040.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086050,my_recorded_signs\blue_1753086050.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086060,my_recorded_signs\blue_1753086060.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086069,my_recorded_signs\blue_1753086069.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753086080,my_recorded_signs\blue_1753086080.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175373,my_recorded_signs\blue_1753175373.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175384,my_recorded_signs\blue_1753175384.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175392,my_recorded_signs\blue_1753175392.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175400,my_recorded_signs\blue_1753175400.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175408,my_recorded_signs\blue_1753175408.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175417,my_recorded_signs\blue_1753175417.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753175426,my_recorded_signs\blue_1753175426.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206662,my_recorded_signs\blue_1753206662.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206671,my_recorded_signs\blue_1753206671.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206699,my_recorded_signs\blue_1753206699.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206707,my_recorded_signs\blue_1753206707.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206718,my_recorded_signs\blue_1753206718.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206726,my_recorded_signs\blue_1753206726.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206735,my_recorded_signs\blue_1753206735.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206744,my_recorded_signs\blue_1753206744.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206753,my_recorded_signs\blue_1753206753.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753206761,my_recorded_signs\blue_1753206761.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262291,my_recorded_signs\blue_1753262291.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262303,my_recorded_signs\blue_1753262303.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262311,my_recorded_signs\blue_1753262311.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262319,my_recorded_signs\blue_1753262319.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262327,my_recorded_signs\blue_1753262327.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262335,my_recorded_signs\blue_1753262335.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262343,my_recorded_signs\blue_1753262343.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262351,my_recorded_signs\blue_1753262351.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262359,my_recorded_signs\blue_1753262359.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262366,my_recorded_signs\blue_1753262366.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262375,my_recorded_signs\blue_1753262375.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262383,my_recorded_signs\blue_1753262383.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262391,my_recorded_signs\blue_1753262391.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262399,my_recorded_signs\blue_1753262399.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262408,my_recorded_signs\blue_1753262408.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262416,my_recorded_signs\blue_1753262416.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262424,my_recorded_signs\blue_1753262424.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262432,my_recorded_signs\blue_1753262432.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262441,my_recorded_signs\blue_1753262441.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262453,my_recorded_signs\blue_1753262453.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262462,my_recorded_signs\blue_1753262462.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262471,my_recorded_signs\blue_1753262471.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262480,my_recorded_signs\blue_1753262480.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262491,my_recorded_signs\blue_1753262491.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262500,my_recorded_signs\blue_1753262500.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262509,my_recorded_signs\blue_1753262509.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262517,my_recorded_signs\blue_1753262517.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262524,my_recorded_signs\blue_1753262524.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262533,my_recorded_signs\blue_1753262533.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262542,my_recorded_signs\blue_1753262542.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262551,my_recorded_signs\blue_1753262551.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262560,my_recorded_signs\blue_1753262560.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262568,my_recorded_signs\blue_1753262568.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262586,my_recorded_signs\blue_1753262586.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262596,my_recorded_signs\blue_1753262596.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262605,my_recorded_signs\blue_1753262605.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262615,my_recorded_signs\blue_1753262615.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262625,my_recorded_signs\blue_1753262625.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262635,my_recorded_signs\blue_1753262635.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753262645,my_recorded_signs\blue_1753262645.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366739,my_recorded_signs\blue_1753366739.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366748,my_recorded_signs\blue_1753366748.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366757,my_recorded_signs\blue_1753366757.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366766,my_recorded_signs\blue_1753366766.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366774,my_recorded_signs\blue_1753366774.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366782,my_recorded_signs\blue_1753366782.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366791,my_recorded_signs\blue_1753366791.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366800,my_recorded_signs\blue_1753366800.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+blue,2,personal_1753366808,my_recorded_signs\blue_1753366808.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086830,my_recorded_signs\brown_1753086830.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086839,my_recorded_signs\brown_1753086839.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086851,my_recorded_signs\brown_1753086851.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086859,my_recorded_signs\brown_1753086859.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086870,my_recorded_signs\brown_1753086870.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086879,my_recorded_signs\brown_1753086879.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086889,my_recorded_signs\brown_1753086889.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086898,my_recorded_signs\brown_1753086898.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086908,my_recorded_signs\brown_1753086908.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086917,my_recorded_signs\brown_1753086917.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086927,my_recorded_signs\brown_1753086927.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086942,my_recorded_signs\brown_1753086942.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086951,my_recorded_signs\brown_1753086951.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086959,my_recorded_signs\brown_1753086959.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086968,my_recorded_signs\brown_1753086968.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086976,my_recorded_signs\brown_1753086976.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086985,my_recorded_signs\brown_1753086985.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753086994,my_recorded_signs\brown_1753086994.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753087007,my_recorded_signs\brown_1753087007.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753087015,my_recorded_signs\brown_1753087015.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753087025,my_recorded_signs\brown_1753087025.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753087034,my_recorded_signs\brown_1753087034.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753087044,my_recorded_signs\brown_1753087044.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172046,my_recorded_signs\brown_1753172046.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172058,my_recorded_signs\brown_1753172058.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172067,my_recorded_signs\brown_1753172067.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172076,my_recorded_signs\brown_1753172076.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172086,my_recorded_signs\brown_1753172086.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172095,my_recorded_signs\brown_1753172095.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172103,my_recorded_signs\brown_1753172103.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172112,my_recorded_signs\brown_1753172112.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172123,my_recorded_signs\brown_1753172123.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172132,my_recorded_signs\brown_1753172132.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172140,my_recorded_signs\brown_1753172140.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172149,my_recorded_signs\brown_1753172149.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172158,my_recorded_signs\brown_1753172158.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172167,my_recorded_signs\brown_1753172167.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172178,my_recorded_signs\brown_1753172178.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172187,my_recorded_signs\brown_1753172187.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172195,my_recorded_signs\brown_1753172195.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172205,my_recorded_signs\brown_1753172205.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172214,my_recorded_signs\brown_1753172214.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172222,my_recorded_signs\brown_1753172222.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172231,my_recorded_signs\brown_1753172231.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172241,my_recorded_signs\brown_1753172241.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172250,my_recorded_signs\brown_1753172250.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753172265,my_recorded_signs\brown_1753172265.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183524,my_recorded_signs\brown_1753183524.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183533,my_recorded_signs\brown_1753183533.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183541,my_recorded_signs\brown_1753183541.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183550,my_recorded_signs\brown_1753183550.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183565,my_recorded_signs\brown_1753183565.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183576,my_recorded_signs\brown_1753183576.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183585,my_recorded_signs\brown_1753183585.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183595,my_recorded_signs\brown_1753183595.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183604,my_recorded_signs\brown_1753183604.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183615,my_recorded_signs\brown_1753183615.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183625,my_recorded_signs\brown_1753183625.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753183635,my_recorded_signs\brown_1753183635.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206451,my_recorded_signs\brown_1753206451.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206460,my_recorded_signs\brown_1753206460.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206468,my_recorded_signs\brown_1753206468.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206477,my_recorded_signs\brown_1753206477.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206486,my_recorded_signs\brown_1753206486.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206494,my_recorded_signs\brown_1753206494.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206504,my_recorded_signs\brown_1753206504.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206514,my_recorded_signs\brown_1753206514.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206523,my_recorded_signs\brown_1753206523.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753206531,my_recorded_signs\brown_1753206531.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262658,my_recorded_signs\brown_1753262658.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262667,my_recorded_signs\brown_1753262667.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262676,my_recorded_signs\brown_1753262676.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262686,my_recorded_signs\brown_1753262686.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262695,my_recorded_signs\brown_1753262695.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262704,my_recorded_signs\brown_1753262704.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262713,my_recorded_signs\brown_1753262713.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262723,my_recorded_signs\brown_1753262723.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262732,my_recorded_signs\brown_1753262732.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262745,my_recorded_signs\brown_1753262745.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262753,my_recorded_signs\brown_1753262753.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262761,my_recorded_signs\brown_1753262761.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262769,my_recorded_signs\brown_1753262769.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262778,my_recorded_signs\brown_1753262778.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262786,my_recorded_signs\brown_1753262786.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262794,my_recorded_signs\brown_1753262794.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262802,my_recorded_signs\brown_1753262802.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262810,my_recorded_signs\brown_1753262810.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262819,my_recorded_signs\brown_1753262819.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262827,my_recorded_signs\brown_1753262827.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262835,my_recorded_signs\brown_1753262835.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262843,my_recorded_signs\brown_1753262843.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262852,my_recorded_signs\brown_1753262852.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262860,my_recorded_signs\brown_1753262860.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262868,my_recorded_signs\brown_1753262868.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262876,my_recorded_signs\brown_1753262876.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262885,my_recorded_signs\brown_1753262885.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262894,my_recorded_signs\brown_1753262894.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262902,my_recorded_signs\brown_1753262902.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262911,my_recorded_signs\brown_1753262911.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262920,my_recorded_signs\brown_1753262920.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262928,my_recorded_signs\brown_1753262928.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262936,my_recorded_signs\brown_1753262936.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262944,my_recorded_signs\brown_1753262944.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262953,my_recorded_signs\brown_1753262953.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753262961,my_recorded_signs\brown_1753262961.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367243,my_recorded_signs\brown_1753367243.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367254,my_recorded_signs\brown_1753367254.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367283,my_recorded_signs\brown_1753367283.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367292,my_recorded_signs\brown_1753367292.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367300,my_recorded_signs\brown_1753367300.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367313,my_recorded_signs\brown_1753367313.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367344,my_recorded_signs\brown_1753367344.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367352,my_recorded_signs\brown_1753367352.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367361,my_recorded_signs\brown_1753367361.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367370,my_recorded_signs\brown_1753367370.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+brown,5,personal_1753367379,my_recorded_signs\brown_1753367379.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086108,my_recorded_signs\can_1753086108.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086118,my_recorded_signs\can_1753086118.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086127,my_recorded_signs\can_1753086127.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086136,my_recorded_signs\can_1753086136.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086146,my_recorded_signs\can_1753086146.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086155,my_recorded_signs\can_1753086155.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086165,my_recorded_signs\can_1753086165.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086174,my_recorded_signs\can_1753086174.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086186,my_recorded_signs\can_1753086186.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086197,my_recorded_signs\can_1753086197.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086207,my_recorded_signs\can_1753086207.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086218,my_recorded_signs\can_1753086218.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086228,my_recorded_signs\can_1753086228.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086241,my_recorded_signs\can_1753086241.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086249,my_recorded_signs\can_1753086249.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086261,my_recorded_signs\can_1753086261.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086339,my_recorded_signs\can_1753086339.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086352,my_recorded_signs\can_1753086352.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086361,my_recorded_signs\can_1753086361.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086369,my_recorded_signs\can_1753086369.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086378,my_recorded_signs\can_1753086378.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086387,my_recorded_signs\can_1753086387.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086396,my_recorded_signs\can_1753086396.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086405,my_recorded_signs\can_1753086405.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086415,my_recorded_signs\can_1753086415.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086423,my_recorded_signs\can_1753086423.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086432,my_recorded_signs\can_1753086432.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086441,my_recorded_signs\can_1753086441.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086452,my_recorded_signs\can_1753086452.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086462,my_recorded_signs\can_1753086462.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086473,my_recorded_signs\can_1753086473.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086501,my_recorded_signs\can_1753086501.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753086512,my_recorded_signs\can_1753086512.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753175438,my_recorded_signs\can_1753175438.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753175448,my_recorded_signs\can_1753175448.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753175456,my_recorded_signs\can_1753175456.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753175465,my_recorded_signs\can_1753175465.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753175477,my_recorded_signs\can_1753175477.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753175485,my_recorded_signs\can_1753175485.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206162,my_recorded_signs\can_1753206162.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206182,my_recorded_signs\can_1753206182.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206191,my_recorded_signs\can_1753206191.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206199,my_recorded_signs\can_1753206199.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206208,my_recorded_signs\can_1753206208.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206218,my_recorded_signs\can_1753206218.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753206226,my_recorded_signs\can_1753206226.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263384,my_recorded_signs\can_1753263384.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263394,my_recorded_signs\can_1753263394.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263403,my_recorded_signs\can_1753263403.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263411,my_recorded_signs\can_1753263411.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263420,my_recorded_signs\can_1753263420.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263431,my_recorded_signs\can_1753263431.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263441,my_recorded_signs\can_1753263441.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263449,my_recorded_signs\can_1753263449.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263458,my_recorded_signs\can_1753263458.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263466,my_recorded_signs\can_1753263466.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263475,my_recorded_signs\can_1753263475.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263483,my_recorded_signs\can_1753263483.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263492,my_recorded_signs\can_1753263492.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263500,my_recorded_signs\can_1753263500.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263508,my_recorded_signs\can_1753263508.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263517,my_recorded_signs\can_1753263517.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263525,my_recorded_signs\can_1753263525.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263533,my_recorded_signs\can_1753263533.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263542,my_recorded_signs\can_1753263542.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263550,my_recorded_signs\can_1753263550.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263561,my_recorded_signs\can_1753263561.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263570,my_recorded_signs\can_1753263570.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263579,my_recorded_signs\can_1753263579.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263589,my_recorded_signs\can_1753263589.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263598,my_recorded_signs\can_1753263598.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753263607,my_recorded_signs\can_1753263607.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753366987,my_recorded_signs\can_1753366987.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753366995,my_recorded_signs\can_1753366995.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367004,my_recorded_signs\can_1753367004.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367013,my_recorded_signs\can_1753367013.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367022,my_recorded_signs\can_1753367022.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367030,my_recorded_signs\can_1753367030.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367039,my_recorded_signs\can_1753367039.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367050,my_recorded_signs\can_1753367050.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+can,3,personal_1753367067,my_recorded_signs\can_1753367067.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087072,my_recorded_signs\cat_1753087072.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087081,my_recorded_signs\cat_1753087081.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087089,my_recorded_signs\cat_1753087089.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087097,my_recorded_signs\cat_1753087097.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087105,my_recorded_signs\cat_1753087105.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087115,my_recorded_signs\cat_1753087115.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087124,my_recorded_signs\cat_1753087124.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087134,my_recorded_signs\cat_1753087134.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087143,my_recorded_signs\cat_1753087143.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087155,my_recorded_signs\cat_1753087155.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087193,my_recorded_signs\cat_1753087193.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087202,my_recorded_signs\cat_1753087202.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087224,my_recorded_signs\cat_1753087224.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087233,my_recorded_signs\cat_1753087233.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087244,my_recorded_signs\cat_1753087244.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087254,my_recorded_signs\cat_1753087254.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087264,my_recorded_signs\cat_1753087264.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087273,my_recorded_signs\cat_1753087273.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087283,my_recorded_signs\cat_1753087283.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087292,my_recorded_signs\cat_1753087292.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753087302,my_recorded_signs\cat_1753087302.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172292,my_recorded_signs\cat_1753172292.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172306,my_recorded_signs\cat_1753172306.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172316,my_recorded_signs\cat_1753172316.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172326,my_recorded_signs\cat_1753172326.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172335,my_recorded_signs\cat_1753172335.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172344,my_recorded_signs\cat_1753172344.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172353,my_recorded_signs\cat_1753172353.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172363,my_recorded_signs\cat_1753172363.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172371,my_recorded_signs\cat_1753172371.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172380,my_recorded_signs\cat_1753172380.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172390,my_recorded_signs\cat_1753172390.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172399,my_recorded_signs\cat_1753172399.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172408,my_recorded_signs\cat_1753172408.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172418,my_recorded_signs\cat_1753172418.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172430,my_recorded_signs\cat_1753172430.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172440,my_recorded_signs\cat_1753172440.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172449,my_recorded_signs\cat_1753172449.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172462,my_recorded_signs\cat_1753172462.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172470,my_recorded_signs\cat_1753172470.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172479,my_recorded_signs\cat_1753172479.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172486,my_recorded_signs\cat_1753172486.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172495,my_recorded_signs\cat_1753172495.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172505,my_recorded_signs\cat_1753172505.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753172514,my_recorded_signs\cat_1753172514.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206359,my_recorded_signs\cat_1753206359.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206368,my_recorded_signs\cat_1753206368.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206377,my_recorded_signs\cat_1753206377.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206386,my_recorded_signs\cat_1753206386.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206395,my_recorded_signs\cat_1753206395.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206404,my_recorded_signs\cat_1753206404.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206412,my_recorded_signs\cat_1753206412.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206421,my_recorded_signs\cat_1753206421.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206431,my_recorded_signs\cat_1753206431.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753206440,my_recorded_signs\cat_1753206440.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263033,my_recorded_signs\cat_1753263033.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263044,my_recorded_signs\cat_1753263044.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263052,my_recorded_signs\cat_1753263052.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263060,my_recorded_signs\cat_1753263060.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263068,my_recorded_signs\cat_1753263068.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263076,my_recorded_signs\cat_1753263076.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263084,my_recorded_signs\cat_1753263084.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263092,my_recorded_signs\cat_1753263092.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263101,my_recorded_signs\cat_1753263101.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263109,my_recorded_signs\cat_1753263109.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263119,my_recorded_signs\cat_1753263119.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263129,my_recorded_signs\cat_1753263129.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263137,my_recorded_signs\cat_1753263137.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263146,my_recorded_signs\cat_1753263146.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263155,my_recorded_signs\cat_1753263155.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263164,my_recorded_signs\cat_1753263164.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263173,my_recorded_signs\cat_1753263173.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263183,my_recorded_signs\cat_1753263183.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263193,my_recorded_signs\cat_1753263193.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263202,my_recorded_signs\cat_1753263202.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263211,my_recorded_signs\cat_1753263211.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263220,my_recorded_signs\cat_1753263220.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263231,my_recorded_signs\cat_1753263231.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263240,my_recorded_signs\cat_1753263240.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263249,my_recorded_signs\cat_1753263249.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263258,my_recorded_signs\cat_1753263258.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263267,my_recorded_signs\cat_1753263267.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263276,my_recorded_signs\cat_1753263276.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263285,my_recorded_signs\cat_1753263285.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263294,my_recorded_signs\cat_1753263294.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263303,my_recorded_signs\cat_1753263303.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263313,my_recorded_signs\cat_1753263313.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263322,my_recorded_signs\cat_1753263322.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263331,my_recorded_signs\cat_1753263331.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753263341,my_recorded_signs\cat_1753263341.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272527,my_recorded_signs\cat_1753272527.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272572,my_recorded_signs\cat_1753272572.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272583,my_recorded_signs\cat_1753272583.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272592,my_recorded_signs\cat_1753272592.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272602,my_recorded_signs\cat_1753272602.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272611,my_recorded_signs\cat_1753272611.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272620,my_recorded_signs\cat_1753272620.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272628,my_recorded_signs\cat_1753272628.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272637,my_recorded_signs\cat_1753272637.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272647,my_recorded_signs\cat_1753272647.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272656,my_recorded_signs\cat_1753272656.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272667,my_recorded_signs\cat_1753272667.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272677,my_recorded_signs\cat_1753272677.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272688,my_recorded_signs\cat_1753272688.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272704,my_recorded_signs\cat_1753272704.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272713,my_recorded_signs\cat_1753272713.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272722,my_recorded_signs\cat_1753272722.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272732,my_recorded_signs\cat_1753272732.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272741,my_recorded_signs\cat_1753272741.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272751,my_recorded_signs\cat_1753272751.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272762,my_recorded_signs\cat_1753272762.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272775,my_recorded_signs\cat_1753272775.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272785,my_recorded_signs\cat_1753272785.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272796,my_recorded_signs\cat_1753272796.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272807,my_recorded_signs\cat_1753272807.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753272817,my_recorded_signs\cat_1753272817.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367412,my_recorded_signs\cat_1753367412.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367421,my_recorded_signs\cat_1753367421.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367435,my_recorded_signs\cat_1753367435.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367446,my_recorded_signs\cat_1753367446.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367454,my_recorded_signs\cat_1753367454.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367463,my_recorded_signs\cat_1753367463.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367472,my_recorded_signs\cat_1753367472.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367483,my_recorded_signs\cat_1753367483.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cat,6,personal_1753367493,my_recorded_signs\cat_1753367493.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085287,my_recorded_signs\chair_1753085287.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085321,my_recorded_signs\chair_1753085321.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085340,my_recorded_signs\chair_1753085340.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085356,my_recorded_signs\chair_1753085356.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085373,my_recorded_signs\chair_1753085373.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085385,my_recorded_signs\chair_1753085385.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085397,my_recorded_signs\chair_1753085397.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085418,my_recorded_signs\chair_1753085418.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085441,my_recorded_signs\chair_1753085441.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085454,my_recorded_signs\chair_1753085454.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085471,my_recorded_signs\chair_1753085471.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085483,my_recorded_signs\chair_1753085483.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085495,my_recorded_signs\chair_1753085495.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085515,my_recorded_signs\chair_1753085515.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753085531,my_recorded_signs\chair_1753085531.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171870,my_recorded_signs\chair_1753171870.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171890,my_recorded_signs\chair_1753171890.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171899,my_recorded_signs\chair_1753171899.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171908,my_recorded_signs\chair_1753171908.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171918,my_recorded_signs\chair_1753171918.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171927,my_recorded_signs\chair_1753171927.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171937,my_recorded_signs\chair_1753171937.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171945,my_recorded_signs\chair_1753171945.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171954,my_recorded_signs\chair_1753171954.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171963,my_recorded_signs\chair_1753171963.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171976,my_recorded_signs\chair_1753171976.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171985,my_recorded_signs\chair_1753171985.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753171994,my_recorded_signs\chair_1753171994.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753172003,my_recorded_signs\chair_1753172003.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753172013,my_recorded_signs\chair_1753172013.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753172022,my_recorded_signs\chair_1753172022.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753172030,my_recorded_signs\chair_1753172030.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753206977,my_recorded_signs\chair_1753206977.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753206986,my_recorded_signs\chair_1753206986.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753206994,my_recorded_signs\chair_1753206994.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753207003,my_recorded_signs\chair_1753207003.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753207012,my_recorded_signs\chair_1753207012.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366491,my_recorded_signs\chair_1753366491.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366500,my_recorded_signs\chair_1753366500.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366508,my_recorded_signs\chair_1753366508.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366516,my_recorded_signs\chair_1753366516.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366526,my_recorded_signs\chair_1753366526.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366535,my_recorded_signs\chair_1753366535.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366544,my_recorded_signs\chair_1753366544.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366553,my_recorded_signs\chair_1753366553.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366562,my_recorded_signs\chair_1753366562.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366571,my_recorded_signs\chair_1753366571.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366579,my_recorded_signs\chair_1753366579.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366588,my_recorded_signs\chair_1753366588.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366596,my_recorded_signs\chair_1753366596.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366609,my_recorded_signs\chair_1753366609.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366617,my_recorded_signs\chair_1753366617.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366626,my_recorded_signs\chair_1753366626.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+chair,0,personal_1753366635,my_recorded_signs\chair_1753366635.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753174998,my_recorded_signs\child_1753174998.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175009,my_recorded_signs\child_1753175009.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175016,my_recorded_signs\child_1753175016.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175025,my_recorded_signs\child_1753175025.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175033,my_recorded_signs\child_1753175033.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175041,my_recorded_signs\child_1753175041.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175050,my_recorded_signs\child_1753175050.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175059,my_recorded_signs\child_1753175059.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175068,my_recorded_signs\child_1753175068.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175076,my_recorded_signs\child_1753175076.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175085,my_recorded_signs\child_1753175085.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175093,my_recorded_signs\child_1753175093.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175102,my_recorded_signs\child_1753175102.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175110,my_recorded_signs\child_1753175110.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175127,my_recorded_signs\child_1753175127.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175136,my_recorded_signs\child_1753175136.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175144,my_recorded_signs\child_1753175144.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175152,my_recorded_signs\child_1753175152.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175160,my_recorded_signs\child_1753175160.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175168,my_recorded_signs\child_1753175168.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175177,my_recorded_signs\child_1753175177.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175184,my_recorded_signs\child_1753175184.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175194,my_recorded_signs\child_1753175194.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175203,my_recorded_signs\child_1753175203.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175212,my_recorded_signs\child_1753175212.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175221,my_recorded_signs\child_1753175221.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175230,my_recorded_signs\child_1753175230.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175239,my_recorded_signs\child_1753175239.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175249,my_recorded_signs\child_1753175249.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175258,my_recorded_signs\child_1753175258.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175268,my_recorded_signs\child_1753175268.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753175277,my_recorded_signs\child_1753175277.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208132,my_recorded_signs\child_1753208132.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208143,my_recorded_signs\child_1753208143.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208152,my_recorded_signs\child_1753208152.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208161,my_recorded_signs\child_1753208161.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208170,my_recorded_signs\child_1753208170.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208179,my_recorded_signs\child_1753208179.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208187,my_recorded_signs\child_1753208187.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208196,my_recorded_signs\child_1753208196.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208204,my_recorded_signs\child_1753208204.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208213,my_recorded_signs\child_1753208213.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208221,my_recorded_signs\child_1753208221.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208229,my_recorded_signs\child_1753208229.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208239,my_recorded_signs\child_1753208239.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208247,my_recorded_signs\child_1753208247.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208256,my_recorded_signs\child_1753208256.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208266,my_recorded_signs\child_1753208266.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208275,my_recorded_signs\child_1753208275.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208284,my_recorded_signs\child_1753208284.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208295,my_recorded_signs\child_1753208295.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208303,my_recorded_signs\child_1753208303.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753208315,my_recorded_signs\child_1753208315.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263792,my_recorded_signs\child_1753263792.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263801,my_recorded_signs\child_1753263801.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263809,my_recorded_signs\child_1753263809.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263817,my_recorded_signs\child_1753263817.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263825,my_recorded_signs\child_1753263825.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263834,my_recorded_signs\child_1753263834.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263843,my_recorded_signs\child_1753263843.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263851,my_recorded_signs\child_1753263851.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263860,my_recorded_signs\child_1753263860.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263868,my_recorded_signs\child_1753263868.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263878,my_recorded_signs\child_1753263878.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263887,my_recorded_signs\child_1753263887.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263896,my_recorded_signs\child_1753263896.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263905,my_recorded_signs\child_1753263905.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263914,my_recorded_signs\child_1753263914.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263923,my_recorded_signs\child_1753263923.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263933,my_recorded_signs\child_1753263933.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263943,my_recorded_signs\child_1753263943.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753263953,my_recorded_signs\child_1753263953.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753272883,my_recorded_signs\child_1753272883.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753272892,my_recorded_signs\child_1753272892.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753272903,my_recorded_signs\child_1753272903.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753272914,my_recorded_signs\child_1753272914.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753272924,my_recorded_signs\child_1753272924.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753272934,my_recorded_signs\child_1753272934.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273064,my_recorded_signs\child_1753273064.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273073,my_recorded_signs\child_1753273073.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273082,my_recorded_signs\child_1753273082.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273091,my_recorded_signs\child_1753273091.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273104,my_recorded_signs\child_1753273104.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273135,my_recorded_signs\child_1753273135.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273166,my_recorded_signs\child_1753273166.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273180,my_recorded_signs\child_1753273180.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273188,my_recorded_signs\child_1753273188.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273198,my_recorded_signs\child_1753273198.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273208,my_recorded_signs\child_1753273208.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273219,my_recorded_signs\child_1753273219.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273230,my_recorded_signs\child_1753273230.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753273239,my_recorded_signs\child_1753273239.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367664,my_recorded_signs\child_1753367664.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367672,my_recorded_signs\child_1753367672.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367681,my_recorded_signs\child_1753367681.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367689,my_recorded_signs\child_1753367689.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367698,my_recorded_signs\child_1753367698.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367707,my_recorded_signs\child_1753367707.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+child,8,personal_1753367716,my_recorded_signs\child_1753367716.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087366,my_recorded_signs\cold_1753087366.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087374,my_recorded_signs\cold_1753087374.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087383,my_recorded_signs\cold_1753087383.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087391,my_recorded_signs\cold_1753087391.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087400,my_recorded_signs\cold_1753087400.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087409,my_recorded_signs\cold_1753087409.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087418,my_recorded_signs\cold_1753087418.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087428,my_recorded_signs\cold_1753087428.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087437,my_recorded_signs\cold_1753087437.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087448,my_recorded_signs\cold_1753087448.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087457,my_recorded_signs\cold_1753087457.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087466,my_recorded_signs\cold_1753087466.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087478,my_recorded_signs\cold_1753087478.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087494,my_recorded_signs\cold_1753087494.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087502,my_recorded_signs\cold_1753087502.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087510,my_recorded_signs\cold_1753087510.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087519,my_recorded_signs\cold_1753087519.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753087528,my_recorded_signs\cold_1753087528.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753172528,my_recorded_signs\cold_1753172528.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753172537,my_recorded_signs\cold_1753172537.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753172546,my_recorded_signs\cold_1753172546.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174197,my_recorded_signs\cold_1753174197.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174209,my_recorded_signs\cold_1753174209.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174220,my_recorded_signs\cold_1753174220.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174231,my_recorded_signs\cold_1753174231.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174250,my_recorded_signs\cold_1753174250.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174265,my_recorded_signs\cold_1753174265.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174274,my_recorded_signs\cold_1753174274.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174283,my_recorded_signs\cold_1753174283.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174293,my_recorded_signs\cold_1753174293.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174301,my_recorded_signs\cold_1753174301.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174310,my_recorded_signs\cold_1753174310.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174320,my_recorded_signs\cold_1753174320.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174329,my_recorded_signs\cold_1753174329.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174338,my_recorded_signs\cold_1753174338.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753174347,my_recorded_signs\cold_1753174347.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206239,my_recorded_signs\cold_1753206239.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206248,my_recorded_signs\cold_1753206248.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206258,my_recorded_signs\cold_1753206258.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206289,my_recorded_signs\cold_1753206289.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206300,my_recorded_signs\cold_1753206300.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206309,my_recorded_signs\cold_1753206309.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206320,my_recorded_signs\cold_1753206320.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206329,my_recorded_signs\cold_1753206329.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206339,my_recorded_signs\cold_1753206339.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753206348,my_recorded_signs\cold_1753206348.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263631,my_recorded_signs\cold_1753263631.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263639,my_recorded_signs\cold_1753263639.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263648,my_recorded_signs\cold_1753263648.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263657,my_recorded_signs\cold_1753263657.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263666,my_recorded_signs\cold_1753263666.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263675,my_recorded_signs\cold_1753263675.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263684,my_recorded_signs\cold_1753263684.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263694,my_recorded_signs\cold_1753263694.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263703,my_recorded_signs\cold_1753263703.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263713,my_recorded_signs\cold_1753263713.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263722,my_recorded_signs\cold_1753263722.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263733,my_recorded_signs\cold_1753263733.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263741,my_recorded_signs\cold_1753263741.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263749,my_recorded_signs\cold_1753263749.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263757,my_recorded_signs\cold_1753263757.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263765,my_recorded_signs\cold_1753263765.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263774,my_recorded_signs\cold_1753263774.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753263782,my_recorded_signs\cold_1753263782.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367505,my_recorded_signs\cold_1753367505.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367513,my_recorded_signs\cold_1753367513.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367521,my_recorded_signs\cold_1753367521.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367530,my_recorded_signs\cold_1753367530.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367538,my_recorded_signs\cold_1753367538.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367547,my_recorded_signs\cold_1753367547.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367555,my_recorded_signs\cold_1753367555.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367569,my_recorded_signs\cold_1753367569.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367578,my_recorded_signs\cold_1753367578.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367587,my_recorded_signs\cold_1753367587.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+cold,7,personal_1753367595,my_recorded_signs\cold_1753367595.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087544,my_recorded_signs\come_1753087544.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087553,my_recorded_signs\come_1753087553.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087561,my_recorded_signs\come_1753087561.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087569,my_recorded_signs\come_1753087569.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087579,my_recorded_signs\come_1753087579.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087589,my_recorded_signs\come_1753087589.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087597,my_recorded_signs\come_1753087597.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087606,my_recorded_signs\come_1753087606.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087614,my_recorded_signs\come_1753087614.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087630,my_recorded_signs\come_1753087630.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087639,my_recorded_signs\come_1753087639.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087648,my_recorded_signs\come_1753087648.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087657,my_recorded_signs\come_1753087657.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087667,my_recorded_signs\come_1753087667.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087676,my_recorded_signs\come_1753087676.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087687,my_recorded_signs\come_1753087687.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087696,my_recorded_signs\come_1753087696.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753087707,my_recorded_signs\come_1753087707.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206777,my_recorded_signs\come_1753206777.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206786,my_recorded_signs\come_1753206786.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206794,my_recorded_signs\come_1753206794.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206804,my_recorded_signs\come_1753206804.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206813,my_recorded_signs\come_1753206813.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206822,my_recorded_signs\come_1753206822.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206831,my_recorded_signs\come_1753206831.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206841,my_recorded_signs\come_1753206841.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206849,my_recorded_signs\come_1753206849.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753206865,my_recorded_signs\come_1753206865.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262044,my_recorded_signs\come_1753262044.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262052,my_recorded_signs\come_1753262052.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262061,my_recorded_signs\come_1753262061.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262070,my_recorded_signs\come_1753262070.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262079,my_recorded_signs\come_1753262079.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262087,my_recorded_signs\come_1753262087.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262096,my_recorded_signs\come_1753262096.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262105,my_recorded_signs\come_1753262105.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262115,my_recorded_signs\come_1753262115.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262125,my_recorded_signs\come_1753262125.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262135,my_recorded_signs\come_1753262135.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262144,my_recorded_signs\come_1753262144.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262158,my_recorded_signs\come_1753262158.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262167,my_recorded_signs\come_1753262167.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262176,my_recorded_signs\come_1753262176.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262184,my_recorded_signs\come_1753262184.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262193,my_recorded_signs\come_1753262193.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262202,my_recorded_signs\come_1753262202.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262210,my_recorded_signs\come_1753262210.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262218,my_recorded_signs\come_1753262218.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262226,my_recorded_signs\come_1753262226.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262236,my_recorded_signs\come_1753262236.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262244,my_recorded_signs\come_1753262244.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262253,my_recorded_signs\come_1753262253.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753262261,my_recorded_signs\come_1753262261.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273473,my_recorded_signs\come_1753273473.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273483,my_recorded_signs\come_1753273483.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273493,my_recorded_signs\come_1753273493.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273503,my_recorded_signs\come_1753273503.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273512,my_recorded_signs\come_1753273512.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273522,my_recorded_signs\come_1753273522.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273532,my_recorded_signs\come_1753273532.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273543,my_recorded_signs\come_1753273543.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273554,my_recorded_signs\come_1753273554.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273563,my_recorded_signs\come_1753273563.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273572,my_recorded_signs\come_1753273572.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273580,my_recorded_signs\come_1753273580.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273589,my_recorded_signs\come_1753273589.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273615,my_recorded_signs\come_1753273615.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273627,my_recorded_signs\come_1753273627.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273636,my_recorded_signs\come_1753273636.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273650,my_recorded_signs\come_1753273650.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273660,my_recorded_signs\come_1753273660.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273680,my_recorded_signs\come_1753273680.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273689,my_recorded_signs\come_1753273689.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273702,my_recorded_signs\come_1753273702.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273711,my_recorded_signs\come_1753273711.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273723,my_recorded_signs\come_1753273723.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273734,my_recorded_signs\come_1753273734.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273744,my_recorded_signs\come_1753273744.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273754,my_recorded_signs\come_1753273754.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273765,my_recorded_signs\come_1753273765.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273775,my_recorded_signs\come_1753273775.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273786,my_recorded_signs\come_1753273786.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273798,my_recorded_signs\come_1753273798.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753273810,my_recorded_signs\come_1753273810.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753367611,my_recorded_signs\come_1753367611.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753367619,my_recorded_signs\come_1753367619.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753367627,my_recorded_signs\come_1753367627.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753367636,my_recorded_signs\come_1753367636.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753367644,my_recorded_signs\come_1753367644.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"
+come,9,personal_1753367652,my_recorded_signs\come_1753367652.npy,personal,train,1,-1,25,"[0, 0, 0, 0]"

ai_model/words/wlasl_10_words_processed.csv

@@ -0,0 +1,90 @@
+gloss,gloss_id,video_id,video_path,signer_id,split,frame_start,frame_end,fps,bbox
+chair,0,09848,.\videos\09848.mp4,0,train,1,-1,25,"[40, 6, 266, 240]"
+chair,0,09869,.\videos\09869.mp4,12,train,1,-1,25,"[183, 55, 560, 400]"
+chair,0,09849,.\videos\09849.mp4,38,train,1,-1,25,"[388, 56, 806, 720]"
+chair,0,09850,.\videos\09850.mp4,31,train,1,-1,25,"[667, 144, 1472, 1061]"
+chair,0,09851,.\videos\09851.mp4,59,val,1,-1,25,"[102, 22, 563, 480]"
+chair,0,65328,.\videos\65328.mp4,90,train,1,-1,25,"[169, 15, 476, 370]"
+chair,0,09854,.\videos\09854.mp4,11,val,1,-1,25,"[81, 15, 233, 192]"
+black,1,69236,.\videos\69236.mp4,118,val,1,-1,25,"[295, 38, 885, 720]"
+black,1,06472,.\videos\06472.mp4,36,train,1,-1,25,"[205, 17, 524, 414]"
+black,1,06473,.\videos\06473.mp4,4,train,1,-1,25,"[544, 60, 1618, 1080]"
+black,1,06474,.\videos\06474.mp4,59,train,1,-1,25,"[77, 18, 515, 480]"
+black,1,06476,.\videos\06476.mp4,11,val,1,-1,25,"[58, 15, 216, 192]"
+black,1,06477,.\videos\06477.mp4,11,train,1,-1,25,"[48, 11, 226, 192]"
+black,1,06478,.\videos\06478.mp4,11,train,1,-1,25,"[63, 8, 245, 192]"
+black,1,65200,.\videos\65200.mp4,90,train,1,-1,25,"[149, 24, 479, 370]"
+black,1,06471,.\videos\06471.mp4,56,train,1,-1,25,"[66, 13, 258, 240]"
+black,1,06486,.\videos\06486.mp4,12,train,1,-1,25,"[131, 52, 555, 400]"
+blue,2,69238,.\videos\69238.mp4,118,val,1,-1,25,"[257, 39, 881, 720]"
+blue,2,06835,.\videos\06835.mp4,59,train,1,-1,25,"[125, 26, 528, 480]"
+blue,2,06839,.\videos\06839.mp4,10,train,1,-1,25,"[90, 21, 213, 192]"
+blue,2,65216,.\videos\65216.mp4,90,train,1,-1,25,"[158, 24, 482, 370]"
+blue,2,06832,.\videos\06832.mp4,8,train,1,-1,25,"[41, 0, 232, 240]"
+blue,2,06845,.\videos\06845.mp4,41,test,1,-1,25,"[232, 39, 483, 400]"
+blue,2,06833,.\videos\06833.mp4,36,train,1,-1,25,"[225, 18, 519, 414]"
+blue,2,06834,.\videos\06834.mp4,46,train,1,-1,25,"[188, 0, 794, 720]"
+can,3,69257,.\videos\69257.mp4,118,test,1,-1,25,"[285, 31, 964, 720]"
+can,3,65294,.\videos\65294.mp4,90,val,1,-1,25,"[137, 15, 500, 370]"
+can,3,08955,.\videos\08955.mp4,12,train,1,-1,25,"[163, 54, 555, 400]"
+can,3,08935,.\videos\08935.mp4,38,train,1,-1,25,"[434, 54, 814, 720]"
+can,3,08936,.\videos\08936.mp4,70,train,1,-1,25,"[201, 5, 823, 480]"
+can,3,08937,.\videos\08937.mp4,59,train,1,-1,25,"[69, 17, 552, 480]"
+can,3,08938,.\videos\08938.mp4,59,train,1,-1,25,"[110, 29, 555, 480]"
+can,3,08942,.\videos\08942.mp4,11,train,1,-1,25,"[76, 18, 237, 192]"
+can,3,08944,.\videos\08944.mp4,11,train,1,-1,25,"[70, 18, 217, 192]"
+apple,4,69213,.\videos\69213.mp4,117,val,1,-1,25,"[304, 41, 916, 720]"
+apple,4,03005,.\videos\03005.mp4,11,train,1,-1,25,"[64, 15, 229, 192]"
+apple,4,03008,.\videos\03008.mp4,12,train,1,-1,25,"[158, 50, 541, 399]"
+apple,4,65085,.\videos\65085.mp4,89,train,1,-1,25,"[207, 0, 520, 370]"
+apple,4,65086,.\videos\65086.mp4,89,val,1,-1,25,"[206, 0, 535, 370]"
+apple,4,65084,.\videos\65084.mp4,88,train,1,-1,25,"[119, 4, 486, 370]"
+apple,4,02999,.\videos\02999.mp4,37,val,1,-1,25,"[44, 0, 232, 240]"
+apple,4,03000,.\videos\03000.mp4,28,train,1,-1,25,"[41, 13, 259, 240]"
+apple,4,03001,.\videos\03001.mp4,38,train,1,-1,25,"[350, 53, 815, 720]"
+apple,4,03002,.\videos\03002.mp4,36,train,1,-1,25,"[212, 17, 526, 414]"
+apple,4,03003,.\videos\03003.mp4,13,train,1,-1,25,"[24, 13, 526, 480]"
+brown,5,69252,.\videos\69252.mp4,118,test,1,-1,25,"[335, 37, 874, 720]"
+brown,5,07963,.\videos\07963.mp4,59,train,1,-1,25,"[102, 30, 523, 480]"
+brown,5,07966,.\videos\07966.mp4,11,train,1,-1,25,"[62, 15, 217, 192]"
+brown,5,65263,.\videos\65263.mp4,90,train,1,-1,25,"[158, 12, 462, 370]"
+brown,5,07960,.\videos\07960.mp4,2,train,1,-1,25,"[37, 6, 250, 240]"
+brown,5,07973,.\videos\07973.mp4,12,val,1,-1,25,"[182, 52, 542, 400]"
+brown,5,07961,.\videos\07961.mp4,36,train,1,-1,25,"[213, 11, 525, 414]"
+brown,5,07962,.\videos\07962.mp4,4,train,1,-1,25,"[381, 38, 1076, 720]"
+cat,6,69261,.\videos\69261.mp4,118,test,1,-1,25,"[357, 37, 874, 720]"
+cat,6,09528,.\videos\09528.mp4,32,train,1,-1,25,"[743, 79, 1579, 1066]"
+cat,6,09532,.\videos\09532.mp4,10,val,1,-1,25,"[84, 21, 216, 192]"
+cat,6,09533,.\videos\09533.mp4,10,train,1,-1,25,"[83, 20, 218, 192]"
+cat,6,65312,.\videos\65312.mp4,102,train,1,-1,25,"[156, 19, 486, 370]"
+cat,6,65313,.\videos\65313.mp4,101,train,1,-1,25,"[168, 18, 462, 370]"
+cat,6,09525,.\videos\09525.mp4,2,val,1,-1,25,"[57, 0, 271, 240]"
+cat,6,09538,.\videos\09538.mp4,41,train,1,-1,25,"[215, 42, 465, 400]"
+cat,6,09526,.\videos\09526.mp4,36,train,1,-1,25,"[210, 18, 523, 414]"
+cold,7,65377,.\videos\65377.mp4,99,test,1,-1,25,"[129, 16, 476, 370]"
+cold,7,11638,.\videos\11638.mp4,12,test,1,-1,25,"[180, 57, 551, 400]"
+cold,7,11622,.\videos\11622.mp4,36,train,1,-1,25,"[210, 16, 540, 414]"
+cold,7,11623,.\videos\11623.mp4,52,train,1,-1,25,"[380, 16, 825, 480]"
+cold,7,11624,.\videos\11624.mp4,6,train,1,-1,25,"[593, 70, 1468, 1080]"
+cold,7,11625,.\videos\11625.mp4,39,val,1,-1,25,"[624, 133, 1621, 1072]"
+cold,7,11627,.\videos\11627.mp4,4,train,1,-1,25,"[411, 31, 1128, 720]"
+cold,7,11628,.\videos\11628.mp4,59,val,1,-1,25,"[116, 29, 513, 480]"
+cold,7,11621,.\videos\11621.mp4,56,train,1,-1,25,"[59, 16, 263, 240]"
+cold,7,11633,.\videos\11633.mp4,10,train,1,-1,25,"[88, 22, 216, 192]"
+cold,7,11634,.\videos\11634.mp4,10,val,1,-1,25,"[65, 21, 233, 192]"
+cold,7,11635,.\videos\11635.mp4,11,train,1,-1,25,"[79, 16, 221, 192]"
+child,8,10450,.\videos\10450.mp4,11,train,1,-1,25,"[69, 17, 215, 192]"
+child,8,65351,.\videos\65351.mp4,90,train,1,-1,25,"[130, 14, 465, 370]"
+child,8,10443,.\videos\10443.mp4,2,train,1,-1,25,"[64, 0, 269, 240]"
+child,8,10453,.\videos\10453.mp4,12,train,1,-1,25,"[130, 54, 550, 400]"
+child,8,10444,.\videos\10444.mp4,32,train,1,-1,25,"[700, 66, 1636, 1073]"
+child,8,10445,.\videos\10445.mp4,4,train,1,-1,25,"[616, 61, 1702, 1079]"
+child,8,10446,.\videos\10446.mp4,32,test,1,-1,25,"[693, 70, 1631, 1073]"
+child,8,10447,.\videos\10447.mp4,59,train,1,-1,25,"[50, 24, 520, 480]"
+child,8,10449,.\videos\10449.mp4,10,val,1,-1,25,"[50, 17, 219, 192]"
+come,9,69275,.\videos\69275.mp4,115,test,1,-1,25,"[140, 37, 943, 720]"
+come,9,11884,.\videos\11884.mp4,12,train,1,-1,25,"[178, 59, 534, 400]"
+come,9,11876,.\videos\11876.mp4,56,train,1,-1,25,"[63, 13, 261, 240]"
+come,9,11877,.\videos\11877.mp4,31,train,1,-1,25,"[540, 64, 1480, 1080]"
+come,9,11878,.\videos\11878.mp4,59,val,1,-1,25,"[84, 27, 590, 480]"
+come,9,11882,.\videos\11882.mp4,11,train,1,-1,25,"[82, 14, 230, 192]"