feat: add fp

app.py

@@ -0,0 +1,135 @@
+import cv2
+import numpy as np
+import mediapipe as mp
+import tensorflow as tf
+from tensorflow.keras.models import Model
+import gradio as gr
+
+# Load model architecture and weights (replace with your actual model setup)
+# def build_bilstm_model(num_vocabs=13, num_frames=19, num_landmarks=42):
+#     # Your model architecture here
+#     pass
+
+# model = build_bilstm_model()
+# model.load_weights('final_model.keras')
+
+
+model = tf.keras.models.load_model('final_model.keras', compile=False)
+
+# Load labels
+with open('label.txt', 'r') as f:
+    label = f.readline().split()
+
+# MediaPipe setup
+mp_holistic = mp.solutions.holistic
+mp_drawing = mp.solutions.drawing_utils
+holistic = mp_holistic.Holistic(
+    min_detection_confidence=0.5, 
+    min_tracking_confidence=0.5
+)
+
+# Global state
+is_recording = False
+sequence = []
+sentence = []
+sequence_length = 19
+threshold = 0.5
+
+# Landmark drawing styles
+STYLES = {
+    "left_hand": ((121, 22, 76), (121, 44, 250), 2, 4),
+    "right_hand": ((245, 117, 66), (245, 66, 230), 2, 4)
+}
+
+def mediapipe_detection(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 draw_styled_landmarks(image, results):
+    for landmark_type in ["left_hand", "right_hand"]:
+        landmarks, connections = (
+            (results.left_hand_landmarks, mp_holistic.HAND_CONNECTIONS) if landmark_type == "left_hand" 
+            else (results.right_hand_landmarks, mp_holistic.HAND_CONNECTIONS)
+        )
+        if landmarks:
+            color, connection_color, thickness, radius = STYLES[landmark_type]
+            mp_drawing.draw_landmarks(
+                image,
+                landmarks,
+                connections,
+                mp_drawing.DrawingSpec(color=color, thickness=thickness, circle_radius=radius),
+                mp_drawing.DrawingSpec(color=connection_color, thickness=thickness, circle_radius=radius//2)
+            )
+
+def extract_landmarks(results):
+    lh = np.array([[res.x, res.y, res.z] for res in results.left_hand_landmarks.landmark]).flatten() if results.left_hand_landmarks else np.zeros(21*3)
+    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 np.concatenate([lh, rh])
+
+def process_frame(image):
+    global is_recording, sequence, sentence
+    
+    if is_recording:
+        # Process frame with MediaPipe
+        image, results = mediapipe_detection(image, holistic)
+        draw_styled_landmarks(image, results)
+        
+        # Extract landmarks and update sequence
+        landmarks = extract_landmarks(results)
+        sequence.append(landmarks)
+        sequence = sequence[-30:]  # Keep last 30 frames
+        
+        # Make prediction when sequence is complete
+        if len(sequence) == sequence_length:
+            res = model.predict(np.expand_dims(sequence, axis=0))[0]
+            predicted_label = label[np.argmax(res)]
+            confidence = np.max(res)
+            
+            if confidence > threshold:
+                if not sentence or sentence[-1] != predicted_label:
+                    sentence.append(predicted_label)
+                    sentence = sentence[-5:]  # Keep last 5 predictions
+            
+            sequence = []  # Reset sequence
+        
+        # Draw prediction text
+        cv2.rectangle(image, (0,0), (640, 40), (245, 117, 16), -1)
+        cv2.putText(image, ' '.join(sentence), (3,30), 
+                   cv2.FONT_HERSHEY_SIMPLEX, 1, (255,255,255), 2)
+    else:
+        # Draw instruction text
+        cv2.rectangle(image, (50, 50), (380, 100), (0, 255, 0), -1)
+        cv2.putText(image, "Press START to begin", (60,85),
+                   cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255,255,255), 2)
+    
+    return image
+
+def toggle_recording():
+    global is_recording
+    is_recording = not is_recording
+    return is_recording
+
+with gr.Blocks() as demo:
+    gr.Markdown("# Sign Language Detection 👐")
+    gr.Markdown("Aplikasi deteksi bahasa isyarat menggunakan MediaPipe dan TensorFlow")
+    
+    with gr.Row():
+        webcam = gr.Image(label="Webcam Input", source="webcam", streaming=True)
+        output = gr.Image(label="Processed Output")
+    
+    btn = gr.Button("Start/Stop Recording")
+    btn.click(toggle_recording)
+    
+    webcam.stream(
+        fn=process_frame,
+        inputs=webcam,
+        outputs=output,
+        every=0.1
+    )
+
+if __name__ == "__main__":
+    demo.launch()

label.txt

@@ -0,0 +1 @@
+hai nama kamu pagi siang malam siapa sudah belum makan suka selamat aku