final update

Dockerfile

@@ -1,35 +1,37 @@
-# Use Python 3.11, as indicated by your traceback
+# 1. Base Image Declaration (Corrected 'slim' tag)
 FROM python:3.11-slim
 
+# 2. Environment Variables & Cleanup
+# Suppress Matplotlib cache permission warning
 ENV MPLCONFIGDIR /tmp/.matplotlib
 
 # Explicitly force CPU and suppress CUDA warnings (if not using a GPU space)
 ENV CUDA_VISIBLE_DEVICES=""
-# Install system dependencies required by OpenCV (cv2)
-# The 'libgl1-mesa-glx' package provides libGL.so.1
+
+# Ensure Python output is immediately available in logs
+ENV PYTHONUNBUFFERED True
+
+# Set the port required by Hugging Face Spaces/Uvicorn
+ENV PORT 7860
+
+# 3. System Dependencies (Required by OpenCV/cv2)
 RUN apt-get update && \
     apt-get install -y libgl1 libglib2.0-0 libsm6 libxext6 && \
     rm -rf /var/lib/apt/lists/*
 
-# Set environment variables for better output and the internal port
-ENV PYTHONUNBUFFERED True
-ENV PORT 5000
-
-# Create the working directory inside the container
+# 4. Working Directory
 WORKDIR /app
 
-# Copy requirements file and install dependencies first (for faster caching)
+# 5. Dependency Installation (Leverages Docker caching)
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Copy the rest of the application code
+# 6. Application Code
 COPY . .
 
-# Expose the application port
-ENV PYTHONUNBUFFERED True
-EXPOSE 5000
+# 7. Port Documentation (EXPOSE is optional but good practice)
+EXPOSE 7860
 
-WORKDIR /app
-# Define the command to start the Uvicorn server
-# This runs your application: `app` instance in the file `app.py`
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
+# 8. Define the command to start the Uvicorn server
+# This runs the 'app' object in the 'app.py' module, listening on port 7860.
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

lettersController.py

@@ -4,17 +4,17 @@ import pickle
 import tensorflow as tf
 import mediapipe as mp
 
-# lettersModel = tf.keras.models.load_model('ai_model/models/detectLettersModel.keras')
-# with open('ai_model/models/labelEncoder.pickle', 'rb') as f:
-#     labelEncoder = pickle.load(f)
+lettersModel = tf.keras.models.load_model('ai_model/models/detectLettersModel.keras')
+with open('ai_model/models/labelEncoder.pickle', 'rb') as f:
+    labelEncoder = pickle.load(f)
 
-# lettersModel2 = tf.keras.models.load_model('ai_model/jz_model/JZModel.keras')
-# with open('ai_model/jz_model/labelEncoder.pickle', 'rb') as f:
-#     labelEncoder2 = pickle.load(f)
+lettersModel2 = tf.keras.models.load_model('ai_model/jz_model/JZModel.keras')
+with open('ai_model/jz_model/labelEncoder.pickle', 'rb') as f:
+    labelEncoder2 = pickle.load(f)
 
-# numbersModel = tf.keras.models.load_model('ai_model/models/detectNumbersModel.keras')
-# with open('ai_model/models/numLabelEncoder.pickle', 'rb') as f:
-#     numLabelEncoder = pickle.load(f)
+numbersModel = tf.keras.models.load_model('ai_model/models/detectNumbersModel.keras')
+with open('ai_model/models/numLabelEncoder.pickle', 'rb') as f:
+    numLabelEncoder = pickle.load(f)
 
 sequenceNum = 20
 hands = mp.solutions.hands.Hands(static_image_mode=True)

test_startup.py

@@ -1,27 +0,0 @@
-# test_startup.py
-
-print("--- Starting Application Object Test ---")
-
-# 1. Import the necessary components
-import os
-import sys
-
-# Add the application root to the path so imports work
-sys.path.append(os.path.dirname(os.path.abspath(__file__)))
-
-# 2. Try to import the application file and object
-try:
-    # Assuming your main file is 'app.py' and the object is 'app'
-    from app import app
-    print("SUCCESS: FastAPI object ('app') imported successfully.")
-
-    # You could add a small model call here if you want to test further
-    # e.g., print(app.dependency_on_model.status)
-
-except Exception as e:
-    print(f"FAILURE: An error occurred during application import.")
-    print(f"Error details: {e}")
-    sys.exit(1)
-
-print("--- Startup Test Finished Successfully ---")
-sys.exit(0)

wordsController.py

@@ -10,13 +10,10 @@ SEQUENCE_LENGTH = 90
 EXPECTED_COORDS_PER_FRAME = 1662
 CONFIDENCE_THRESHOLD = 0.1
 
-# model = load_model(MODEL_PATH)
-# df = pd.read_csv(CSV_PATH)
-# unique_glosses = df['gloss'].unique()
-# id_to_gloss = {i: g for i, g in enumerate(unique_glosses)}
-# Insert these lines immediately after the commented-out block
-model = None # Placeholder
-id_to_gloss = {0: "placeholder_word"} # Minimal placeholder for the dictionary
+model = load_model(MODEL_PATH)
+df = pd.read_csv(CSV_PATH)
+unique_glosses = df['gloss'].unique()
+id_to_gloss = {i: g for i, g in enumerate(unique_glosses)}
 
 mp_holistic = mp.solutions.holistic.Holistic(
     static_image_mode=True,
@@ -132,15 +129,11 @@ def detectWords(image_paths):
     sequence = pad_or_truncate_sequence(sequence, SEQUENCE_LENGTH, EXPECTED_COORDS_PER_FRAME)
     sequence = np.expand_dims(sequence, axis=0) 
 
-    # preds = model.predict(sequence, verbose=0)
-    # predicted_id = int(np.argmax(preds))
-    # confidence = float(np.max(preds))
-    # predicted_word = id_to_gloss.get(predicted_id, "Unknown")
-
-    predicted_id = 0 # Use the placeholder ID
-    confidence = 0.99 # Use a dummy confidence
+    preds = model.predict(sequence, verbose=0)
+    predicted_id = int(np.argmax(preds))
+    confidence = float(np.max(preds))
     predicted_word = id_to_gloss.get(predicted_id, "Unknown")
-    
+
     result = {"word": predicted_word if confidence >= CONFIDENCE_THRESHOLD else "",
               "confidence": confidence}