How pranav6266/realtime-handsign-to-audio-converter works
Project type and purpose
• This is a desktop Python application for real-time American Sign Language (ASL) recognition and audio output. It includes two logical subsystems:
a) Offline training pipeline (data collection → dataset creation → model training)
b) Real-time inference pipeline (camera input → hand-landmark detection → classification → audio output)
File structure
Top-level files and folders:
• collect_images.py, create_dataset.py, train_classifer.py → training pipeline scripts
• hand-sign-recognizer_copy.py → main inference application
• labels_alphabet.txt, labels_words.txt → class label definitions
• model_alphabet.p, model_words.p → serialized classification models
• Words/ (presumably contains sample word images)
• Requirements: OpenCV, MediaPipe, NumPy, Tkinter, pyttsx3, pickle
README for additional insights
• The README confirms two models (alphabet & words), real-time webcam capture, 50%+ accuracy threshold over 10 frames, and audio output via pyttsx3.