Sign Language Detection

A real-time Sign Language Detection application using LSTM and MediaPipe. This project captures motion data from hand gestures and translates them into text/sign labels.

Project Structure

• app.py: Main application entry point.
• collectdata.py: Script to collect data for training new signs.
• trainmodel.py: Script to train the LSTM model.
• function.py: Helper functions for MediaPipe detection.
• MP_Data/: Pre-processed numpy arrays used for training/inference.
• Models/: Contains trained model weights (model.h5, etc.).
• docs/: Project documentation (Reports, Research Paper, etc.).

Dataset

The raw video dataset for this project is available on Kaggle: Link to Kaggle Dataset

How to Run

1. Install Dependencies Ensure you have the required Python packages (TensorFlow, OpenCV, MediaPipe, etc.).

   pip install tensorflow opencv-python mediapipe numpy matplotlib

2. Run the Application

   python app.py

Training (Optional)

If you want to retrain the model on new data:

1. Run collectdata.py to capture new sequences.
2. Run trainmodel.py to train the LSTM model.