Sign Speak : Empowering Deaf And Dumb Through Gesture Recognition

Project Title: Sign Speak – Real-time Sign Language Detection and Recognition Description: This project aims to bridge the communication gap for deaf and dumb individuals by recognizing Indian Sign Language (ISL) gestures in real-time using computer vision and deep learning. It leverages MediaPipe to extract hand landmarks, which are then used as input to a Long Short-Term Memory (LSTM) neural network model built with TensorFlow and Keras. The system detects hand gestures via a webcam feed, processes the keypoints, classifies them using the trained model, and translates them into corresponding text or actions. The project also features a Flask-based web application with user authentication and optional face recognition login. It demonstrates practical deployment of AI models, handling both accuracy and real-time performance. Key Technologies: Python, OpenCV, MediaPipe NumPy, TensorFlow, Keras LSTM Neural Network Flask Web Framework Face Recognition for Secure Login Highlights: Real-time detection of ISL gestures using webcam Landmark extraction via MediaPipe Hands Trained LSTM model for gesture classification Accuracy: ~99.4% (Tanh/Softmax activation, 1000 epochs) Lightweight and efficient compared to image-based CNN models Interactive web interface with face login