An AI-powered system designed to bridge communication gaps by recognizing and translating Sign Language gestures into text in real-time using Computer Vision and Deep Learning.
This project implements an end-to-end Machine Learning pipeline—from training a custom model to testing its metrics and running real-time predictions. It captures hand gestures via a camera feed, processes the visual data, and instantly outputs the corresponding alphanumeric text.
- Language: Python
- Libraries & Frameworks: OpenCV, MediaPipe, NumPy, TensorFlow/Keras
The core functionality is split into three main scripts:
- Purpose: Handles the model training pipeline.
- Details: Loads the processed gesture datasets, defines the neural network architecture, handles data augmentation, tunes hyperparameters, and saves the trained weights/model file upon completion.
- Purpose: Evaluates the model's performance on unseen data.
- Details: Computes crucial metrics such as validation accuracy, loss curves, confusion matrices, and precision/recall to ensure the system is highly accurate before deployment.
- Purpose: The real-time inference and deployment script.
- Details: Integrates with OpenCV to access the webcam feed, captures live hand landmarks, feeds the processed frames into the trained model, and displays the translated text live on-screen.