Sign language recognition

This project focuses on recognizing sign language letters using image classification techniques. By training and comparing multiple convolutional neural network (CNN) models, it aims to accurately identify 24 hand signs. The use of data augmentation, performance evaluation, and visualization tools ensures robust model development and analysis.

• Builds an efficient sign language recognition system using image-based classification.
• Implements multiple CNN models and compares their performance on hand sign images.
• Incorporates data augmentation for better generalization and training diversity.
• Achieves high accuracy in recognizing 24 different hand signs.
• Demonstrates superior performance of ResNet50 over traditional CNN architectures.
• Uses confusion matrices and prediction plots to evaluate model correctness.
• Provides clear accuracy comparisons through visualized bar charts.
• Enables real-time prediction visualization for quick model validation.
• Establishes a baseline for future improvements in gesture and sign recognition tasks.
• Offers a simple, scalable method for recognizing hand signs from grayscale images.