Oluwamuyiwa01/Development-of-a-Deep-Learning-Based-Static-Sign-Language-Recognition-Model

This project presents a Convolutional Neural Network (CNN) developed for recognizing static sign language gestures. It was undertaken as my undergraduate thesis in Computer Science at the Federal University Lokoja, Nigeria. # Development-of-a-Deep-Learning-Based-Static-Sign-Language-Recognition-Model This project presents a Convolutional Neural Network (CNN) developed for recognizing static sign language gestures. It was undertaken as my undergraduate thesis in Computer Science at the Federal University Lokoja, Nigeria. ## Overview. This project, titled "Deep Learning-Based Static Sign Language Recognition Model," was designed to assist the deaf and hard-of-hearing in basic communication. The system captures and classifies static hand gestures using a Convolutional Neural Network (CNN) trained on a self-generated dataset. Key components include a data acquisition module using a webcam, preprocessing techniques for noise reduction and segmentation, feature extraction using CNNs, and model evaluation for performance tuning. The model serves as both an educational tool and a foundational communication aid. ## Data Collection and Preprocessing. * Type: Image dataset of static hand signs. * Number of classes: 12. * Source: Self-generated using OpenCV via webcam. * Collection Method: Captured multiple images per class using OpenCV in real-time with hand tracking and segmentation. * Preprocessing: Resizing, grayscale conversion, normalization, and one-hot encoding. * Dataset split: 80% training, 20% testing. ## Dataset. To access the dataset I used in this project, please request permission via this link: Download Dataset ## Model Architecture. * I built the model using TensorFlow/Keras * Input image size: 224 X 224 * Layers: Conv2D, MaxPooling2D, Dropout, Flatten, Dense * Loss: Categorical crossentropy; Optimizer: Adam * Trained for 20 epochs * Achieved high accuracy and generalization ## Performance Evaluation. ## Confusion matrix. ## Presentations of some of the results. ## Contribution. This project enhances basic communication for individuals with hearing impairments. Leveraging Convolutional Neural Networks (CNNs) and advanced preprocessing techniques, the system accurate …