Sign Language Using Deep Learning

Overview

This contains my Undergraduate Final Year Project titled "Sign Language Using Deep Learning".

The project aims to recognize sign language gestures from images using deep learning techniques. The system was developed in MATLAB and utilizes Convolutional Neural Networks (CNN) and AlexNet to classify hand gestures and assist in communication between hearing-impaired individuals and others.

Objective

The primary objective of this project is to automatically identify sign language gestures from images and convert them into meaningful outputs using deep learning-based image classification techniques.

Technologies Used

• MATLAB R2018a
• Deep Learning Toolbox
• Image Processing Toolbox
• Convolutional Neural Networks (CNN)
• AlexNet
• Computer Vision Techniques

Methodology

The system follows the following workflow:

1. Input Image

• Capture or load sign language gesture images
• Display image for processing

2. Image Preprocessing

• Image resizing
• Noise removal
• Image enhancement
• Color conversion (RGB to grayscale when required)

3. Image Segmentation

• Region extraction
• Hand gesture isolation
• Background removal

4. Feature Extraction

• Automatic feature learning using CNN
• Deep feature extraction using AlexNet

5. Classification

• CNN-based gesture classification
• AlexNet-based transfer learning approach
• Prediction of sign language symbols

6. Output Generation

• Recognized gesture display
• Classification results
• Performance evaluation

Deep Learning Models

Convolutional Neural Network (CNN)

The CNN architecture includes:

• Input Layer
• Convolution Layer
• ReLU Layer
• Max Pooling Layer
• Batch Normalization Layer
• Fully Connected Layer
• Softmax Layer

AlexNet

AlexNet was explored as a transfer learning model for sign language recognition and image classification tasks. The pre-trained architecture was adapted for gesture recognition experiments.

Features

• Automated sign language recognition
• Deep learning-based image classification
• MATLAB implementation
• CNN and AlexNet comparison
• Gesture prediction and recognition
• Assistive communication application

Repository Contents

• Source Code
• Project Documentation
• Output Screenshots
• Project Images
• Experimental Results

Software Requirements

• MATLAB R2018a or later
• Deep Learning Toolbox
• Image Processing Toolbox

Results

The system successfully classified sign language gestures using deep learning techniques and demonstrated the effectiveness of CNN-based image recognition for assistive communication applications.

Learning Outcomes

Through this project, I gained practical experience in:

• Deep Learning
• Computer Vision
• Image Processing
• CNN Architecture Design
• Transfer Learning with AlexNet
• MATLAB Development
• Pattern Recognition

Academic Information

Project Type: Undergraduate Final Year Project

Domain: Deep Learning and Computer Vision

Development Platform: MATLAB