🧠 Sign Language Recognition

This project implements a deep learning model to recognize American Sign Language (ASL) hand gestures. The model is trained to classify alphabet signs (A–Z), numbers(0-9) from image data using a Convolutional Neural Network (CNN). It includes training, evaluation, and testing components, along with necessary preprocessing steps.

The application aims to enhance communication accessibility for the hearing- and speech-impaired by converting hand gestures into readable text or voice output using computer vision techniques.

---

📁 Project Structure

• Sign_Language.ipynb: Notebook for data preprocessing, model building, training, and saving the CNN model.
• Test_model.ipynb: Notebook for loading the trained model and testing it on new input images.
• requirements.txt: Lists all required Python packages to run the notebooks.

---

🚀 Features

• Classifies ASL alphabet signs (A–Z)
• CNN-based architecture with convolution, pooling, and dense layers
• Evaluation using accuracy metrics
• Easy testing with custom images
• Clean and modular notebooks

---

🛠️ Installation

1. Clone the repository:

git clone https://github.com/RafiAhamed07/Sign_Language_Classcification.git
cd sign-language-recognition

2. ⚙️ Environment Setup (with CUDA support)

⚠️ Note: This project is best run in a conda environment with proper CUDA support for GPU acceleration.

Create and activate a Conda environment:

conda create -n signlang python=3.9
conda activate signlang

Install TensorFlow with GPU support (CUDA):

conda install -c conda-forge cudatoolkit=11.2 cudnn=8.1.0
pip install tensorflow==2.10.1

Ensure that your system has a compatible NVIDIA GPU and the correct drivers installed. You can verify GPU access in Python using:

import tensorflow as tf
print("Num GPUs Available: ", len(tf.config.list_physical_devices('GPU')))

3. Install required packages:

pip install -r requirements.txt

---

🧪 Usage

1. Training the model:

   • Open Sign_Language.ipynb
   • Run cells step-by-step: data loading → preprocessing → model training → save model

2. Testing the model:

   • Open Test_model.ipynb
   • Load the saved model and test it on new images

---

🖼️ Dataset

You will need a dataset of ASL hand gesture images. For example, the Sign Language Gesture Images Dataset from Kaggle is a great option.

Make sure the dataset is structured as expected in the training notebook.

---

📊 Model Architecture

• Input Layer: Preprocessed image data
• Convolutional Layers: Feature extraction using filters
• MaxPooling Layers: Downsampling to reduce complexity
• Fully Connected Layers: Classification into 26 alphabet classes
• Softmax Output: Multi-class probability prediction

---

✅ Requirements

All dependencies are listed in requirements.txt. Key libraries include:

• TensorFlow
• Keras
• NumPy
• Matplotlib
• Scikit-learn
• Gradio (optional for UI)

Install them using:

pip install -r requirements.txt

---

✨ Acknowledgments

• Dataset from Kaggle
• Built with TensorFlow and Keras
• Inspired by inclusive tech and accessibility solutions