✋ Sign Language Recognition (SLR) using Computer Vision and Deep Learning

📌 About the Project

This is a BTech Final Year Project focused on Sign Language Recognition.
The project uses Python, OpenCV, and Deep Learning to capture, process, and classify hand gestures.

Key functionality includes:

• Hand detection and histogram creation
• Gesture image collection and preprocessing
• Training and evaluating a CNN model for gesture recognition
• Real-time display of recognized gestures

---

🎯 Objectives

• Capture hand gestures using a webcam
• Preprocess images and extract features using HSV color histograms
• Train a Convolutional Neural Network (CNN) for gesture classification
• Recognize and display gestures in real time

---

🛠️ Technologies Used

• Python
• OpenCV
• NumPy
• TensorFlow / Keras (for CNN model)
• Pickle (for saving hand histograms)
• Jupyter Notebooks

---

⚙️ How the Project Works

1. Hand Histogram Setup:

   • Run set_hand_histogram.ipynb to capture your hand and generate a histogram.
   • Press C to capture histogram and S to save and exit.

2. Create Gestures:

   • Run Create_gesture.ipynb to capture gesture images for the dataset.

3. Preprocessing:

   • Use Rotate_images.ipynb and load_images.ipynb to augment and load images.

4. CNN Training:

   • Train the model using cnn_modle_train.ipynb.
   • The trained model is saved as cnn_model_keras2.h5.

5. Display Gestures:

   • Run display_gesturers.ipynb or final.ipynb for real-time gesture recognition.

---

▶️ How to Run

🔹 Prerequisites

pip install opencv-python numpy tensorflow keras

🔹Running Notebook Scripts

Open Jupyter Notebook in the gestures/ folder:

jupyter notebook

Run the notebooks in the order:

1. set_hand_histogram.ipynb
2. Create_gesture.ipynb
3. Rotate_images.ipynb
4. cnn_modle_train.ipynb
5. display_gesturers.ipynb or final.ipynb

Output

• hist → Hand histogram file
• cnn_model_keras2.h5 → Trained CNN model
• Gesture images → stored in train_images, val_images, test_images
• Labels → train_labels, val_labels, test_labels

---

🚀 Future Scope

• Enhance CNN model for higher accuracy
• Real-time sentence recognition using multiple gestures
• Integrate text-to-speech for sign language output
• Deploy as a web or mobile application

---

👩‍💻 Contributors

1. Riteeka Purnekar
2. Richa Patil
3. Nilesh Mahajan