Sign Talk Web App

The Sign Talk Web App is a platform designed to facilitate communication with deaf individuals by translating their signs into understandable text. This web application employs a combination of HTML, CSS, JavaScript, and Python to create an interactive and user-friendly experience.

Features

• Sign-to-Text Translation: Easily understand what deaf individuals are communicating through sign language.
• User-Friendly Interface: Intuitive design for seamless navigation and a positive user experience.
• Multi-Platform Compatibility: Accessible through various web browsers for desktop and mobile devices.

Technologies Used

• Frontend:

  • HTML
  • CSS
  • JavaScript

• Backend:

  • Python

Tools and IDEs

• Visual Studio Code
• Kaggle (for data-related tasks)
• PyCharm
• Affinity Designer (for logo design)

For any inquiries or feedback please contact

-Nirmal Khakda at nirmal.201719@ncit.edu.np -Deepu Shah at deepu.201707@ncit.edu.np -puspha raj sangrula -skanda Nupana

Hand Gesture Recognition and Data Collection

Introduction

This repository contains Python code for two related projects: a hand gesture data collection script and a hand gesture recognition script. The data collection script captures images of hand gestures using a webcam, while the recognition script utilizes a pre-trained model to classify the hand gestures in real-time.

Features

• Data Collection Script:

  • Utilizes OpenCV and the cvzone library for hand detection.
  • Captures and saves images of hand gestures with predefined labels.
  • Allows users to customize the folder for saving images and set the number of images to capture.

• Recognition Script:

  • Combines hand detection using cvzone with a pre-trained gesture recognition model.
  • Recognizes and classifies hand gestures in real-time.
  • Displays bounding boxes and text annotations around the detected hands and their classifications.

Requirements

• Python 3.x
• OpenCV
• cvzone library
• TensorFlow (for the recognition script)

Setup

1. Clone the repository:

   git clone https://github.com/Nepal-College-of-Information-Technology/ai-mini-project-pands

2. Install the required libraries:

   pip install -r requirements.txt

3. Download the pre-trained gesture recognition model and labels from Model.zip and extract it into the Model folder.

Usage

Data Collection Script

1. Run the data collection script:

   python data_collection.py

2. Press the 's' key to save images. Images will be saved in the specified folder.

Recognition Script

1. Run the recognition script:

   python recognition.py

2. The script will display real-time hand gesture recognition with bounding boxes and text annotations.

Configuration

• Folder and Counter:

  • You can customize the folder for saving images in the data collection script by modifying the folder variable.
  • The counter variable keeps track of the number of captured images.

• Model:

  • The recognition script uses a pre-trained model located in the Model folder. Ensure the model and labels are correctly downloaded.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Hand detection powered by the cvzone library.
• Gesture recognition model trained using TensorFlow.

Author

[Nirmal Khadka]

Feel free to contribute, report issues, or suggest improvements. Happy coding!