gradio-ASL-fingerspelling-recognition

This code is for a kaggle compeition for American Sign Language fingerspelling recognition. Refer to the competition page for more information about the competition.

This repository contains the code to perform american sign language fingerspelling recognition on input from your Webcam or by uploading a video(with a Gradio interface).

Table of Contents

• Prerequisites
• Setup
• Running the App
• Usage

Prerequisites

• Docker (recommended)
• Python version 3.8.17

Setup

1. Clone the repository

   git clone https://github.com/SamratThapa120/gradio-ASL-fingerspelling-recognition.git
   

2. Enter the directory

   cd gradio-ASL-fingerspelling-recognition
   

3. If you're using Docker, use the following commands:

   docker pull python:3.8.17
   docker run -it --rm -v "$(pwd)":/app -w /app python:3.8.17 bash
   

   This will setup a Docker container with Python 3.8.17 and map your current directory into the Docker container.

4. Install the requirements

   pip install -r requirements.txt
   

5. Place your model.tflite and inference_args.json under the weights directory

Running the App

To start the Gradio application, use the following command:

python3 app.py

This command will start a local server that hosts the application. You can then access the app by visiting http://localhost:7860/ in your web browser.

UI Preview