darksuei / imagevision

:octocat:💻 An image recognition REST API created using theTensorflow mobile-net computer vision model and Fastify(_dev)

Image Vision - API for Image Recognition

Image vision is a restful API that utilizes the tensorflow mobile-net computer vision model to identify objects and patterns in images. Make requests to this API with image(s) and get a response with the identified objects with their respective confidence scores [in plain terms, how confident the model is of success].

Checkout the fastify-dev branch to see the implementation with the fastify framework.

Wanna try it out? >> https://image-vision-api.onrender.com/ <<

Minimalistic sample usage of the API >> HERE <<

Response caching and response time enhancement coming soon!

• Image recognition using TensorFlow.js and MobileNet model.
• Confidence threshold customization for object detection result fine-tuning.

Installation

1. Clone the repository:

git clone https://github.com/darksuei/imagevision.git

cd imagevision

2. Install the dependencies:

npm install

3. Start the server in development mode:

npm run dev

Usage

1. Grab an API KEY, make a GET request to https://image-vision-api.onrender.com/api/auth for specific details.

2. Make API requests:

Send a POST request to the /api/image-recognition endpoint with an image in the request body. You can also customize the API by setting the confidence threshold in the request body.

Example:

curl -X POST -F "image=@path/to/your/image.jpg" -F "confidenceThreshold=0.5" -F https://image-vision-api.onrender.com/api/image-recognition

Available Endpoints

• GET * : Welcome message/ Health Check 👏.

• GET /api/auth : Get more information on how to acquire an API KEY.

• POST /api/auth : Grab your API KEY by making a request with your email and preferred password.

• GET /api/image-recognition : Get more information on how to use the image recognition endpoint.

• POST /api/image-recognition : Perform image recognition on the provided image with optional customization parameters.

  • Request body:
    • image: The image file to be recognized.
    • confidenceThreshold (optional): The confidence threshold for object detection, between 0 and 1 (default: 0.1).

• GET /api/save : Get more information on how to use this endpoint to temporarily store images and files.

• POST /api/save : Temporarily save your files and images.

• This readme would be updated with the latest features!!

Contributing

Contributions from the community are welcome! If you find any bugs or have suggestions for improvements, please feel free to open an issue or submit a pull request.

License

This project is licensed under the MIT license.

imagevision's Issues

Strict Error handling

Request Rate Limiting

Image Enhancement [Feature]

Add Dropbox for optional image storage

v2.0

• Functional tests for all features
• Temporary Image storage
• Response Caching
• Request rate limiting per user
• Improved error handling

Complete demo

Response caching for imporved performance

Functional Tests

Usage Documentation