/http_scanit

A simple tensorflowjs based image recognition webapp

Primary LanguageJavaScript

http_scanit

A simple tensorflowjs based image recognition webap

This project is a web application with no backend requirement that utilizes the TensorFlow.js library to recognize images shown on the user’s webcam and categorize them into one of four different images. If a good enough fit (>95%) is found, the webapp redirects the user to the HTML page with the image’s name.

Installation

  • Clone this repository.

Usage

  • localhost the file through python -m http.server
  • Allow access to your webcam.
  • Show an image to your webcam and wait for the application to recognize it.
    • Images can be located at the try_using_these_images folder.
  • See that you have been redirected

Methods

- Tensorflow model trained with [teachable_machine](https://teachablemachine.withgoogle.com/train) and around 8 images per class.

Try the webapp by visiting [dibbywang.github.io/http_scanit](https://dibbywang.github.io/http_scanit)