ClassifAi Live is an open-source Image Classifier IA Progressive Web App project, made for Arweave's open web hackathon.
This is a Progressive Web App, meaning that it has everything needed to be installed and used as a "normal app" on your device, if installed ClassifAi Live will work even if you're offline !
It can recognizes images and classify them into different categories.
Users create "models" with at least two categories and then "train" ClassifyAi to recognize these categories by feeding it images from the camera. Moreover logged in users with an Arweave's keyfile can share their models with the community or use a model made by the community.
ClassifAi Live works on both computers and mobile devices, it requires access to the camera.
- Use the Example model or create a new one with at least two categories
- Give access to the camera when asked
- Train ClassifAi Live by clicking on the categories buttons, this tells it that the current image belongs to that category
- The more ClassifAi Live is trained, the more its guess will be accurate !
- Go to the account page by clicking the "person" icon on the home page
- Drop your arweave's keyfile in the drop zone to log in
- You now have access to the community models and to the publish feature
The publishing process may take some time to be fully completed, the published model will only be usable after the process is fully done (from a few minutes to hours depending on the level of training of the model).
The main technologies and frameworks used in this project are :
- React / Redux
- Material-ui
- TensorFlow.js (A library from Google for Machine Learning in JavaScript)
- Arweave (A blockchain based network and permanent storage system)
Installing :
yarn
Running :
yarn start
Building :
yarn build