CoolFace
Apppublic

Iris-Auto-ae/IRIS__fault-detection-demo-v0.1

sourceHugging Facemitupdated 3y agoView on Hugging Face
0likes
App README

IRIS fault detection Demo

  • —Run in a terminal with python app.py and open http://localhost:7861 in your browser.
  • —Alternatively, run with gradio app.py to start a dev server with hot reloading enabled.

Demo1

Demo2

Customize a fault detection model

  • —Add a few images of your own and see how the model performs

Note: This demo runs on CPU with num_workers=0, therefore is slow. For demonstration purposes, the model is trained only for 1 epoch. If you want to change run configuration, update the config.ini file accordingly. The model is not properly pre-trained, it is just built with pretrained weights. Segmentation is not yet implemented.

  • —Give some normal images. Prepare a directory with two folders: one folder 'normal' with normal images and one folder 'anomaly' with faulty images. (e.g.{DATASETS_DIR})

Note: For demo purposes this folder is not editable. If you want to use different input data, then add files in the 'data/' folder and edit the config.ini file accordingly.

Note: If the screen changes size and you can't go up, resize the page using Ctrl + '+' or with the magnifying lens.