This model was trained within Edge Impulse using the "cookies dataset three". It is configured with a threshold of 5.0, meaning that any prediction above this value will classify the image as an anomaly. When an anomaly is detected in the live feed, it is highlighted using the "jet" colormap, which shows the model's confidence levels across different parts of the image.

This is a compressed version of the Edge Impulse mobile-client project. It doesn't need an API KEY to work. After training, simply provide the exported WebAssembly .zip file and set the threshold in the 'config.yaml' file based on the Edge Impulse logs.