About

This demo shows where a vision model looks. Its main view is the Label-free Activation Map (LaFAM): a heatmap of the regions the model responds to most, computed without choosing a class, so it also works for models trained without labels (Karjauv & Albayrak, 2024; blog post).

The demo also reads the classifier region by region. A classifier divides a photo into a grid of regions, describes each region, and averages these descriptions into one prediction. Because averaging is the last step, its answer can be read for every region, and the average of these answers is exactly its prediction. This shows where it sees each class (CAM, class activation map) and lets it find several objects in one image (Karjauv, 2026).

Everything runs in your browser, and your images never leave your device.

How to use

Pick an example, upload an image or video, or use your camera. Pause a video to interact with it.

• Click a region to see what the model sees there.
• Click a class to see where the model finds it; click again or press × to clear.
• Modes: heatmaps, a class for every region (multi-instance classification), boxes around objects (object detection), and following an object through a video (object tracking).
• Models are downloaded when you select them. They run on your graphics card if the browser supports WebGPU, otherwise on the processor, which is slower.