Private image classification

Local Image Analyzer

Upload an image and inspect the five categories a compact vision model considers most likely. Processing happens inside your browser.

Local Image Analyzer

Image stays on this device

What the model sees

After the device check and one-time model download, the model returns its five most likely ImageNet categories. It recognizes broad objects and scenes; it does not identify people, verify authenticity, or understand every image.
The image is decoded and classified inside this browser. Best-AI.org does not upload or save it. The model host receives normal model-file requests, but never your selected image.

What this experiment can do

The model was trained on ImageNet categories. It works best on recognizable objects such as animals, vehicles, food, and common products. Results are predictions, not verified facts.

Model transparency

Model
onnx-community/mobilenetv4s-webnn
Revision
4d0797f67ae5973e65432967711289f4f2f47190
Runtime
WASM in a Web Worker
License
Apache-2.0
View model source