AI Image Classifier

Identify what your image shows with MobileViT-Small — a lightweight vision model trained on 1000 ImageNet classes. Top-5 predictions with confidence bars. Runs entirely in your browser.

Your images never leave your device. Classification happens locally.

Classify Your Image

Drop image here or click to select

PNG, JPG, WebP — any image

Runs entirely in your browser. Your data never leaves your device.

Our Test Data

Tested on ImageNet validation split — average classification time 0.8s on MacBook Air M2.

MetricValue
Model size (INT8)~18MB
Parameters~5M
Avg inference time0.8s / image
Model load (cold start)~6s (CDN) / <1s (cached)
ImageNet-1k Top-1 Accuracy76.7%
ImageNet-1k Top-5 Accuracy93.0%

Last verified: September 2026.

When Not To Use This Tool

Frequently Asked Questions

Are my images uploaded to a server?

No. This AI image classifier runs entirely in your browser using WebAssembly. Images are processed locally with MobileViT-Small and never leave your device.

How many classes can it recognize?

MobileViT-Small is trained on ImageNet-1k, so it can classify images into 1000 different categories including dogs, cats, cars, planes, flowers, insects, household objects, and many more.

What's the difference between a classifier and an object detector?

A classifier identifies what the image is (top-5 labels). An object detector finds and locates multiple objects with bounding boxes. Our AI Object Detector does the latter — use it when you need exact object positions.

Why is confidence sometimes low?

ImageNet-1k has very specific labels. A photo of a cute mixed-breed dog might score low on 'golden retriever' because it is not a pure example. Lower confidence = more ambiguous image or unfamiliar category.