Find and count objects in a photo or on your webcam
80 everyday things — people, cars, bikes, dogs, cups, laptops. The detector runs on this device: nothing is uploaded and there is nothing to sign in to.
Drop photos here, paste one, or
JPEG, PNG, WebP or HEIC your browser can open. Several at once are counted together.
Live detection from your camera
Your browser asks first. Frames are read in this tab and never sent anywhere.
Loading the detector…
Count over time
The most of each object seen in any frame of each second, for the last two minutes.
All photos
How it works
The detector is YOLOX, trained on the 80 COCO object classes. It runs in this tab with onnxruntime-web — on your graphics card through WebGPU where the browser has it, otherwise on the processor. The model file is downloaded once from SkillSafe’s model host, checked against its fingerprint and kept on this device for next time.
Counts over time keep the largest number of each object seen in any frame of each second, so a person who flickers out of one frame does not dent the line. Exports: a PNG of the picture with its boxes, a CSV of every box, a per-photo counts CSV and COCO-format JSON.