Object Detection with Your Camera
Open your phone or computer camera and let AI mark and count the objects it sees. The image never leaves your device.
Your browser cannot run AI models in the page. A recent version of Chrome, Edge, Firefox or Safari is required.
Your browser will ask for camera permission. It only applies while this page is open; stopping turns the camera off.
Your image is not uploaded. The model is downloaded to your browser and detection runs on your device.
The image never leaves your device
The detection model is downloaded to your browser with the page and every frame is processed on your device. The camera feed is never sent to us at any point; no recordings, frames or results are kept on the server.
What it recognises
80 everyday object types such as people, bicycles, cars, buses, cats, dogs, chairs, tables, laptops, phones, bottles, cups and books. Each object gets a box with its name and how confident the model is; the list next to it shows how many of each type there are.
Install it as an app
VisionSpace is a web app: use "Add to Home screen" or "Install app" in your browser menu to put it on your phone. Once opened, it also works without an internet connection.
VisionSpace
- Object detection with your camera — mark and count objects in the live view
- Object detection in photos — find the objects in an image
- Saved results — what you saved, in this browser only
Limitations
The model recognises the 80 everyday object types of the COCO dataset: people, vehicles, animals, kitchenware, furniture, electronics and so on. Anything outside that list is not recognised and may be mistaken for the closest type. It does not tell brands, models or specific products apart. People are marked only as "person": there is no face recognition, identification or number-plate reading.
On first use the model and the runtime are downloaded to your browser (about 23–25 MB). After that they come from the cache and work without an internet connection. Speed depends on your device: several frames per second in browsers that can use the graphics card, slower in those that cannot.
Open source components
Object detection model YOLOX-Tiny (Megvii, Apache-2.0); in-browser runtime ONNX Runtime Web (Microsoft, MIT). The licence texts are kept next to the files on the server.