AI Object Detector
Spot and label people, animals, vehicles and everyday objects in your photos.
How to use: Choose a photo. The AI draws a labelled box around each object it recognises and lists what it found. Move the slider to show more or fewer objects.
Runs the open-source EfficientDet-Lite0 model (MediaPipe, Apache-2.0) on your device. It recognises 80 everyday object types. About 16 MB downloads once.
How to use the AI Object Detector
Object detection is the computer-vision task behind self-driving cars, photo search and security cameras: finding things in an image and drawing a box around each one. This tool lets you try it on your own photos for free, with the model running inside your browser.
Choose a photo and the detector loads the open-source EfficientDet-Lite0 model the first time (about 16 MB, then cached). It recognises the 80 everyday categories of the well-known COCO dataset, including people, bicycles, cars, buses, dogs, cats, birds, chairs, sofas, laptops, phones, bottles, cups, pizza and many more. Each detection gets a coloured box with its name and a confidence percentage, and a table counts how many of each object were found.
Use the confidence slider to hide uncertain guesses or reveal more of them. Download the labelled image to share or use in a presentation. It is a handy way to count items, to learn how AI vision works or to check what a model can and cannot see.
The model can miss small, unusual or partly hidden objects and sometimes confuses similar things. It does not know about objects outside its 80 categories, and it does not recognise specific people. Your photo is never uploaded.
Frequently asked questions
What objects can it detect?
The 80 COCO categories: people, common animals, vehicles, furniture, kitchen items, electronics, sports equipment and food such as pizza or bananas.
Does it recognise faces or identities?
No. It detects a “person” but never identifies who someone is.
Is my photo sent anywhere?
No. Detection runs on your device using a downloaded open-source model.