FreeAIAI Object Detector
Detect and count objects in an image
Boxes, labels and a count for everything the AI recognises, or only what you type.
Drop an image here, or .
Free AI that runs on your device. Nothing is uploaded. Only edit photos you own or have permission to edit; see our Acceptable Use Policy.
AI Object Detector: how to use it
- Drop in a photo.
- Leave Find on Everything it recognises, or type the object you are looking for.
- Press Find objects, then download the labelled image or the boxes as JSON.
Tips
- Typing one simple word (car, bottle, dog) finds more of that object than the general mode does.
- Crop a busy photo to the area you care about with the Image Cropper for a more accurate count.
AI Object Detector: what it does and when to use it
The AI Object Detector finds the things in a photo, draws a labelled box around each one and counts them: cars in a car park, people in a queue, apples on a tree, chairs in a room. Choose Everything it recognises for a full inventory, or type a word such as dog, red car or bottle and it marks only those.
It runs Florence-2, an open-source vision model, on your own device, so a photo of your home, office or warehouse is never uploaded. You get the image with boxes drawn on it, a table of how many of each object were found, and the exact box positions as a JSON file for a spreadsheet, a dataset or your own code.
Like every detector it is best at everyday objects of a reasonable size. It can miss things that are tiny, overlapping, partly hidden or unusual, and it may give one object two names, so count by eye when the number matters. For reading the text in a picture rather than its objects, use Image to Text (OCR); for a written description, the Alt Text Generator.
AI Object Detector: frequently asked questions
How do I count objects in a picture?
Drop the picture in and press Find objects. Every object the AI recognises gets a box and a label, and the table under the image counts them. To count one kind of thing, type it in the What to find box first, for example car or apple. Related tool: Image to Prompt.
How accurate is the count?
Good for everyday objects that are clearly visible. It can miss small, overlapping or partly hidden objects, and occasionally names one thing twice, so treat the number as a fast first count and check it by eye when it matters. Related tool: Alt Text Generator.
Can I export the boxes?
Yes. Boxes as JSON downloads every object's label and its position in pixels (x, y, width and height), ready for a spreadsheet, a dataset or your own code. More help: ImgVX AI help docs.
What objects can it recognise?
Hundreds of everyday things: people, vehicles, animals, furniture, food, kitchenware, electronics, plants and more. It is a general model, so it will not know a specific part number or plant species.