Photo Object Splitter
Upload a photo, pick which detected objects you want, and download each as its own image.
Uploading a photo downloads the object-detection AI model (18 MB) the first time in this browser.
Frequently asked questions
- What kinds of objects can it detect?
- The detector recognizes 80 common object categories (people, animals, vehicles, furniture, food, and everyday items) using the COCO-SSD model. It won't recognize objects outside that list, and may miss small, overlapping, or partially hidden objects.
- Does my photo get uploaded anywhere?
- No. Detection, cropping, and background removal all run in your browser using on-device AI models. Your photo never leaves your device.
- Why does the first detection take a few seconds?
- The detection model is an 18 MB download the first time you use it in a browser (shown above the upload button) and is cached afterward — later photos in the same session, or a later visit, detect faster. Turning on "Remove background from exports" downloads a separate, larger AI model the first time you use it, sized the same way.
- What does the padding slider do?
- It adds a margin around each detected object before cropping, so the exported image isn't cropped exactly to the edge of the object. It's clamped to the photo's own edges, so it never adds blank space beyond the original photo.