Click on an object to create an intelligent mask and download it with a transparent background.
PNG, JPG or WebP, up to 20 MB.
SlimSAM is a lightweight version of the Segment Anything approach. The image is encoded once, then you select the object to keep by placing points inside or outside the region.
Select the object directly on the photo.
Add points within the region you want to keep.
Excludes areas that the mask has mistakenly included.
Review the selection overlay before downloading.
Download only the selected object.
The image is not sent to UtiliTools Pro.
Select a picture.
Wait for SlimSAM to prepare the image.
Click on the object and add points if needed.
Download the cutout with transparency.
The tool uses Xenova/slimsam-77-uniform, a web-ready conversion for Transformers.js. The image is processed locally, and the model is reused during the session.
It is a smaller version of Segment Anything designed to preserve segmentation capabilities while using fewer resources.
Yes. Use positive points to include and negative points to exclude areas.
No. The model processes the photo directly in the browser.
A PNG with the selected object and transparency outside the mask.
Yes. The browser must download and prepare the model before calculating the image representation.
