All tools

aiquick/segment-anything

image-to-masks

SAM-2 automatic mask generation. It segments the whole image without being told what to look for, and this playground draws the resulting objects as boxes over your image, largest first, with each mask's area and confidence. The per-pixel masks stay on the server — one image's worth of run-length encoding runs to megabytes.

Inference/gpu/segment

Input

Image *Up to 6 MB. Takes about half a minute.
Maximum objects1–100. The largest masks are kept when the model finds more.

Result

Idle
Fill in the input and run the tool to see results here.