Back to all skills

✂️

Documents

SAM Image Masking

Precise image segmentation — click a point, get a mask.

4.6rating

4,800 installs

mlops/models/segment-anything

Max required

About this skill

SAM Image Masking runs Meta's Segment Anything Model locally. Click a point on an image and get a pixel-perfect mask for that object — no labelling, no training. Perfect for isolating a subject before inpainting, composing an image, or cutting a product out for an e-commerce listing. Also supports box prompts and "segment everything" mode.

What it does

Point-and-click masking, no training needed
Box-prompt and full-image segmentation modes
Export masks as PNG, SVG, or JSON polygons
Runs on Metal or CUDA
Pairs with Stable Diffusion Lab for inpainting

Use cases

Isolate a product from its background for a store listing
Create a clean mask before an inpaint
Auto-segment every object in a scene for a dataset

Related skills

Document Q&A

Paste a PDF or text and ask questions — answers cite the exact passage.

29,500 installs

Long-form Summarizer

Turn long articles, papers, or reports into a usable summary at the length you want.

23,800 installs

Bulk Document Extraction

Extract structured fields from a folder of PDFs into a spreadsheet.

10,200 installs

YouTube Summarizer

Summarize any YouTube video into chapters, key points, and quotes.

18,900 installs