facebookresearch / facebookresearch/segment-anything

Using SAM image encoder with different image size

Open
#374 9 comments 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
54.9k
Forks
6.4k
PR merge metrics
No merged PRs in 30d

Description

Thanks for the great work.

@likethesky @Celebio @colesbury @pdollar
@minqi

I am gonna try SAM image encoder with image size 400*400. But the model weights are for size 1024*1024 and I get mismatch error. Is there any way to use SAM with different image sizes.

I think It would be great for the vision community to be able to use SAM encoder for feature extraction.

Contributor guide

Open the contributing guide

Research direction

Start by tracing how the SAM image encoder validates input dimensions and loads the 1024×1024 checkpoint. Investigate whether 400×400 inputs can be supported without incompatible weights, then confirm the intended behavior for feature extraction and validate it with a representative 400×400 input.

Written by the indexing model from the issue text.

Assessment

Domain
computer-vision, machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.