The problem it solves
Repeated or unusually similar image regions can be difficult to notice during figure review. A detector can surface candidates, but reviewers still need the source images and visual context to assess them responsibly.
Who it is for
- Researchers checking figures before submission
- Editors and reviewers investigating possible duplicated image regions
- Research-integrity teams documenting an image review
How it works
- Load the source image or comparison set at the best available quality.
- Run the available similarity review and inspect candidate regions in the overlay.
- Compare every candidate with its source context and record a reviewer conclusion.
Output you can keep
A visual review workspace containing candidate matches, region overlays, and the evidence needed for a documented human assessment.
Limitations to check
- A similarity match is a review lead; it does not prove fabrication, misconduct, or intent.
- Compression, repeated structures, labels, and legitimate reuse can produce similar regions.
- Conclusions should include acquisition history, figure context, and expert review.