CVPR 2026 Accepted Paper Copied Your Work Without Credit—Here's What to Do

A researcher discovered that a paper accepted to CVPR 2026 reproduced substantial technical content from their arXiv paper posted in June 2025—five months before the conference deadline. The CVPR paper reworded concepts and reframed terminology, but copied equations verbatim without changing notation or citing the original work. Pipeline diagrams and figures also show suspicious stylistic similarities. Despite attempts to contact the authors, the situation remains unresolved. This raises critical questions about peer review oversight and what recourse exists when major conferences accept potentially plagiarized work. The post discusses detection methods, documentation strategies, and escalation paths through conference organizers and institutional channels.