CAPTOR proposes text-encoder-free visual grounding via caption co-occurrence counting
A new arXiv paper, CAPTOR, grounds text queries in images without a text encoder, scoring terms against patches by counting caption co-occurrence over the visual words of a frozen encoder.
Read the full story at arXiv cs.CV ↗
Timeline · 1 report
- 2026-11-05 05:00 · arXiv cs.CV
CAPTOR: text-encoder-free grounding by counting caption co-occurrence over the visual words of a frozen encoder