Presents MMIO benchmark and RTVP method achieving state-of-the-art 42.2% AP in zero-shot industrial defect detection.
arXiv preprint arXiv:2306.13731 , year=
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
verdicts
UNVERDICTED 4roles
background 1polarities
background 1representative citing papers
Presents MMIOC-1M benchmark with 1M+ samples across 14 super-categories and RTVPNet with domain projection, sparse sampling, and bidirectional interaction, claiming SOTA on MMIOC-1M, LVIS, and COCO.
A2ONet improves robustness of liver surface landmark detection in laparoscopic surgery via illumination field compensation, frequency-orientation selective filtering, and alternating seg-curve optimization, with reported gains on three datasets.
The paper surveys data-centric strategies for foundation models in computational healthcare and supplies a curated list of related models and datasets.
citing papers explorer
No citing papers match the current filters.