Research map / Detection and segmentation / CNN features
Explaining Object Detectors via Collective Contribution of Pixels
Toshinori Yamauchi, Hiroshi Kera, Kazuhiko Kawamoto
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace SSGrad-CAM++ with VX-CODE (r=1) | explanation method · MS-COCO DETR | overall (Ins-Del): .757 -> .851 | Improved |
| replace SSGrad-CAM++ with VX-CODE (r=1) | explanation method · PASCAL VOC Faster R-CNN | insertion AUC: .890 -> .827 | Hurt |
| increase patches per step r from 1 to 3 (ablation) | VX-CODE · MS-COCO DETR | overall: .851 -> .857 | Improved |
| increase patches per step r from 1 to 3 (ablation) | VX-CODE · PASCAL VOC Faster R-CNN | overall: .760 -> .787 | Improved |
| increase r from 1 to 3 (ablation) | VX-CODE · MS-COCO misclassification failures | insertion AUC: .707 -> .738 | Improved |
| add patch selection and step restriction (r=2) (ablation) | VX-CODE · MS-COCO DETR 50 instances | time per instance: 768.126s -> 96.253s | Improved |
| add patch selection and step restriction (r=2) (ablation) | VX-CODE · MS-COCO DETR 50 instances | overall: .860 -> .863 | No clear effect |
| add step restriction only (r=2) (ablation) | VX-CODE · MS-COCO DETR 50 instances | overall: .860 -> .866 | Improved |
| replace VPS with VX-CODE | explanation method · Grounding DINO MS-COCO | overall: .646 -> .751 | Improved |
| replace SSGrad-CAM++ with VX-CODE (r=1) | explanation method · MS-COCO mislocalization failures | deletion AUC: .120 -> .079 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ODAM — ODAM: Object Detection, Association, and Mapping Using Posed RGB Video (ICCV 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Identifying Important Group of Pixels using Interactions (CVPR 2024 · similarity 0.85)
- Solving Instance Detection from an Open-World Perspective (CVPR 2025 · similarity 0.85)
- Union-over-Intersections: Object Detection beyond Winner-Takes-All (ICLR 2025 · similarity 0.85)
- InstructDET: Diversifying Referring Object Detection with Generalized Instructions (ICLR 2024 · similarity 0.84)
- FFAM: Feature Factorization Activation Map for Explanation of 3D Detectors (NeurIPS 2024 · similarity 0.84)
- Interpreting Object-level Foundation Models via Visual Precision Search (CVPR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
