Research map / Detection and segmentation / CNN features
Gradient-Based Instance-Specific Visual Explanations for Object Specification and Object Discrimination
Chenyang Zhao, Janet H. Hsiao, Antoni B. Chan
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 |
|---|---|---|---|
| add Odam-Train to detector (ablation) | ODAM · MS COCO | energy PG(b) (%): 73.1 -> 79.6 | Improved |
| add Odam-Train to detector (ablation) | ODAM · CrowdHuman | compactness: 1.04 -> 0.91 | Improved |
| add Odam-Train to detector (ablation) | ODAM-NMS · CrowdHuman FCOS | JI: 78.5 -> 81.1 | Improved |
| add Odam-Train to detector (ablation) | classical NMS · CrowdHuman FCOS | JI: 78.4 -> 78.9 | No clear effect |
| add Odam-Train to detector (ablation) | ODAM-NMS · CrowdHuman FCOS | MR: 54.0 -> 44.5 | Improved |
| add Odam-Train to detector (ablation) | ODAM · user study object discrimination | user accuracy (%): 94.00 -> 94.67 | No clear effect |
| replace NMS with Odam-NMS | FCOS · CrowdHuman | JI: 78.4 -> 81.1 | Improved |
| replace NMS with Odam-NMS | Faster R-CNN · CrowdHuman | JI: 79.5 -> 80.5 | Improved |
| replace NMS with Soft-NMS | FCOS · CrowdHuman | MR: 45.5 -> 89.0 | Hurt |
| replace NMS with FeatureNMS | Faster R-CNN · CrowdHuman | AP: 86.9 -> 82.0 | Hurt |
| stricter correlation thresholds for Odam-NMS without Odam-Train (ablation) | ODAM-NMS · CrowdHuman FCOS | MR: 54.0 -> 45.5 | Improved |
| use gradient-based instance-specific heat map ODAM instead of D-RISE | D-RISE · MS COCO | PG(b) (%): 82.6 -> 91.9 | Improved |
Closest papers by meaning
- ODAM: Gradient-based Instance-Specific Visual Explanations for Object Detection (ICLR 2023 · similarity 0.98)
- Black-Box Explanation of Object Detectors via Saliency Maps (CVPR 2021 · similarity 0.89)
- FFAM: Feature Factorization Activation Map for Explanation of 3D Detectors (NeurIPS 2024 · similarity 0.87)
- Interpreting Object-level Foundation Models via Visual Precision Search (CVPR 2025 · similarity 0.86)
- Finer-CAM: Spotting the Difference Reveals Finer Details for Visual Explanation (CVPR 2025 · similarity 0.86)
- The Devil is in the Fine-Grained Details: Evaluating Open-Vocabulary Object Detectors for Fine-Grained Understanding (CVPR 2024 · similarity 0.85)
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.
