Research map / Detection and segmentation / Weakly supervised detection and segmentation
Min-Entropy Latent Model for Weakly Supervised Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use MELM instead of OICR | VGGF weakly supervised detector · PASCAL VOC 2007 test | mAP: 38.4 vs. 37.9 | Helped |
| Use MELM instead of WeakRPN | VGG16 weakly supervised detector · PASCAL VOC 2007 test | mAP: 47.3% vs. 45.3% | Helped |
| Use MELM instead of WCCN | VGG16 detector · PASCAL VOC 2012 | mAP: 42.4% vs. 37.9% | Helped |
| Use MELM instead of WCCN | weakly supervised detector · ILSVRC 2013 | mAP: 13.4% vs. 9.8% | Helped |
| Add proposal clique partition (ablation) | MELM-base · PASCAL VOC 2007 test | mAP: 31.5% to 33.9%; 2.4% | Helped |
| Use the object clique discovery branch (ablation) | MELM-base · PASCAL VOC 2007 test | mAP: 31.5% to 33.6%; 2.1% | Helped |
| Use the object localization branch (ablation) | MELM-base · PASCAL VOC 2007 test | mAP: 31.5% to 36.0%; 5.5% | Helped |
| Add recurrent learning (ablation) | MELM-L · PASCAL VOC 2007 test | mAP: 36.0% to 38.4% | Helped |
| Add a second object localization branch with accumulation (ablation) | MELM-L-RL · PASCAL VOC 2007 test | mAP: 42.6% to 46.4% (+3.8%) | Helped |
| Increase accumulated localization branches from two to three (ablation) | MELM with accumulated recurrent learning · PASCAL VOC 2007 test | mAP: 46.4% to 47.3% (+0.9%) | Helped |
| Increase accumulated localization branches to four (ablation) | MELM with accumulated recurrent learning · PASCAL VOC 2007 test | mAP | No clear effect |
| Set clique overlap threshold τ to 0.7 rather than other tested values (ablation) | MELM · PASCAL VOC 2007 val | mAP: 32.6, 34.3, 34.4, 35.3, 33.5, 34.4 | Mixed |
Closest papers by meaning
- Dissimilarity Coefficient Based Weakly Supervised Object Detection (CVPR 2019 · similarity 0.87)
- Instance-Aware, Context-Focused, and Memory-Efficient Weakly Supervised Object Detection (CVPR 2020 · similarity 0.86)
- W2F: A Weakly-Supervised to Fully-Supervised Framework for Object Detection (CVPR 2018 · similarity 0.86)
- C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection (CVPR 2019 · similarity 0.86)
- SLV: Spatial Likelihood Voting for Weakly Supervised Object Detection (CVPR 2020 · similarity 0.85)
- Relational Matching for Weakly Semi-Supervised Oriented Object Detection (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.