Research map / Detection and segmentation / Weakly supervised detection and segmentation
EHSOD: CAM-Guided End-to-End Hybrid-Supervised Object Detection with Cascade Refinement
Linpu Fang, Hang Xu, Zhili Liu, Sarah Parisot, Zhenguo Li
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 |
|---|---|---|---|
| Train jointly with weakly labeled images and limited fully labeled images | EHSOD · VOC07 test, 10% fully supervised | mAP50: 50.9 -> 55.3 | Improved |
| Train with 30% fully annotated data and image-level labels from the remaining data | EHSOD · MS-COCO 2017 val | AP | Improved |
| Add MID loss to cascade heads (ablation) | EHSOD · VOC07 test, 10% fully supervised | mAP50: 35.4 -> 44.5; +9.1 | Improved |
| Add proposal-confidence loss to the MID-loss head (ablation) | EHSOD · VOC07 test, 10% fully supervised | mAP50: 44.5 -> 50.8; +6.3 | Improved |
| Add the CAM branch to the hybrid-supervised detector (ablation) | EHSOD · VOC07 test, 10% fully supervised | mAP50: 50.8 -> 55.3; +4.5 | Improved |
| Increase cascade modules from 1 to 2 (ablation) | EHSOD · VOC07 test, 50% fully supervised | mAP50: 67.8 -> 70.3; +2.5 | Improved |
| Increase cascade modules from 2 to 3 (ablation) | EHSOD · VOC07 test, 50% fully supervised | mAP50: 70.3 -> 71.5; +3.7 | Improved |
| Increase cascade modules from 3 to 4 (ablation) | EHSOD · VOC07 test, 50% fully supervised | mAP50: 71.5 -> 71.6 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports EHSOD as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Many-shot from Low-shot: Learning to Annotate using Mixed Supervision for Object Detection (ECCV 2020 · 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
- W2N: Switching from Weak Supervision to Noisy Supervision for Object Detection (ECCV 2022 · similarity 0.89)
- Diverse Learner: Exploring Diverse Supervision for Semi-Supervised Object Detection (ECCV 2022 · similarity 0.89)
- UWSOD: Toward Fully-Supervised-Level Capacity Weakly Supervised Object Detection (NeurIPS 2020 · similarity 0.89)
- Label, Verify, Correct: A Simple Few Shot Object Detection Method (CVPR 2022 · similarity 0.88)
- Relational Matching for Weakly Semi-Supervised Oriented Object Detection (CVPR 2024 · similarity 0.87)
- LEOD: Label-Efficient Object Detection for Event Cameras (CVPR 2024 · similarity 0.86)
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.
