Research map / Detection and segmentation / Weakly supervised detection and segmentation
Object Discovery via Contrastive Learning for Weakly Supervised Object Detection
Jinhwan Seo, Wonho Bae, Danica J. Sutherland, Junhyug Noh, Daijin Kim
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 bounding box regression and dropblock to OICR (ablation) | OICR · VOC07 (SS proposals) | mAP: 45.0 -> 52.3 | Improved |
| add object discovery (ablation) | OICR+ · VOC07 ablation (OICR+ baseline) | mAP: 52.3 -> 56.1 | Improved |
| add WSCL (ablation) | OICR+ · VOC07 ablation (OICR+ baseline) | mAP: 52.3 -> 54.5 | Improved |
| add WSCL to object discovery (ablation) | OICR+ + OD · VOC07 ablation (OICR+ baseline) | mAP: 56.1 -> 58.7 | Improved |
| add random masking feature augmentation (ablation) | OICR+ + Ours · VOC07 feature augmentation ablation | mAP: 52.3 -> 55.1 | Improved |
| add Gaussian noise feature augmentation (ablation) | OICR+ + Ours · VOC07 feature augmentation ablation | mAP: 52.3 -> 54.8 | Improved |
| add IoU sampling to random masking (ablation) | OICR+ + Ours · VOC07 feature augmentation ablation | mAP: 55.1 -> 57.4 | Improved |
| add IoU sampling to Gaussian noise (ablation) | OICR+ + Ours · VOC07 feature augmentation ablation | mAP: 54.8 -> 57.4 | Improved |
| combine all three feature augmentations (ablation) | OICR+ + Ours · VOC07 feature augmentation ablation | mAP: 52.3 -> 58.7 | Improved |
| use MCG instead of SS proposals (ablation) | Ours · VOC07 | mAP: 56.1 -> 58.7 | Improved |
| use COB instead of MCG proposals (ablation) | Ours · VOC07 | mAP: 58.7 -> 61.8 | Improved |
| vary tau_nms, tau_drop, tau_IoU, lambda, epsilon around chosen values (ablation) | Ours · VOC07 | mAP: no more than 2.6% gap | No clear effect |
Closest papers by meaning
- Object Instance Mining for Weakly Supervised Object Detection (AAAI 2020 · similarity 0.92)
- Object-Aware Instance Labeling for Weakly Supervised Object Detection (ICCV 2019 · similarity 0.92)
- W2N: Switching from Weak Supervision to Noisy Supervision for Object Detection (ECCV 2022 · similarity 0.90)
- UWSOD: Toward Fully-Supervised-Level Capacity Weakly Supervised Object Detection (NeurIPS 2020 · similarity 0.90)
- Weakly Supervised Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.89)
- Weakly Supervised Few-Shot Object Detection with DETR (AAAI 2024 · similarity 0.88)
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.
