Research map / Detection and segmentation / Weakly supervised detection and segmentation
Geometry Constrained Weakly Supervised Object Localization
Weizeng Lu, Xi Jia, Weicheng Xie, Linlin Shen, Yicong Zhou, Jinming Duan
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 background (negative-entropy) loss Lb to Lo (ablation) | GC-Net · CUB-200-2011 (VGG16, learning-driven rotated ellipse) | Top-1 LocErr (%): 59.22 -> 47.03 | Improved |
| add area loss La to Lo+Lb (ablation) | GC-Net · CUB-200-2011 (VGG16, learning-driven rotated ellipse) | Top-1 LocErr (%): 47.03 -> 41.15 | Improved |
| add area loss La alone to Lo (ablation) | GC-Net · CUB-200-2011 (VGG16, learning-driven rotated ellipse) | Top-1 LocErr (%): 59.22 -> 69.89 | Hurt |
| add Lb to Lo+La (ablation) | GC-Net · CUB-200-2011 (VGG16, learning-driven rotated ellipse) | CorLoc (%): 39.89 -> 74.89 | Improved |
| use rotated rectangle instead of rectangle (learning-driven) (ablation) | GC-Net · CUB-200-2011 (VGG16) | Top-1 LocErr (%): 44.35 -> 36.76 | Improved |
| use rotated ellipse instead of rectangle (learning-driven) (ablation) | GC-Net · CUB-200-2011 (VGG16) | Top-1 LocErr (%): 44.35 -> 41.15 | Improved |
| use rotated rectangle instead of rectangle (model-driven) (ablation) | GC-Net · CUB-200-2011 (VGG16) | Top-1 LocErr (%): 44.10 -> 44.25 | No clear effect |
| use learning-driven instead of model-driven generator (rotated rectangle) (ablation) | GC-Net · CUB-200-2011 (VGG16) | Top-1 LocErr (%): 44.25 -> 36.76 | Improved |
| use learning-driven instead of model-driven generator (rotated ellipse) (ablation) | GC-Net · CUB-200-2011 (VGG16) | Top-1 LocErr (%): 41.73 -> 41.15 | Improved |
| use GC-Net-Rect-VGG instead of DA-Net-VGG | GC-Net · CUB-200-2011 | CorLoc (%): 67.7 -> 81.1 | Improved |
| use GC-Net-Rect with Inception-V3 instead of DA-Net on ILSVRC2012 | GC-Net · ILSVRC2012 val | Top-1 LocErr (%): 52.47 -> 50.94 | No clear effect |
Closest papers by meaning
- Weakly Supervised Object Localization Using Size Estimates (ECCV 2016 · similarity 0.88)
- Pairwise Similarity Knowledge Transfer for Weakly Supervised Object Localization (ECCV 2020 · similarity 0.88)
- Rethinking the Route Towards Weakly Supervised Object Localization (CVPR 2020 · similarity 0.88)
- Evaluation for Weakly Supervised Object Localization: Protocol, Metrics, and Datasets (TPAMI 2022 · similarity 0.87)
- Inter-Image Communication for Weakly Supervised Localization (ECCV 2020 · similarity 0.87)
- Bridging the Gap Between Classification and Localization for Weakly Supervised Object Localization (CVPR 2022 · similarity 0.87)
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.
