Research map / Detection and segmentation / Weakly supervised detection and segmentation
Adversarial Complementary Learning for Weakly Supervised Object Localization
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 two jointly trained classifiers, with Classifier A erasing its discriminative feature regions to guide Classifier B | ACoL · ILSVRC 2016 val | Top-1 localization error: 56.40 -> 54.17 | Helped |
| Use two jointly trained classifiers, with Classifier A erasing its discriminative feature regions to guide Classifier B | ACoL · ILSVRC 2016 val | Top-1 localization error: 54.53 -> 53.28 | Helped |
| Use two jointly trained classifiers, with Classifier A erasing its discriminative feature regions to guide Classifier B | ACoL · CUB-200-2011 test | Top-1 localization error: 59.00 -> 54.08 | Helped |
| Add the third classifier, guided by the fused maps from the first two (ablation) | ACoL · ILSVRC 2016 val | localization performance | No clear effect |
| Set the erasing threshold to 0.6 rather than the other tested values (ablation) | ACoL · ILSVRC 2016 val | Top-1 localization error: 62.62; 54.17; 54.55; 56.61; 55.72 | Mixed |
| Set the erasing threshold to 0.6 rather than the other tested values (ablation) | ACoL · ILSVRC 2016 val | Top-5 localization error: 52.03; 40.57; 42.53; 45.45; 44.42 | Helped |
| Vary the erasing threshold (ablation) | ACoL · CUB-200-2011 test | Top-1 localization error: 58.34; 54.15; 54.08; 55.78; 55.22 | Mixed |
| Vary the erasing threshold (ablation) | ACoL · CUB-200-2011 test | Top-5 localization error: 48.11; 42.79; 43.49; 45.17; 45.76 | Mixed |
| Use only predictions with high scores for Top-5 localization (ablation) | VGGnet-ACoL · CUB-200-2011 test | Top-5 localization error: 43.49 -> 39.05 | Helped |
| Use DPN ensemble classification results for localization | VGGnet-ACoL · ILSVRC 2016 val | Top-1 localization error: 49.82/26.22; 49.26/24.90; 48.96/24.39; 46.30/17.70; 46.16/17.42; 46.06/17.08; 45.14/15.47 | Helped |
Closest papers by meaning
- Erasing Integrated Learning: A Simple Yet Effective Approach for Weakly Supervised Object Localization (CVPR 2020 · similarity 0.92)
- Bridging the Gap Between Classification and Localization for Weakly Supervised Object Localization (CVPR 2022 · similarity 0.88)
- Unveiling the Potential of Structure Preserving for Weakly Supervised Object Localization (CVPR 2021 · similarity 0.87)
- Rethinking the Route Towards Weakly Supervised Object Localization (CVPR 2020 · similarity 0.86)
- Weakly Supervised Complementary Parts Models for Fine-Grained Image Classification From the Bottom Up (CVPR 2019 · similarity 0.86)
- LCTR: On Awakening the Local Continuity of Transformer for Weakly Supervised Object Localization (AAAI 2022 · 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.