Research map / Detection and segmentation / Weakly supervised detection and segmentation
Inter-Image Communication for Weakly Supervised Localization
Xiaolin Zhang, Yunchao Wei, Yi Yang
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 Stochastic Consistency (SC) loss (ablation) | InceptionV3 plain · ILSVRC val | Gt-known localization error: 37.32 -> 31.63 | Improved |
| add SC loss (ablation) | InceptionV3 plain · ILSVRC val | Top-1 localization error: 53.71 -> 49.07 | Improved |
| add Global Consistency (GC) loss on top of SC (ablation) | InceptionV3 + SC · ILSVRC val | Gt-known localization error: 31.63 -> 31.04 | Improved |
| add GC loss on top of SC (ablation) | InceptionV3 + SC · ILSVRC val | Top-1 localization error: 49.07 -> 48.08 | Improved |
| update class centers with momentum rule instead of back-propagation (ablation) | GC memory · ILSVRC val | localization / Gt-known error: 49.03 -> 48.08; 32.09 -> 31.04 | Improved |
| enlarge input resolution from 224x224 to 320x320 (ablation) | I2C · ILSVRC val | Gt-known localization error: 31.50 -> 31.04 | Improved |
| enlarge input resolution from 224x224 to 320x320 (ablation) | I2C · CUB test | Gt-known localization error: 27.40 -> 22.99 | Improved |
| vary lambda1 over 40x range from 0.002 to 0.08 (ablation) | SC only · ILSVRC val | localization error: 49.07 vs 50.82 (best vs worst) | No clear effect |
| vary lambda2 from 0.0001 to 0.1 (ablation) | SC + GC · ILSVRC val | localization error: fluctuates within 0.96% | No clear effect |
| increase number of seeds K from 3 up to 100 (ablation) | SC · ILSVRC val | localization error: fluctuates within 0.46%; lowest at K=3; classification error lowest 28.40% at K=60 | Hurt |
| increase lambda1 (SC weight) (ablation) | SC only · ILSVRC val | classification vs Gt-known error: best 29.24 / 49.07 / 31.63; worst 32.13 / 50.82 / 34.44 | Mixed |
| use I2C with ResNet50 instead of ADL (ResNet50-SE) | I2C · ILSVRC val | Top-1 localization error: 51.47 -> 45.17 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- CutMix — CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features (ICCV 2019) · better on 2 of 2 shared results
- DANet — DANet: Divergent Activation for Weakly Supervised Object Localization (ICCV 2019) · better on 2 of 2 shared results
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
- Bridging the Gap Between Classification and Localization for Weakly Supervised Object Localization (CVPR 2022 · similarity 0.92)
- Pairwise Similarity Knowledge Transfer for Weakly Supervised Object Localization (ECCV 2020 · similarity 0.92)
- Learning to Localize Objects with Noisy Labeled Instances (AAAI 2019 · similarity 0.91)
- Spatial-Aware Token for Weakly Supervised Object Localization (ICCV 2023 · similarity 0.91)
- Rethinking the Route Towards Weakly Supervised Object Localization (CVPR 2020 · similarity 0.91)
- Coarse2Fine: Local Consistency Aware Re-prediction for Weakly Supervised Object Localization (AAAI 2023 · similarity 0.90)
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.
