Counterfactual Co-occurring Learning for Bias Mitigation in Weakly-supervised Object Localization
Not on the map. Why?
Abstract (arXiv)
Contemporary weakly-supervised object localization (WSOL) methods have primarily focused on addressing the challenge of localizing the most discriminative region while largely overlooking the relatively less explored issue of biased activation -- incorrectly spotlighting co-occurring background with the foreground feature. In this paper, we conduct a thorough causal analysis to investigate the origins of biased activation. Based on our analysis, we attribute this phenomenon to the presence of co-occurring background confounders. Building upon this profound insight, we introduce a pioneering paradigm known as Counterfactual Co-occurring Learning (CCL), meticulously engendering counterfactual representations by adeptly disentangling the foreground from the co-occurring background elements. Furthermore, we propose an innovative network architecture known as Counterfactual-CAM. This architecture seamlessly incorporates a perturbation mechanism for counterfactual representations into the vanilla CAM-based model. By training the WSOL model with these perturbed representations, we guide the model to prioritize the consistent foreground content while concurrently reducing the influence of distracting co-occurring backgrounds. To the best of our knowledge, this study represents the initial exploration of this research direction. Our extensive experiments conducted across multiple benchmarks validate the effectiveness of the proposed Counterfactual-CAM in mitigating biased activation.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 53 of its 92 references that are in the map. First posted on arXiv 24 May 2023.
Builds on
53 of 92 references.
References · 53
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2014Adam: A Method for Stochastic Optimizationcited by 37,240
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
- 2014Microsoft COCO: Common Objects in Contextcited by 17,608
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
- 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
- 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
- 2019CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Featurescited by 2,884
- 2015Learning Deep Features for Discriminative Localizationcited by 2,436
- 2013Rich feature hierarchies for accurate object detection and semantic segmentationcited by 2,162
- 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,432
38 more
- 2018Object Detection with Deep Learning: A Reviewcited by 528
- 2018Deep Learning for Generic Object Detection: A Surveycited by 392
- 2020Unbiased Scene Graph Generation from Biased Trainingcited by 356
- 2020Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal Effectcited by 256
- 2020Causal Intervention for Weakly-Supervised Semantic Segmentationcited by 229
- 2017Hide-and-Seek: Forcing a Network to be Meticulous for Weakly-supervised Object and Action Localizationcited by 207
- 2015Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentationcited by 189
- 2018Adversarial Complementary Learning for Weakly Supervised Object Localizationcited by 166
- 2020Counterfactual Samples Synthesizing for Robust Visual Question Answeringcited by 153
- 2019Attention-based Dropout Layer for Weakly Supervised Object Localizationcited by 125
- 2021Counterfactual Attention Learning for Fine-Grained Visual Categorization and Re-identificationcited by 114
- 2020Interventional Few-Shot Learningcited by 112
- 2020Evaluating Weakly Supervised Object Localization Methods Rightcited by 105
- 2018Self-produced Guidance for Weakly-supervised Object Localizationcited by 72
- 2021A Prototype-Oriented Framework for Unsupervised Domain Adaptationcited by 62
- 2022The Devil is in the Labels: Noisy Label Correction for Robust Scene Graph Generationcited by 60
- 2020Inter-Image Communication for Weakly Supervised Localizationcited by 52
- 2016Weakly Supervised Cascaded Convolutional Networkscited by 47
- 2020Rethinking the Route Towards Weakly Supervised Object Localizationcited by 46
- 2021Unveiling the Potential of Structure Preserving for Weakly Supervised Object Localizationcited by 46
- 2021Causal Inference Under Unmeasured Confounding With Negative Controls: A Minimax Learning Approachcited by 44
- 2020Geometry Constrained Weakly Supervised Object Localizationcited by 36
- 2021Shallow Feature Matters for Weakly Supervised Object Localizationcited by 34
- 2017Exploiting saliency for object segmentation from image level labelscited by 33
- 2021Boosting Entity-aware Image Captioning with Multi-modal Knowledge Graphcited by 31
- 2021Online Refinement of Low-level Feature Based Activation Map for Weakly Supervised Object Localizationcited by 30
- 2018Macro-Micro Adversarial Network for Human Parsingcited by 28
- 2022Bridging the Gap between Classification and Localization for Weakly Supervised Object Localizationcited by 25
- 2019Combinational Class Activation Maps for Weakly Supervised Object Localizationcited by 24
- 2022Prompt-based Learning for Unpaired Image Captioningcited by 23
- 2018TS2C: Tight Box Mining with Surrounding Segmentation Context for Weakly Supervised Object Detectioncited by 22
- 2021Background Activation Suppression for Weakly Supervised Object Localizationcited by 22
- 2017Two-Phase Learning for Weakly Supervised Object Localizationcited by 21
- 2022CREAM: Weakly Supervised Object Localization via Class RE-Activation Mappingcited by 21
- 2021Normalization Matters in Weakly Supervised Object Localizationcited by 19
- 2022Active Learning for Point Cloud Semantic Segmentation via Spatial-Structural Diversity Reasoningcited by 16
- 2021Improving Weakly-supervised Object Localization via Causal Interventioncited by 14
- 2022Label Semantic Knowledge Distillation for Unbiased Scene Graph Generationcited by 14
50 most cited of 53 All 53 (JSON)
Built on by
Citing papers per year: 2026: 3.
3 papers.