May 24, 2023 · We introduce a pioneering paradigm known as Counterfactual Co-occurring Learning (CCL), meticulously engendering counterfactual representations.

In this paper, we focus on an under-explored issue of biased activation in prior weakly-supervised object localization methods based on Class Activation ...

May 24, 2023 · Extensiveexperiments on several benchmarks demonstrate that Counterfactual-CAMsuccessfully mitigates the biased activation problem, achieving ...

Mar 9, 2024 · Weakly-supervised object localization (WSOL) focuses on localizing target objects in images using only image-level labels [20, 21, 22] .

The proposed method, dubbed CI-CAM, explores the causalities among image features, contexts, and categories to eliminate the biased object-context ...

Mitigating Biased Activation in Weakly-supervised Object Localization via Counterfactual Learning. Preprint. May 2023. Feifei Shao · Yawei ...

Previous weakly-supervised object localization (WSOL) methods aim to expand activation map discriminative areas to cover the whole objects, yet neglect two ...

Counterfactual Co-occurring Learning for Bias Mitigation in Weakly-supervised Object Localization · Knowledge-guided Causal Intervention for Weakly-supervised ...

Apr 16, 2024 · Previous weakly-supervised object localization (WSOL) methods aim to expand activation map discriminative areas to cover the whole objects, ...

Sep 2, 2024 · We introduce a semantic matching (SM) module to mitigate divergent activation. This module mines co-object knowledge from image pairs generated using a local ...