Counterfactual Co-occurring Learning for Bias Mitigation in Weakly-supervised Object Localization

TL;DR

This paper introduces Counterfactual Co-occurring Learning (CCL) and Counterfactual-CAM, novel methods that use causal analysis and counterfactual representations to reduce background bias in weakly-supervised object localization, improving localization accuracy.

Contribution

The paper pioneers the use of causal analysis and counterfactual representations to address biased activation in WSOL, proposing a new network architecture and training paradigm.

Findings

Counterfactual-CAM effectively reduces background bias.

The method improves localization accuracy on multiple benchmarks.

Extensive experiments validate the approach's effectiveness.

Abstract

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.