Semantic-Constraint Matching Transformer for Weakly Supervised Object Localization

TL;DR

This paper introduces a Semantic-Constraint Matching Network using a transformer and a local patch shuffle strategy to improve weakly supervised object localization, achieving state-of-the-art results by addressing divergent activation issues.

Contribution

The paper proposes a novel transformer-based network with a semantic-constraint matching module and a local patch shuffle strategy to enhance object localization accuracy in weakly supervised settings.

Findings

Achieves new state-of-the-art performance on CUB-200-2011 and ILSVRC datasets.

Outperforms previous methods by a large margin.

Effectively mitigates divergent activation in transformer-based WSOL.

Abstract

Weakly supervised object localization (WSOL) strives to learn to localize objects with only image-level supervision. Due to the local receptive fields generated by convolution operations, previous CNN-based methods suffer from partial activation issues, concentrating on the object's discriminative part instead of the entire entity scope. Benefiting from the capability of the self-attention mechanism to acquire long-range feature dependencies, Vision Transformer has been recently applied to alleviate the local activation drawbacks. However, since the transformer lacks the inductive localization bias that are inherent in CNNs, it may cause a divergent activation problem resulting in an uncertain distinction between foreground and background. In this work, we proposed a novel Semantic-Constraint Matching Network (SCMN) via a transformer to converge on the divergent activation. Specifically, we first propose a local patch shuffle strategy to construct the image pairs, disrupting local patches while guaranteeing global consistency. The paired images that contain the common object in spatial are then fed into the Siamese network encoder. We further design a semantic-constraint matching module, which aims to mine the co-object part by matching the coarse class activation maps (CAMs) extracted from the pair images, thus implicitly guiding and calibrating the transformer network to alleviate the divergent activation. Extensive experimental results conducted on two challenging benchmarks, including CUB-200-2011 and ILSVRC datasets show that our method can achieve the new state-of-the-art performance and outperform the previous method by a large margin.