Rethinking the Localization in Weakly Supervised Object Localization

TL;DR

This paper proposes a new approach for weakly supervised object localization that replaces single-class regression with a binary-class detector to localize multiple objects and introduces a weighted entropy loss to mitigate noise from pseudo bounding boxes, improving performance on standard datasets.

Contribution

It introduces a binary-class detector for multi-object localization and a weighted entropy loss to handle noisy pseudo bounding boxes in WSOL.

Findings

Effective in localizing multiple objects per image

Reduces noise impact from pseudo bounding boxes

Improves performance on CUB-200-2011 and ImageNet-1K datasets

Abstract

Weakly supervised object localization (WSOL) is one of the most popular and challenging tasks in computer vision. This task is to localize the objects in the images given only the image-level supervision. Recently, dividing WSOL into two parts (class-agnostic object localization and object classification) has become the state-of-the-art pipeline for this task. However, existing solutions under this pipeline usually suffer from the following drawbacks: 1) they are not flexible since they can only localize one object for each image due to the adopted single-class regression (SCR) for localization; 2) the generated pseudo bounding boxes may be noisy, but the negative impact of such noise is not well addressed. To remedy these drawbacks, we first propose to replace SCR with a binary-class detector (BCD) for localizing multiple objects, where the detector is trained by discriminating the foreground and background. Then we design a weighted entropy (WE) loss using the unlabeled data to reduce the negative impact of noisy bounding boxes. Extensive experiments on the popular CUB-200-2011 and ImageNet-1K datasets demonstrate the effectiveness of our method.