WeakPolyp: You Only Look Bounding Box for Polyp Segmentation

TL;DR

WeakPolyp introduces a cost-effective weakly supervised polyp segmentation method using only bounding box annotations, employing novel transformations and loss functions to achieve performance comparable to fully supervised models.

Contribution

The paper proposes a novel weakly supervised segmentation approach that leverages bounding box annotations with new M2B and SC modules, reducing labeling costs and maintaining high accuracy.

Findings

Achieves comparable performance to fully supervised models.

Uses only bounding box annotations during training.

Modules do not add inference computational cost.

Abstract

Limited by expensive pixel-level labels, polyp segmentation models are plagued by data shortage and suffer from impaired generalization. In contrast, polyp bounding box annotations are much cheaper and more accessible. Thus, to reduce labeling cost, we propose to learn a weakly supervised polyp segmentation model (i.e., WeakPolyp) completely based on bounding box annotations. However, coarse bounding boxes contain too much noise. To avoid interference, we introduce the mask-to-box (M2B) transformation. By supervising the outer box mask of the prediction instead of the prediction itself, M2B greatly mitigates the mismatch between the coarse label and the precise prediction. But, M2B only provides sparse supervision, leading to non-unique predictions. Therefore, we further propose a scale consistency (SC) loss for dense supervision. By explicitly aligning predictions across the same image at different scales, the SC loss largely reduces the variation of predictions. Note that our WeakPolyp is a plug-and-play model, which can be easily ported to other appealing backbones. Besides, the proposed modules are only used during training, bringing no computation cost to inference. Extensive experiments demonstrate the effectiveness of our proposed WeakPolyp, which surprisingly achieves a comparable performance with a fully supervised model, requiring no mask annotations at all.