MARIO: A Mixed Annotation Framework For Polyp Segmentation
This addresses the challenge of high labeling costs and underutilized datasets in medical imaging for polyp segmentation, though it is incremental in leveraging multiple annotation types.
The paper tackles the problem of limited and costly annotations in polyp segmentation by introducing MARIO, a mixed supervision model that uses five annotation types, and it outperforms existing methods across five benchmark datasets.
Existing polyp segmentation models are limited by high labeling costs and the small size of datasets. Additionally, vast polyp datasets remain underutilized because these models typically rely on a single type of annotation. To address this dilemma, we introduce MARIO, a mixed supervision model designed to accommodate various annotation types, significantly expanding the range of usable data. MARIO learns from underutilized datasets by incorporating five forms of supervision: pixel-level, box-level, polygon-level, scribblelevel, and point-level. Each form of supervision is associated with a tailored loss that effectively leverages the supervision labels while minimizing the noise. This allows MARIO to move beyond the constraints of relying on a single annotation type. Furthermore, MARIO primarily utilizes dataset with weak and cheap annotations, reducing the dependence on large-scale, fully annotated ones. Experimental results across five benchmark datasets demonstrate that MARIO consistently outperforms existing methods, highlighting its efficacy in balancing trade-offs between different forms of supervision and maximizing polyp segmentation performance