Active Boundary Loss for Semantic Segmentation

TL;DR

This paper introduces an active boundary loss that enhances boundary accuracy in semantic segmentation by guiding predicted boundaries to align better with ground-truth boundaries during training, leading to improved segmentation quality.

Contribution

It presents a novel, model-agnostic active boundary loss that explicitly enforces boundary alignment in end-to-end training of segmentation networks.

Findings

Improves boundary F-score on segmentation datasets.

Enhances mean Intersection-over-Union metrics.

Effective on challenging image and video segmentation tasks.

Abstract

This paper proposes a novel active boundary loss for semantic segmentation. It can progressively encourage the alignment between predicted boundaries and ground-truth boundaries during end-to-end training, which is not explicitly enforced in commonly used cross-entropy loss. Based on the predicted boundaries detected from the segmentation results using current network parameters, we formulate the boundary alignment problem as a differentiable direction vector prediction problem to guide the movement of predicted boundaries in each iteration. Our loss is model-agnostic and can be plugged in to the training of segmentation networks to improve the boundary details. Experimental results show that training with the active boundary loss can effectively improve the boundary F-score and mean Intersection-over-Union on challenging image and video object segmentation datasets.