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.
@article{arxiv.2102.02696,
title = {Active Boundary Loss for Semantic Segmentation},
author = {Chi Wang and Yunke Zhang and Miaomiao Cui and Peiran Ren and Yin Yang and Xuansong Xie and XianSheng Hua and Hujun Bao and Weiwei Xu},
journal= {arXiv preprint arXiv:2102.02696},
year = {2022}
}