Active Boundary Loss for Semantic Segmentation

Wang, Chi; Zhang, Yunke; Cui, Miaomiao; Liu, Jinlin; Ren, Peiran; Yang, Yin; Xie, Xuansong; Hua, Xian-Sheng; Bao, Hujun; Xu, Weiwei

doi:10.1609/aaai.v36i2.20139

Cited by 31 publications

(21 citation statements)

References 71 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(3) Boundary-aware loss. These works guide the model to focus more on boundary accuracy by designing the boundary-aware loss function with high sensitivity to the boundary changes [3,20,45]. Inverseform [3] introduced a boundary distance-based measure into the popular segmentation loss functions.…”

Section: Related Workmentioning

confidence: 99%

“…[20] proposed a boundary loss taking the form of a distance metric on the space of contours for highly unbalanced situations. ABL [45] designed an active boundary loss function for progressively encouraging the predicted boundaries matching the ground-truth boundaries. Instead of imposing various boundary-related constraints, we consider the intrinsic reason for fuzzy boundary from the operator level, and design a boundary-sensitive operator to make the network naturally capable of modeling and perceiving boundaries.…”

Section: Related Workmentioning

confidence: 99%

“…A key challenge in semantic segmentation is how to improve the boundary quality of segmentation results (Figure 1). In general, people have been working on this problem in three different ways: (i) adopting a post-processing module to refine the boundary [26,23,43,39,58], (ii) jointly learned the edge detection and semantic segmentation task for boundary-awareness [60,7,30], and (iii) making training more focused on the boundary by designing a boundary-aware loss with higher sensitivity to the boundary changes [3,20,45].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Semantic Diffusion Network for Semantic Segmentation

Tan¹,

Wu²,

Pi³

2023

Preprint

View full text Add to dashboard Cite

Precise and accurate predictions over boundary areas are essential for semantic segmentation. However, the commonly-used convolutional operators tend to smooth and blur local detail cues, making it difficult for deep models to generate accurate boundary predictions. In this paper, we introduce an operator-level approach to enhance semantic boundary awareness, so as to improve the prediction of the deep semantic segmentation model. Specifically, we first formulate the boundary feature enhancement as an anisotropic diffusion process. We then propose a novel learnable approach called semantic diffusion network (SDN) to approximate the diffusion process, which contains a parameterized semantic difference convolution operator followed by a feature fusion module. Our SDN aims to construct a differentiable mapping from the original feature to the inter-class boundary-enhanced feature. The proposed SDN is an efficient and flexible module that can be easily plugged into existing encoder-decoder segmentation models. Extensive experiments show that our approach can achieve consistent improvements over several typical and state-of-the-art segmentation baseline models on challenging public benchmarks. The code will be released soon.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Semantic Diffusion Network for Semantic Segmentation

Tan¹,

Wu²,

Pi³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In SPP, we split the whole training into two steps, and only add C c L d c in the second step. This idea draws on [48], where a loss called ABL is added at the last 20% epochs, since the gradient of ABL is not useful when semantic edges output by the network are far from the semantic edge ground truth at the beginning of the training, much similar to our case.…”

Section: Ablation Studiesmentioning

confidence: 99%

TriangleNet: Edge Prior Augmented Network for Semantic Segmentation through Cross-Task Consistency

Zhang¹,

Zheng²

2022

Preprint

View full text Add to dashboard Cite

Semantic segmentation is a classic computer vision problem dedicated to labeling each pixel with its corresponding category. As a basic task for advanced tasks such as industrial quality inspection, remote sensing information extraction, medical diagnostic aid, and autonomous driving, semantic segmentation has been developed for a long time in combination with deep learning, and a lot of works have been accumulated. However, neither the classic FCN-based works nor the popular Transformer-based works have attained fine-grained localization of pixel labels, which remains the main challenge in this field. Recently, with the popularity of autonomous driving, the segmentation of road scenes has received more and more attention. Based on the cross-task consistency theory, we incorporate edge priors into semantic segmentation tasks to obtain better results. The main contribution is that we provide a model-agnostic method that improves the accuracy of semantic segmentation models with zero extra inference runtime overhead, verified on the datasets of road and non-road scenes. From our experimental results, our method can effectively improve semantic segmentation accuracy.

show abstract

“…The next goal is to formulate the detection problem using a multi-task loss that learns a distributional loss using the log-likelihood term, refines the boundaries of the crack, and align it to the ground truth using a boundary loss term. This formulation borrows from Wang et al (2021) to compute a boundary loss that learns to better align predicted and ground truth boundaries. To demonstrate the effectiveness of the method, we conduct experiments on two commonly used crack segmentation datasets and report results on model performance, calibration, and uncertainty for both within and out of distribution samples.…”

Section: Introductionmentioning

confidence: 99%