BASNet: Boundary-Aware Salient Object Detection

By Xuebin Qin, Chenyang Huang, Chao Gao, Masood Dehghan, Martin Jägersand. Computer Vision and Pattern Recognition, 2019

1,441 citations recorded by the source corpus.

Deep Convolutional Neural Networks have been adopted for salient object detection and achieved the state-of-the-art performance. Most of the previous works however focus on region accuracy but not on the boundary quality. In this paper, we propose a predict-refine architecture, BASNet, and a new hybrid loss for Boundary-Aware Salient object detection. Specifically, the architecture is composed of a densely supervised Encoder-Decoder network and a residual refinement module, which are respectively in charge of saliency prediction and saliency map refinement. The hybrid loss guides the network to learn the transformation between the input image and the ground truth in a three-level hierarchy -- pixel-, patch- and map- level -- by fusing Binary Cross Entropy (BCE), Structural SIMilarity (SSIM) and Intersection-over-Union (IoU) losses. Equipped with the hybrid loss, the proposed predict-refine architecture is able to effectively segment the salient object regions and accurately predict the fine structures with clear boundaries. Experimental results on six public datasets show that our method outperforms the state-of-the-art methods both in terms of regional and boundary evaluation measures. Our method runs at over 25 fps on a single GPU. The code is available at: https://github.com/NathanUA/BASNet.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/cebd4ab4ab52be88b26d976aa7d4fb35cc19c2a2

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

Citations in the corpus

Automatic Polyp Segmentation via Multi-scale Subtraction Network (2021, methodology): “We use the same definitions as in [7,25,19] and their effectiveness has been validated in theses works.”
Distilling Localization for Self-Supervised Representation Learning (2020, methodology): “Deep supervised methods [32] train a segmentation network to regress the foreground mask, outper-”
Suppress and Balance: A Simple Gated Network for Salient Object Detection (2020, methodology): “And many methods aim at combining multilevel features in either the encoder [69,53,67,57,37,59] or the decoder [29,20,70,59].”
Interactive Two-Stream Decoder for Accurate and Fast Saliency Detection (2020, methodology): “More recently, many researches [18, 24, 48, 26] further improve the U-shape structure by incorporating contour information.”
Concealed Object Detection (2021, background): “According to previous studies [59], [60], multistage refinement can improve performance.”
F3Net: Fusion, Feedback and Focus for Salient Object Detection (2019, methodology): “We follow (Wu, Su, and Huang 2019; Qin et al. 2019) to use DUTS-TR as the training dataset and others as testing datasets.”
PraNet: Parallel Reverse Attention Network for Polyp Segmentation (2020, background): “ortance. In addition, compared with the standard BCE loss, Lw BCE pays more attention to hard pixels rather than assigning all pixels equal weights. The definitions of these losses are the same as in [21,26] and their eectiveness has been validated in the eld of salient object detection. Here, we ad”
Global Context-Aware Progressive Aggregation Network for Salient Object Detection (2020, methodology): “Different from previous works (Qin et al. 2019; Liu et al. 2019) that often simply fuse the high-level features after up-sampling with the low-level features by concatenation or addition operation, we adopt a more aggressive yet efﬁcient operation, i.e., multiplication.”
Camouflaged Object Segmentation with Distraction Mining (2021, methodology): “…detection methods DSC [20] and BDRAR [72]; medical image segmentation methods UNet++ [71] and PraNet [13]; salient object detection methods PiCANet [33], BASNet [44], CPD [55], PFANet [69], EGNet [67], F3Net [53], GC-PANet [5], and MINet-R [40]; and camouﬂaged object segmentation method SINet [12].”
Uncertainty-aware Joint Salient Object and Camouflaged Object Detection (2021, methodology): “In this way, the SOD models [48, 46, 42, 38] are designed based on both global contrast and local contrast, while the COD models [11, 26, 35, 52, 34] usually avoid searching the camouﬂaged objects in those salient regions.”
Context-aware Cross-level Fusion Network for Camouflaged Object Detection (2021, background): “In order to make up for its deficiency of ignoring the global structure, [Qin et al., 2019] introduced IoU loss(LIoU) to focus on the global structure to form a global restriction on the network.”
Dual Attention Guided Gaze Target Detection in the Wild (2021, methodology): “For these reasons, gaze analysis has widely been used in neurophysiology studies [36, 10], relevant saliency prediction [8, 34], and social awareness tracking [7, 29, 30].”
UC-Net: Uncertainty Inspired RGB-D Saliency Detection via Conditional Variational Autoencoders (2020, methodology): “We first hide the ground truth salient object in the RGB image, and feed the modified image to an existing RGB saliency detection model [42] to produce a saliency map and treat it as one candidate annotation.”