Mining Cross-Image Semantics for Weakly Supervised Semantic Segmentation

By Guolei Sun, Wenguan Wang, Jifeng Dai, Luc Van Gool. 2020

356 citations recorded by the source corpus.

This paper studies the problem of learning semantic segmentation from image-level supervision only. Current popular solutions leverage object localization maps from classifiers as supervision signals, and struggle to make the localization maps capture more complete object content. Rather than previous efforts that primarily focus on intra-image information, we address the value of cross-image semantic relations for comprehensive object pattern mining. To achieve this, two neural co-attentions are incorporated into the classifier to complimentarily capture cross-image semantic similarities and differences. In particular, given a pair of training images, one co-attention enforces the classifier to recognize the common semantics from co-attentive objects, while the other one, called contrastive co-attention, drives the classifier to identify the unshared semantics from the rest, uncommon objects. This helps the classifier discover more object patterns and better ground semantics in image regions. In addition to boosting object pattern learning, the co-attention can leverage context from other related images to improve localization map inference, hence eventually benefiting semantic segmentation learning. More essentially, our algorithm provides a unified framework that handles well different WSSS settings, i.e., learning WSSS with (1) precise image-level supervision only, (2) extra simple single-label data, and (3) extra noisy web data. It sets new state-of-the-arts on all these settings, demonstrating well its efficacy and generalizability. Moreover, our approach ranked 1st place in the Weakly-Supervised Semantic Segmentation Track of CVPR2020 Learning from Imperfect Data Challenge.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Incremental Learning in Semantic Segmentation from Image Labels (2021, methodology): “To address this issue, Weakly Supervised Semantic Segmentation (WSSS) methods aim to obtain effective segmentation models using cheaper supervisions such as bounding boxes [18, 32, 49], scribbles [39, 60], points [6, 53], and image-level labels [31, 34, 35, 59].”
GroupViT: Semantic Segmentation Emerges from Text Supervision (2022, background): “Another line of related research is of weaklysupervised semantic segmentation [1,9,26,31,36,40,65,68, 80], which aims to learn semantic segmentation with only image-level object category supervision.”
Multi-class Token Transformer for Weakly Supervised Semantic Segmentation (2022, background): “Prior works also exploited sub-categories [4] and cross-image semantics [13, 25, 33] to localize more accurate object regions.”
Causal Intervention for Weakly-Supervised Semantic Segmentation (2020, background): “There is a recent work that explicitly uses contexts to improve the multi-label classifier [49]: given a pair of images, it encourages the similarity of the foreground features of the same class and the contrast of the rest.”
Webly Supervised Concept Expansion for General Purpose Vision Models (2022, methodology): “[73] Guolei Sun, Wenguan Wang, Jifeng Dai, and Luc Van Gool.”
Railroad is not a Train: Saliency as Pseudo-pixel Supervision for Weakly Supervised Semantic Segmentation (2021, methodology): “To address this problem, cross-image affinity [15], knowledge graph [31] and contrastive optimization [38, 50] are used to improve the quality of pseudo-masks.”
PseudoSeg: Designing Pseudo Labels for Semantic Segmentation (2020, methodology): “…saliency estimation model (Oh et al., 2017; Huang et al., 2018; Wei et al., 2018), utilizing pixel similarity to propagate the initial score map (Ahn & Kwak, 2018; Wang et al., 2020), or mining and co-segment the same category of objects across images (Sun et al., 2020; Zhang et al., 2020b).”
LayerCAM: Exploring Hierarchical Class Activation Maps for Localization (2021, methodology): “Besides, because of the popularity of attention methods [1], [2], [3], many WSSS approaches [74], [75], [6], [76], [77], [78], [79], [80], [81] use the object localization cues extracted from the class activation maps.”
Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation (2021, methodology): “[62], MCIS [44], IRN [1], FickleNet[30], SSDD [42], SEAM [49], SCE [7], CONTA [60].”
Rethinking Semantic Segmentation: A Prototype View (2022, background): “Since it was proposed, numerous efforts have been devoted to improving FCN, by, for example, enlarging the receptive field [15, 16, 25, 124, 128, 135]; strengthening context cues [4, 43, 47, 48, 56, 57, 70, 75, 77, 81, 90, 126, 128, 129, 132, 138, 141]; leveraging boundary information [6, 14, 27, 66”
Hypercorrelation Squeeze for Few-Shot Segmenation (2021, background): “To cope with the challenge, there have been various attempts in semi- and weakly-supervised segmentation approaches [6, 26, 39, 66, 72, 77, 88] which in turn effectively alleviated the data-hunger issue.”
SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation (2020, background): “Diﬀerent from semantic segmentation [32,10,6,34,39], instance segmentation also diﬀerentiates multiple object instances.”
DiscoBox: Weakly Supervised Instance Segmentation and Semantic Correspondence from Box Supervision (2021, unknown): “A number of methods have been proposed to learn semantic segmentation with image-level class labels [69–73], points [74,75], scribbles [76–79] and bounding boxes [46,80–83].”