Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision

Chen, Xiaokang; Yuan, Yuhui; Zeng, Gang; Wang, Jingdong

doi:10.1109/cvpr46437.2021.00264

Cited by 688 publications

(370 citation statements)

References 19 publications

Supporting

Mentioning

369

Contrasting

Order By: Relevance

“…Contrastive learning [8] was introduced to learn a representation from unlabeled data for image classification. Recent approaches use teacher-student architectures to constraint a consistency on the Siamese-networks applied on the same input [4,5,9]. In our work, we improve the CNN training process by regularizing the loss with a dense-contrastive-learning approach for image segmentation [25], and by the teacherstudent consistency [9].…”

Section: Previous Workmentioning

confidence: 99%

“…An alternative to the semi-supervised loss that was introduced in Section 5 is to use a consistency loss instead as described in [9]. The idea is to train a network by an unsupervised loss that assumes a teacher-student consistency.…”

Section: Consistency Lossmentioning

confidence: 99%

“…This process can be done with a supervised binary-cross-entropy (BCE) loss [15]. To improve our robustness to the lack of labeled data, we use the semisupervised approaches of dense contrastive-learning (Dense Sim-Clr) [8,25] and teacher-student consistency [9]. Our experiments demonstrate that we succeed to segment real defects in SEM images, by training on defect-implanted reference images.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automatic defect segmentation by unsupervised anomaly learning

Ofir,

Yacobi,

Granoviter

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper addresses the problem of defect segmentation in semiconductor manufacturing. The input of our segmentation is a scanning-electron-microscopy (SEM) image of the candidate defect region. We train a U-net shape network to segment defects using a dataset of clean background images. The samples of the training phase are produced automatically such that no manual labeling is required. To enrich the dataset of clean background samples, we apply defect implant augmentation. To that end, we apply a copy-and-paste of a random image patch in the clean specimen. To improve robustness to the unlabeled data scenario, we train the features of the network with unsupervised learning methods and loss functions. Our experiments show that we succeed to segment real defects with high quality, even though our dataset contains no defect examples. Our approach performs accurately also on the problem of supervised and labeled defect segmentation.

show abstract

Section: Previous Workmentioning

confidence: 99%

Section: Consistency Lossmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automatic defect segmentation by unsupervised anomaly learning

Ofir,

Yacobi,

Granoviter

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…As the basic task of scene understanding, semantic segmentation technology based on pixel-by-pixel classification has been widely studied [22][23][24]. Many semantic segmentation methods based on deep learning have been proposed [25][26][27][28]. Currently, there are four main types of networks, namely, the fully convolutional network (FCN) [29], the convolutional neural network (CNN) [30], the recurrent neural network (RNN) [31], and the generative adversarial network (GAN) [32].…”

Section: Semantic Segmentation Based On Deep Learningmentioning

confidence: 99%

MAGNet: A camouflaged object detection network simulating the observation effect of a magnifier

Jiang

Cai

Jiang

et al. 2022

Preprint

View full text Add to dashboard Cite

In recent years, protecting important objects by simulating animal camouflage has been widely used in many fields. Therefore, camouflaged object detection (COD) technology has emerged. COD is more difficult than traditional object detection techniques because of the high degree of fusion of camouflaged objects with the background. In this paper, we strive to more accurately identify camouflaged objects. Inspired by the use of magnifiers to search for hidden objects in pictures, we propose a COD network that simulates the observation effect of a magnifier, called the MAGnifier Network (MAGNet). Specifically, our MAGNet contains two parallel modules, i.e., the ergodic magnification module (EMM) and the attention focus module (AFM). The EMM is designed to mimic the process of a magnifier enlarging an image, and AFM is used to simulate the observation process in which human attention is highly focused on a region. The two sets of output camouflaged object maps are merged to simulate the observation of an object by a magnifier. In addition, a weighted key point area perception loss function, which is more applicable to COD, is designed based on two modules to give higher attention to the camouflaged object. Extensive experiments demonstrate that compared with 14 cutting-edge detection models, MAGNet can achieve the best comprehensive effect on eight evaluation metrics on the public COD dataset, and the segmentation accuracy is significantly improved. We also validate the models' generalization ability on a military camouflaged object dataset constructed in-house. Finally, we experimentally explore some extended applications of COD.

show abstract

“…To alleviate the dependence on massive labeled datasets, semi-supervised learning (SSL) has gained more and more attention and become an active research area due to its desired ability to exploit unlabeled data effectively. Since unlabeled data can often be obtained at low cost, SSL has demonstrated superior performance on various tasks such as semantic segmentation [8], image classification [29], and object detection [35].…”

Section: Introductionmentioning

confidence: 99%

CLS: Cross Labeling Supervision for Semi-Supervised Learning

Yao¹,

Shen²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

It is well known that the success of deep neural networks is greatly attributed to large-scale labeled datasets. However, it can be extremely time-consuming and laborious to collect sufficient highquality labeled data in most practical applications. Semi-supervised learning (SSL) provides an effective solution to reduce the cost of labeling by simultaneously leveraging both labeled and unlabeled data. In this work, we present Cross Labeling Supervision (CLS), a framework that generalizes the typical pseudo-labeling process. Based on FixMatch [29], where a pseudo label is generated from a weakly-augmented sample to teach the prediction on a strong augmentation of the same input sample, CLS allows the creation of both pseudo and complementary labels to support both positive and negative learning. To mitigate the confirmation bias of self-labeling and boost the tolerance to false labels, two different initialized networks with the same structure are trained simultaneously. Each network utilizes high-confidence labels from the other network as additional supervision signals. During the label generation phase, adaptive sample weights are assigned to artificial labels according to their prediction confidence. The sample weight plays two roles: quantify the generated labels' quality and reduce the disruption of inaccurate labels on network training. Experimental results on the semi-supervised classification task show that our framework outperforms existing approaches by large margins on the CIFAR-10 and CIFAR-100 datasets.

show abstract

Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision

Cited by 688 publications

References 19 publications

Automatic defect segmentation by unsupervised anomaly learning

Automatic defect segmentation by unsupervised anomaly learning

MAGNet: A camouflaged object detection network simulating the observation effect of a magnifier

CLS: Cross Labeling Supervision for Semi-Supervised Learning

Contact Info

Product

Resources

About