Semi-Supervised Semantic Segmentation With High- and Low-Level Consistency

Mittal, Sudhanshu; Tatarchenko, Maxim; Brox, Thomas

doi:10.1109/tpami.2019.2960224

Cited by 327 publications

(221 citation statements)

References 28 publications

Supporting

Mentioning

219

Contrasting

Order By: Relevance

“…After FCN [18] and U-net [19] created the basic semantic segmentation network architecture under supervised learning, researchers have also strived to leverage weakly supervision instead such as multiple instance learning [20], EM algorithm [21] and constrained CNN [22], or semisupervision by additionally using a few pixel-wise segmentation labels [23], [24]. Similar to this paper, some weakly supervised methods [25], [26] used attention masks and classification tags. They achieved an excellent level of semantic segmentation.…”

Section: Weakly Supervised Semantic Segmentationmentioning

confidence: 99%

Self-Attention-Masking Semantic Decomposition and Segmentation for Facial Attribute Manipulation

Xia

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Many face attribute manipulation methods can only provide global attribute manipulation according to the attribute labels. In this paper, we propose a self-attention-masking semantic decomposition method which is able to learn an attribute attention mask for each attribute. User can adjust the strength and color of each attribute smoothly and more freely. We decouple the attention of different attributes and overcome the disadvantage of overlap between different attribute attention masks by an attention weighting module. Thanks to the attribute attention masks, our method allows manipulate facial attribute without generator after only once generation. Moreover, we can perform facial semantic segmentation without pixel level semantic labels. Experiments show that our method simultaneously improves the freedom of attribute manipulation and the authenticity of synthetic face. The mean intersection over union of semantic segmentation is over 65% for hair and skin. Our code is available at github.com/flyfeatherok/SAMSD.

show abstract

Section: Weakly Supervised Semantic Segmentationmentioning

confidence: 99%

Self-Attention-Masking Semantic Decomposition and Segmentation for Facial Attribute Manipulation

Xia

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Only four recent works consider true semisupervised learning for semantic segmentation, which uses a large number of unlabeled samples and a small number of fully labeled samples to train the network. Three works [14], [22], [28] use Generative Adversarial Networks (GAN) as a backbone network, but have different training manners. Souly et al [28] utilizes GAN to generate additional data for improving the feature extraction capabilities of the segmentation network.…”

Section: Semi-supervised Semantic Segmentationmentioning

confidence: 99%

“…Souly et al [28] utilizes GAN to generate additional data for improving the feature extraction capabilities of the segmentation network. Hung et al [14] and Mittal et al [22] proposed a GAN-based segmentation network that can learn beneficial feature information from unlabeled data to assist in the training of labeled data. On the other hand, Bellver et al [3] uses the aforementioned self-learning framework for semantic segmentation, where the prediction results of unlabeled images are accepted as ground truth labels.…”

Section: Semi-supervised Semantic Segmentationmentioning

confidence: 99%

“…Only several previous works have consider true semisupervised learning for semantic segmentation. These methods are divided into two types: one is to learn directly from unlabeled samples using GANs [14], [22], [28], and the other [3] is to follow the standard self-training framework in image classification [34], [35] which has three main steps: 1) train a teacher model on ground truth(GT) labeled data, 2) use the teacher to predict pesudo labels on unlabeled images, 3) train a student model with all GT annotated and pseudo-annotated samples. Follow the same training pipeline with Bellver et al [3], the training pipeline of semi-supervised learning consists of two networks: a teacher network f θ trained with GT labels which can generate pseudoannotations for unlabeled data, and a student network g ϕ trained with the combination of GT and pseudo labeled samples.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Digging Into Pseudo Label: A Low-Budget Approach for Semi-Supervised Semantic Segmentation

Chen

Zhang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

The capability to understand visual scenes with limited labeled data has been widely concerned in the field of computer vision. Although semi-supervised learning for image classification has been extensively studied in some cases, semantic segmentation with limited data has only recently gained attention. In this work, we follow the standard semi-supervised segmentation pipeline for image classification and propose a two-branch network that can encode strong and pseudo label spaces respectively, extracting reliable supervision information from pseudo-labels to assist in training network with strong labels. Our method outperforms previous semi-supervised methods with limited annotation cost. On standard benchmark PASCAL VOC 2012 for semi-supervised semantic segmentation, the proposed approach gains fresh state-of-the-art performance.

show abstract

“…Semisupervised segmentation methods are becoming increasingly prevalent in the computer vision domain, which attempt to reduce the annotation effort by utilizing a large set of unlabeled data for model learning. Most of those methods 35‐37 employ adversarial loss for semisupervised learning on unlabeled data. Inspired by them, in this article, our proposed SemiTongue attempts to conduct segmentation in the semisupervised setting by utilizing a large number of unlabeled data, while imposes a self‐reconstruction constraint between labeled and unlabeled data to improve the accuracy of tongue segmentation.…”

Section: Related Workmentioning

confidence: 99%

Reconstruction enhanced probabilistic model for semisupervised tongue image segmentation

Zhou

Fan

Zhao

et al. 2020

Concurrency and Computation

View full text Add to dashboard Cite

Tongue segmentation is a key step of automatic tongue diagnosis, and the major challenges for the effective segmentation lie in the large appearance variations of tongue caused by different diseases, for example, tongue coating and tongue texture. Moreover, the limited labeled data also hinders traditional supervised methods from their powerful learning ability. To alleviate these challenges, in this work, we propose a reconstruction enhanced probabilistic model for semisupervised tongue segmentation, named SemiTongue, in which, image reconstruction constraint combined with adversarial learning is used to improve the accuracy of tongue segmentation. Specifically, based on a shared feature encoder that served as an inference model, two separate branches in SemiTongue as the generative model, which are composed of a segmentation decoder and a reconstruction decoder, are utilized to generate the tongue segmentation and reconstruct original tongue image respectively. Then, a discriminator is employed to differentiate the generated segmentation map from the ground truth segmentation distribution. Moreover, semisupervised learning is conducted through discriminator by discovering the reliable region in the generated segmentation map of unlabeled images, which is further utilized to supervise the segmentation branch. Experimental results compared with state-of-the-art methods on real-world datasets demonstrate the effectiveness of SemiTongue.

show abstract

Semi-Supervised Semantic Segmentation With High- and Low-Level Consistency

Cited by 327 publications

References 28 publications

Self-Attention-Masking Semantic Decomposition and Segmentation for Facial Attribute Manipulation

Self-Attention-Masking Semantic Decomposition and Segmentation for Facial Attribute Manipulation

Digging Into Pseudo Label: A Low-Budget Approach for Semi-Supervised Semantic Segmentation

Reconstruction enhanced probabilistic model for semisupervised tongue image segmentation

Contact Info

Product

Resources

About