Two-Phase Pseudo Label Densification for Self-training Based Domain Adaptation

Shin, Inkyu; Woo, Sanghyun; Pan, Fang; Kweon, In So

doi:10.1007/978-3-030-58601-0_32

Cited by 89 publications

(47 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Many works have focused on the first step, how to get believable pseudo labels. They presented weakly-supervised learning by image-level domain transfer and annotation information [66], exploiting temporal cues using a tracker on unlabeled videos is presented [67], relying on pseudo labels of the intermediate domain and the ground-truth of source domain in an imbalanced sampling way [68], using k-reciprocal encoding and clustering algorithm in feature space [69], and treating easy and hard pseudo labels differently [70].…”

Section: Related Workmentioning

confidence: 99%

“…Like other computer vision problems, this cut-and-paste approach also suffers from performance degradation when there is a distribution mismatch between the training domain (or source domain) and the test domain (or target domain). Domain adaptation methods have been introduced to tackle this domain shift problem by aligning features [39]- [46], [61]- [65] and self-training scheme [66]- [70]. However, these approaches require knowledge of the target domain.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Cut-and-Paste Dataset Generation for Balancing Domain Gaps in Object Instance Detection

et al. 2021

View full text Add to dashboard Cite

Training an object instance detector where only a few training object images are available is a challenging task. One solution is a cut-and-paste method that generates a training dataset by cutting object areas out of training images and pasting them onto other background images. A detector trained on a dataset generated with a cut-and-paste method suffers from the conventional domain shift problem, which stems from a discrepancy between the source domain (generated training dataset) and the target domain (real test dataset). Though state-of-the-art domain adaptation methods are able to reduce this gap, it is limited because they do not consider the difference of domain gaps of foreground and background. In this study, we present that the conventional domain gap can be divided into two sub-domain gaps for foreground and background. Then, we show that the original cut-and-paste approach suffers from a new domain gap problem, an unbalanced domain gaps, because it has two separate source domains for foreground and background, unlike the conventional domain shift problem. Then, we introduce an advanced cut-and-paste method to balance the unbalanced domain gaps by diversifying the foreground with GAN (generative adversarial network)-generated seed images and simplifying the background using image processing techniques. Experimental results show that our method is effective for balancing domain gaps and improving the accuracy of object instance detection in a cluttered indoor environment using only a few seed images. Furthermore, we show that balancing domain gaps can improve the detection accuracy of state-of-the-art domain adaptation methods.INDEX TERMS Artificial neural networks, image processing, learning (artificial intelligence), object detection.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Cut-and-Paste Dataset Generation for Balancing Domain Gaps in Object Instance Detection

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Semantic segmentation adaptation. We divide the existing UDA semantic segmentation methods into two categories: domain alignment [6,7,18,19] and self-training [11,12,15,24,25], and the existing state-of-art approaches are usually a combination of two methods. The main motivation of domain alignment is to reduce the discrepancy between two domains.…”

Section: Related Workmentioning

confidence: 99%

“…In CRST [25], a confidence regularized self-training method is proposed to address the problem of overconfident wrong label. [15] presents a two-phase pseudo label densification framework through voting-based and easy-hard classification based method. In [12], weak labels are explored to enhance pseudo label learning.…”

Section: Related Workmentioning

confidence: 99%

Multiple Fusion Adaptation: A Strong Framework for Unsupervised Semantic Segmentation Adaptation

Zhang¹,

Wang

et al. 2021

Preprint

View full text Add to dashboard Cite

This paper challenges the cross-domain semantic segmentation task, aiming to improve the segmentation accuracy on the unlabeled target domain without incurring additional annotation. Using the pseudo-label-based unsupervised domain adaptation (UDA) pipeline, we propose a novel and effective Multiple Fusion Adaptation (MFA) method. MFA basically considers three parallel information fusion strategies, i.e., the cross-model fusion, temporal fusion and a novel online-offline pseudo label fusion. Specifically, the online-offline pseudo label fusion encourages the adaptive training to pay additional attention to difficult regions that are easily ignored by offline pseudo labels, therefore retaining more informative details. While the other two fusion strategies may look standard, MFA pays significant efforts to raise the efficiency and effectiveness for integration, and succeeds in injecting all the three strategies into a unified framework. Experiments on two widely used benchmarks, i.e., GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes, show that our method significantly improves the semantic segmentation adaptation, and sets up new state of the art (58.2% and 62.5% mIoU, respectively). The code will be available at https://github.com/KaiiZhang/MFA. This paper considers the unsupervised domain adaptation (UDA) for semantic segmentation. In real-world segmentation tasks, there usually exists a domain gap between the training (source domain) and testing data (target domain), which substantially compromises the segmentation accuracy. Instead of using additional annotated data on the target domain for adaptation, which is notoriously expensive, an alternative way is to adapt the already-learned

show abstract

“…Recently, self-training based UDA presents a powerful means to counter unknown labels in the target domain [33], surpassing the adversarial learning-based methods in many discriminative UDA benchmarks, e.g., classification and segmentation (i.e., pixel-wise classification) [31,23,26]. The core idea behind the deep self-training based UDA is to iteratively generate a set of one-hot (or smoothed) pseudo-labels in the target domain, followed by retraining the network based on these pseudo-labels with target data [33].…”

Section: Introductionmentioning

confidence: 99%

Generative Self-training for Cross-Domain Unsupervised Tagged-to-Cine MRI Synthesis

Liu

Xing

Stone

et al. 2021

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Self-training based unsupervised domain adaptation (UDA) has shown great potential to address the problem of domain shift, when applying a trained deep learning model in a source domain to unlabeled target domains. However, while the self-training UDA has demonstrated its effectiveness on discriminative tasks, such as classification and segmentation, via the reliable pseudo-label selection based on the softmax discrete histogram, the self-training UDA for generative tasks, such as image synthesis, is not fully investigated. In this work, we propose a novel generative self-training (GST) UDA framework with continuous value prediction and regression objective for cross-domain image synthesis. Specifically, we propose to filter the pseudo-label with an uncertainty mask, and quantify the predictive confidence of generated images with practical variational Bayes learning. The fast test-time adaptation is achieved by a round-based alternative optimization scheme. We validated our framework on the tagged-to-cine magnetic resonance imaging (MRI) synthesis problem, where datasets in the source and target domains were acquired from different scanners or centers. Extensive validations were carried out to verify our framework against popular adversarial training UDA methods. Results show that our GST, with tagged MRI of test subjects in new target domains, improved the synthesis quality by a large margin, compared with the adversarial training UDA methods.

show abstract

Two-Phase Pseudo Label Densification for Self-training Based Domain Adaptation

Cited by 89 publications

References 31 publications

Cut-and-Paste Dataset Generation for Balancing Domain Gaps in Object Instance Detection

Cut-and-Paste Dataset Generation for Balancing Domain Gaps in Object Instance Detection

Multiple Fusion Adaptation: A Strong Framework for Unsupervised Semantic Segmentation Adaptation

Generative Self-training for Cross-Domain Unsupervised Tagged-to-Cine MRI Synthesis

Contact Info

Product

Resources

About