Instance Adaptive Self-training for Unsupervised Domain Adaptation

Mei, Ke; Zhu, Chuang; Zou, Jiaqi; Zhang, Shanghang

doi:10.1007/978-3-030-58574-7_25

Cited by 220 publications

(148 citation statements)

References 29 publications

Supporting

Mentioning

147

Contrasting

Order By: Relevance

“…The goal of unsupervised domain adaptive segmentation is to train a segmentation network that achieves good performance in an unlabeled target domain when only the source domain data are annotated. Methodology-wise, existing methods build on three techniques: 1) adversarial learning [7,22,32,33,36,42,44,46], 2) image-toimage translation [2,10,12,14,17,25,31,40] and 3) self-training [1,15,18,19,24,36,43,47,49,53].…”

Section: Related Workmentioning

confidence: 99%

“…Versatility. As Table 6 shows, we justify the versatility of our method by testing its performance with a different pseudo label thresholding scheme: instance adaptive selection (IAS) [24]. Instead Table 6: Justification of the versatility of our method on GTA→Cityscapes.…”

Section: Training Detailsmentioning

confidence: 99%

“…Instead Table 6: Justification of the versatility of our method on GTA→Cityscapes. For PLA + IAS and PLA + CTS + IAS, we assign pseudo labels using instance adaptive selector (IAS) [24] rather than our pseudo label thresholding scheme (Eq. 3).…”

Section: Training Detailsmentioning

confidence: 99%

“…3). of determining a global set of per-class thresholds for all samples in the dataset (section 4.1), IAS [24] applies a unique set of perclass thresholds to each sample, which is determined by a convex combination of the current sample thresholds and a historical one during pseudo label generation. It can be observed that our method can also achieve a significant performance gain when IAS is used.…”

Section: Training Detailsmentioning

confidence: 99%

See 3 more Smart Citations

Pixel-level Intra-domain Adaptation for Semantic Segmentation

Yan

Qin

et al. 2021

Proceedings of the 29th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Recent advances in unsupervised domain adaptation have achieved remarkable performance on semantic segmentation tasks. Despite such progress, existing works mainly focus on bridging the interdomain gaps between the source and target domain, while only few of them noticed the intra-domain gaps within the target data. In this work, we propose a pixel-level intra-domain adaptation approach to reduce the intra-domain gaps within the target data. Compared with image-level methods, ours treats each pixel as an instance, which adapts the segmentation model at a more fine-grained level. Specifically, we first conduct the inter-domain adaptation between the source and target domain; Then, we separate the pixels in target images into the easy and hard subdomains; Finally, we propose a pixel-level adversarial training strategy to adapt a segmentation network from the easy to the hard subdomain. Moreover, we show that the segmentation accuracy can be further improved by incorporating a continuous indexing technique in the adversarial training. Experimental results show the effectiveness of our method against existing state-of-the-art approaches.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Training Detailsmentioning

confidence: 99%

Section: Training Detailsmentioning

confidence: 99%

Section: Training Detailsmentioning

confidence: 99%

See 2 more Smart Citations

Pixel-level Intra-domain Adaptation for Semantic Segmentation

Yan

Qin

et al. 2021

Proceedings of the 29th ACM International Conference on Multimedia

View full text Add to dashboard Cite

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