Source-Free Domain Adaptation for Semantic Segmentation

Liu, Yuang; Zhang, Wei; Wang, Jun

doi:10.1109/cvpr46437.2021.00127

Cited by 181 publications

(97 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the target dataset, retraining can be limited to the feature-based classifier as its predictions-unlike the raw data-do not exhibit a significant domain shift if the same semantic classes are being segmented. In more detail, we propose the following sequence of steps (see also Figure 1): (2021) and Liu et al (2021). At the same time, we can fully profit from all advances in the field of pseudo-label rectification (Prabhu et al, 2021;Wu et al, 2021;Zhang et al, 2021;Zhao et al, 2021), applying those to pseudo-labels generated by the PE network.…”

Section: Methodsmentioning

confidence: 99%

From Shallow to Deep: Exploiting Feature-Based Classifiers for Domain Adaptation in Semantic Segmentation

Matskevych¹,

Wolny²,

Pape³

et al. 2022

Front. Comput. Sci.

View full text Add to dashboard Cite

The remarkable performance of Convolutional Neural Networks on image segmentation tasks comes at the cost of a large amount of pixelwise annotated images that have to be segmented for training. In contrast, feature-based learning methods, such as the Random Forest, require little training data, but rarely reach the segmentation accuracy of CNNs. This work bridges the two approaches in a transfer learning setting. We show that a CNN can be trained to correct the errors of the Random Forest in the source domain and then be applied to correct such errors in the target domain without retraining, as the domain shift between the Random Forest predictions is much smaller than between the raw data. By leveraging a few brushstrokes as annotations in the target domain, the method can deliver segmentations that are sufficiently accurate to act as pseudo-labels for target-domain CNN training. We demonstrate the performance of the method on several datasets with the challenging tasks of mitochondria, membrane and nuclear segmentation. It yields excellent performance compared to microscopy domain adaptation baselines, especially when a significant domain shift is involved.

show abstract

Section: Methodsmentioning

confidence: 99%

From Shallow to Deep: Exploiting Feature-Based Classifiers for Domain Adaptation in Semantic Segmentation

Matskevych¹,

Wolny²,

Pape³

et al. 2022

Front. Comput. Sci.

View full text Add to dashboard Cite

show abstract

“…However, the main drawback of HTL is that it requires a small set of labeled target data. Inspired by HTL, source-free domain adaptation (SF-DA) [4,18,21,20,19,14,33,11] has recently flourished in the domain adaptation community. In SF-DA, instead of the source data, the source pre-trained model is provided in the target training stage.…”

Section: Related Workmentioning

confidence: 99%

“…In addition, an instance-level weighting mechanism is proposed for effective target adaptation. Lately, Liu et al [21] introduce source-free domain adaptation for semantic segmentation and utilize the batch normalization statistics of the source model to recover source-like samples. In this work, we introduce the source-free open compound domain adaptation (SF-OCDA) for semantic segmentation, extending SF-DA to a more realistic setting.…”

Section: Related Workmentioning

confidence: 99%

“…However, the source data are not always available due to data privacy. In addition, the source data are generally very large, DA [34] SF-DA [21] DG [38] OCDA [22] SF-OCDA which require plenty of storage space (e.g., GTA5 [29]≈57GB). This further limits the applications of existing methods, especially when transferring to a lightweight self-driving device.…”

Section: Introductionmentioning

confidence: 99%

“…These facts motivate us to introduce a more challenging but practical setting for OCDA, called source-free OCDA (SF-OCDA), where only the source pre-trained model and the unlabeled target data are available during the training of the target model. In the literature, source-free domain adaptation (SF-DA) has recently been developed in image classification [19,14] and semantic segmentation [21] for the single target case. However, as shown in Fig.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Source-Free Open Compound Domain Adaptation in Semantic Segmentation

Zhao¹,

Zhang²,

Luo³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this work, we introduce a new concept, named source-free open compound domain adaptation (SF-OCDA), and study it in semantic segmentation. SF-OCDA is more challenging than the traditional domain adaptation but it is more practical. It jointly considers (1) the issues of data privacy and data storage and ( 2) the scenario of multiple target domains and unseen open domains. In SF-OCDA, only the source pre-trained model and the target data are available to learn the target model. The model is evaluated on the samples from the target and unseen open domains. To solve this problem, we present an effective framework by separating the training process into two stages: (1) pre-training a generalized source model and (2) adapting a target model with self-supervised learning. In our framework, we propose the Cross-Patch Style Swap (CPSS) to diversify samples with various patch styles in the feature-level, which can benefit the training of both stages. First, CPSS can significantly improve the generalization ability of the source model, providing more accurate pseudo-labels for the latter stage. Second, CPSS can reduce the influence of noisy pseudo-labels and also avoid the model overfitting to the target domain during self-supervised learning, consistently boosting the performance on the target and open domains. Experiments demonstrate that our method produces state-of-the-art results on the C-Driving dataset. Furthermore, our model also achieves the leading performance on CityScapes for domain generalization. * Equal contribution. Preprint. Under review.

show abstract

Union-Set Multi-source Model Adaptation for Semantic Segmentation

Togo

Ogawa

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Source-Free Domain Adaptation for Semantic Segmentation

Cited by 181 publications

References 29 publications

From Shallow to Deep: Exploiting Feature-Based Classifiers for Domain Adaptation in Semantic Segmentation

From Shallow to Deep: Exploiting Feature-Based Classifiers for Domain Adaptation in Semantic Segmentation

Source-Free Open Compound Domain Adaptation in Semantic Segmentation

Union-Set Multi-source Model Adaptation for Semantic Segmentation

Contact Info

Product

Resources

About