Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaption

Chao, Chen-Hao; Cheng, Bo-Wun; Lee, Chun-Yi

doi:10.1109/cvprw53098.2021.00295

Cited by 11 publications

(4 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tables 1 and 2 compare the results of ELDA against multiple baselines. Please note that these baselines do not include works that resort to ensemble distillation methods [2,36,37] or transformer based architectures [13] for fair comparisons. We also include the performance of ELDA trained solely in the source domains, denoted as source only, for reference.…”

Section: Quantitative Results On the Benchmarksmentioning

confidence: 99%

“…ELDA utilizes edges as the domain invariant information by incorporating edge extraction into its training process as an auxiliary task. The experimental results show that without resorting to ensemble distillation methods [2,36,37] or transformer based architectures [13], ELDA is able to achieve the state-of-the-art performance on two commonly adopted benchmarks [7,23,24]. The contributions of this work are summarized as follows:…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

ELDA: Using Edges to Have an Edge on Semantic Segmentation Based UDA

Liao¹,

Liao²,

Yang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Many unsupervised domain adaptation (UDA) methods have been proposed to bridge the domain gap by utilizing domain invariant information. Most approaches have chosen depth as such information and achieved remarkable successes. Despite their effectiveness, using depth as domain invariant information in UDA tasks may lead to multiple issues, such as excessively high extraction costs and difficulties in achieving a reliable prediction quality. As a result, we introduce Edge Learning based Domain Adaptation (ELDA), a framework which incorporates edge information into its training process to serve as a type of domain invariant information. In our experiments, we quantitatively and qualitatively demonstrate that the incorporation of edge information is indeed beneficial and effective, and enables ELDA to outperform the contemporary state-of-the-art methods on two commonly adopted benchmarks for semantic segmentation based UDA tasks. In addition, we show that ELDA is able to better separate the feature distributions of different classes. We further provide ablation analysis to justify our design decisions.

show abstract

Section: Quantitative Results On the Benchmarksmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

ELDA: Using Edges to Have an Edge on Semantic Segmentation Based UDA

Liao¹,

Liao²,

Yang³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Chao et al [ 18 ] assumes the existence of a set of semantic segmentation models independently pre-trained according to some UDA technique. Then, the pseudo-label confidences coming from such models are unified, fused, and finally distilled into a student model.…”

Section: Related Workmentioning

confidence: 99%

“…It is well-known that training deep models on synthetic images for performing on real-world ones requires domain adaptation [ 8 , 9 ], which must be unsupervised if we have no labels from real-world images [ 10 ]. Thus, this paper falls into the realm of unsupervised domain adaptation (UDA) for semantic segmentation [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ], i.e., in contrast to assuming access to labels from the target domain [ 23 , 24 ]. Note that the great relevance of UDA in this context comes from the fact that, until now, pixel-level semantic image segmentation labels are obtained by cumbersome and error-prone manual work.…”

Section: Introductionmentioning

confidence: 99%

Co-Training for Unsupervised Domain Adaptation of Semantic Segmentation Models

Gómez

Villalonga

López

2023

Sensors

View full text Add to dashboard Cite

Semantic image segmentation is a core task for autonomous driving, which is performed by deep models. Since training these models draws to a curse of human-based image labeling, the use of synthetic images with automatically generated labels together with unlabeled real-world images is a promising alternative. This implies addressing an unsupervised domain adaptation (UDA) problem. In this paper, we propose a new co-training procedure for synth-to-real UDA of semantic segmentation models. It performs iterations where the (unlabeled) real-world training images are labeled by intermediate deep models trained with both the (labeled) synthetic images and the real-world ones labeled in previous iterations. More specifically, a self-training stage provides two domain-adapted models and a model collaboration loop allows the mutual improvement of these two models. The final semantic segmentation labels (pseudo-labels) for the real-world images are provided by these two models. The overall procedure treats the deep models as black boxes and drives their collaboration at the level of pseudo-labeled target images, i.e., neither modifying loss functions is required, nor explicit feature alignment. We test our proposal on standard synthetic and real-world datasets for onboard semantic segmentation. Our procedure shows improvements ranging from approximately 13 to 31 mIoU points over baselines.

show abstract

Monitored Distillation for Positive Congruent Depth Completion

Agrawal

Chen

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaption

Cited by 11 publications

References 37 publications

ELDA: Using Edges to Have an Edge on Semantic Segmentation Based UDA

ELDA: Using Edges to Have an Edge on Semantic Segmentation Based UDA

Co-Training for Unsupervised Domain Adaptation of Semantic Segmentation Models

Monitored Distillation for Positive Congruent Depth Completion

Contact Info

Product

Resources

About