Cross-Domain Correlation Distillation for Unsupervised Domain Adaptation in Nighttime Semantic Segmentation

Gao, Huan; Guo, Jichang; Wang, Guoli; Zhang, Qian

doi:10.1109/cvpr52688.2022.00968

Cited by 59 publications

(9 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, limited research has been conducted on domain adaptation in adverse domain contexts, particularly concerning adverse scenarios in autonomous driving. Algorithms for removing images captured under adverse conditions in the context of autonomous driving tend to focus predominantly on single-task solutions [45][46][47][48][49]. Within the context of snow scene removal, one can encounter both traditional snow denoising models founded on matrix factorization [50], as well as contemporary deep learning-based approaches, like the deep dense multi-scale network, DDMSNet [51].…”

Section: Adaptation Domain Of Adverse Conditionsmentioning

confidence: 99%

Efficient Decoder and Intermediate Domain for Semantic Segmentation in Adverse Conditions

Chen,

Jiang,

et al. 2024

Smart Cities

View full text Add to dashboard Cite

In smart city contexts, traditional methods for semantic segmentation are affected by adverse conditions, such as rain, fog, or darkness. One challenge is the limited availability of semantic segmentation datasets, specifically for autonomous driving in adverse conditions, and the high cost of labeling such datasets. To address this problem, unsupervised domain adaptation (UDA) is commonly employed. In UDA, the source domain contains data from good weather conditions, while the target domain contains data from adverse weather conditions. The Adverse Conditions Dataset with Correspondences (ACDC) provides reference images taken at different times but in the same location, which can serve as an intermediate domain, offering additional semantic information. In this study, we introduce a method that leverages both the intermediate domain and frequency information to improve semantic segmentation in smart city environments. Specifically, we extract the region with the largest difference in standard deviation and entropy values from the reference image as the intermediate domain. Secondly, we introduce the Fourier Exponential Decreasing Sampling (FEDS) algorithm to facilitate more reasonable learning of frequency domain information. Finally, we design an efficient decoder network that outperforms the DAFormer network by reducing network parameters by 28.00%. When compared to the DAFormer work, our proposed approach demonstrates significant performance improvements, increasing by 6.77%, 5.34%, 6.36%, and 5.93% in mean Intersection over Union (mIoU) for Cityscapes to ACDC night, foggy, rainy, and snowy, respectively.

show abstract

Section: Adaptation Domain Of Adverse Conditionsmentioning

confidence: 99%

Efficient Decoder and Intermediate Domain for Semantic Segmentation in Adverse Conditions

Chen,

Jiang,

et al. 2024

Smart Cities

View full text Add to dashboard Cite

show abstract

“…The realm of industrial automation relies significantly on domain adaptation techniques to facilitate the seamless integration of robotic vision systems across various production settings. Domain adaptation empowers robotic vision to maintain consistent and accurate object recognition and manipulation by adapting to changes in illumination, object textures, and camera perspectives [316,317]. For instance, domain adaptation enables robots to proficiently handle various parts and components from diverse sources within robotic assembly lines, ensuring precise grasping and assembly, while optimizing production efficiency and minimizing errors by aligning the robot's vision with the unique characteristics of each component, as shown in Figure 9.…”

Section: Domain Adaptation In Robotic Visionmentioning

confidence: 99%

An In-Depth Analysis of Domain Adaptation in Computer and Robotic Vision

Tanveer,

Fatima,

Zardari

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

This review article comprehensively delves into the rapidly evolving field of domain adaptation in computer and robotic vision. It offers a detailed technical analysis of the opportunities and challenges associated with this topic. Domain adaptation methods play a pivotal role in facilitating seamless knowledge transfer and enhancing the generalization capabilities of computer and robotic vision systems. Our methodology involves systematic data collection and preparation, followed by the application of diverse assessment metrics to evaluate the efficacy of domain adaptation strategies. This study assesses the effectiveness and versatility of conventional, deep learning-based, and hybrid domain adaptation techniques within the domains of computer and robotic vision. Through a cross-domain analysis, we scrutinize the performance of these approaches in different contexts, shedding light on their strengths and limitations. The findings gleaned from our evaluation of specific domains and models offer valuable insights for practical applications while reinforcing the validity of the proposed methodologies.

show abstract

“…Cross-domain segmentation. As a typical type of DA, UDA assumes no annotation on the target domain and thus has been a prominent problem setting in many tasks [37]- [39]. Given the great success of UDA methods in many classification tasks [9], they have also been applied in segmentation tasks, which involve more challenging structured prediction.…”

Section: Related Workmentioning

confidence: 99%

WDA-Net: Weakly-Supervised Domain Adaptive Segmentation of Electron Microscopy

Dafei

Yi²

2022

2022 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

View full text Add to dashboard Cite

Accurate segmentation of organelle instances from electron microscopy (EM) images plays an essential role in many neuroscience researches. However, practical scenarios usually suffer from high annotation costs, label scarcity, and large domain diversity. While unsupervised domain adaptation (UDA) that assumes no annotation effort on the target data is promising to alleviate these challenges, its performance on complicated segmentation tasks is still far from practical usage. To address these issues, we investigate a highly annotation-efficient weak supervision, which assumes only sparse center-points on a small subset of object instances in the target training images. To achieve accurate segmentation with partial point annotations, we introduce instance counting and center detection as auxiliary tasks and design a multitask learning framework to leverage correlations among the counting, detection, and segmentation, which are all tasks with partial or no supervision. Building upon the different domain-invariances of the three tasks, we enforce counting estimation with a novel soft consistency loss as a global prior for center detection, which further guides the per-pixel segmentation. To further compensate for annotation sparsity, we develop a cross-position cut-and-paste for label augmentation and an entropy-based pseudo-label selection. The experimental results highlight that, by simply using extremely weak annotation, e.g., 15% sparse points, for model training, the proposed model is capable of significantly outperforming UDA methods and produces comparable performance as the supervised counterpart. The high robustness of our model shown in the validations and the low requirement of expert knowledge for sparse point annotation further improve the potential application value of our model. Code is available at https://github.com/x-coral/WDA.

show abstract

Cross-Domain Correlation Distillation for Unsupervised Domain Adaptation in Nighttime Semantic Segmentation

Cited by 59 publications

References 40 publications

Efficient Decoder and Intermediate Domain for Semantic Segmentation in Adverse Conditions

Efficient Decoder and Intermediate Domain for Semantic Segmentation in Adverse Conditions

An In-Depth Analysis of Domain Adaptation in Computer and Robotic Vision

WDA-Net: Weakly-Supervised Domain Adaptive Segmentation of Electron Microscopy

Contact Info

Product

Resources

About