CoChat
Sign Up
Cited by 160Domain Adaptation and Few-Shot LearningAdvanced Neural Network ApplicationsAutonomous Vehicle Technology and Safety

Domain Adaptive Object Detection for Autonomous Driving under Foggy Weather

Open access
Li, Jinlong, Xu, Runsheng, Ma, Jin, Zou, Qin, Ma, Jiaqi, Yu, Hongkai·2023·2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
DOI 10.1109/wacv56688.2023.00068
Open this paper in CoChat — free accountOpen PDF

Abstract

Most object detection methods for autonomous driving usually assume a consistent feature distribution between training and testing data, which is not always the case when weathers differ significantly. The object detection model trained under clear weather might be not effective enough on the foggy weather because of the domain gap. This paper proposes a novel domain adaptive object detection framework for autonomous driving under foggy weather. Our method leverages both image-level and object-level adaptation to diminish the domain discrepancy in image style and object appearance. To further enhance the model’s capabilities under challenging samples, we also come up with a new adversarial gradient reversal layer to perform adversarial mining for the hard examples together with domain adaptation. Moreover, we propose to generate an auxiliary domain by data augmentation to enforce a new domain-level metric regularization. Experimental results on public benchmarks show the effectiveness and accuracy of the proposed method. The code is available at https://github.com/jinlong17/DA-Detect.

Integrity signals

Registry

Crossref · 10.1109/wacv56688.2023.00068

Cites

12 of 12 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Weather conditions, lighting variations, and complex traffic scenarios can affect the perception of the surrounding environment.

    “Factors such as weather conditions, lighting variations, and complex traffic scenarios can affect the perception of the surrounding environment, and addressing these challenges is crucial for reliable operation [33].”

    From Vehicle-to-Everything Communication in Intelligent Connected Vehicles: A Survey and Taxonomy (2025) · cites this paper as [33]

Cites

Flagged first, then by citations

Research works printed but not linked: 47

  1. 10.1016/j.eswa.2021.116334 · DOI
  2. 10.1109/CVPR42600.2020.01170 · DOI
  3. 10.1109/CVPR.2019.01172 · DOI
  4. Chen, Learning domain adaptive object detection with probabilistic teacher, 3040
  5. 10.1109/CVPR.2018.00352 · DOI · Paper page
  6. 10.1007/s43684-021-00010-2 · DOI
  7. 10.1109/TITS.2021.3096854 · DOI
  8. 10.1109/TCSVT.2021.3081999 · DOI
  9. Ganin, Unsupervised domain adaptation by backpropagation, 1180
  10. 10.1145/1141911.1141985 · DOI
  11. 10.1109/CVPR.2012.6248074 · DOI · Paper page
  12. 10.1109/ICCV.2015.169 · DOI · Paper page
  13. 10.1109/CVPR.2014.81 · DOI · Paper page
  14. 10.1109/TMM.2021.3082687 · DOI
  15. 10.1609/aaai.v35i2.16239 · DOI
  16. 10.1109/ICCV48922.2021.01500 · DOI
  17. 10.1109/ICCV.2019.00677 · DOI
  18. 10.1007/978-3-030-58586-0_19 · DOI
  19. Hendrycks, AugMix: A simple data processing method to improve robustness and uncertainty
  20. 10.1109/TPAMI.2020.2977911 · DOI
  21. 10.1109/TIE.2014.2321342 · DOI
  22. 10.1109/CVPR.2019.01274 · DOI · Paper page
  23. 10.1109/IV47402.2020.9304562 · DOI
  24. 10.1016/j.trc.2020.102946 · DOI · Paper page
  25. 10.1109/CVPR52688.2022.00522 · DOI
  26. 10.1109/ICCV.2017.324 · DOI · Paper page
  27. 10.1007/978-3-319-46448-0_2 · DOI · Paper page
  28. 10.1109/TIP.2020.3034487 · DOI
  29. 10.1109/IROS45743.2020.9341738 · DOI
  30. 10.1109/ICCV48922.2021.00907 · DOI
  31. 10.1007/s11263-018-1072-8 · DOI
  32. 10.1016/j.neucom.2019.08.022 · DOI
  33. 10.1007/978-3-030-58568-6_45 · DOI
  34. 10.1609/aaai.v34i07.6879 · DOI
  35. 10.1109/CVPR46437.2021.01499 · DOI
  36. 10.1109/CVPR42600.2020.01237 · DOI
  37. 10.1109/ICCV48922.2021.01516 · DOI
  38. 10.1109/ITSC48978.2021.9564825 · DOI
  39. 10.1109/ICPR48806.2021.9413070 · DOI
  40. 10.1109/CVPR46437.2021.01023 · DOI
  41. 10.1109/CVPR.2019.00283 · DOI
  42. 10.1109/TMM.2021.3078141 · DOI
  43. 10.1109/CVPR46437.2021.01224 · DOI
  44. 10.1109/JSEN.2020.2966034 · DOI
  45. 10.1007/978-3-030-58604-1_4 · DOI
  46. 10.1109/CVPR42600.2020.01378 · DOI
  47. 10.1109/CVPR52688.2022.00936 · DOI

Cited by

Sorted by citations

Authors
Li, J., Xu, R., Ma, J., Zou, Q., Ma, J., & Yu, H.
Affiliations
  • Cleveland State University (US)
  • University of California, Los Angeles (US)
  • Wuhan University (CN)
ORCID
Venue
2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
Publisher
IEEE
Published
2023-01
Pages
612-622
ISSN
2472-6737 (ISSN-L)
DOI
10.1109/wacv56688.2023.00068
OpenAlex
W4319300947
Cited by
160
References
12
Citations by year
2026: 24 · 2025: 60 · 2024: 58 · 2023: 18
FWCI
28.35
Type
conference paper
Language
en
Open access
Licence
Locations
  • IEEE Winter Conference on Applications of Computer Vision (publishedVersion)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Funders
National Science Foundation (2215388)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
Keywords
domain adaptation, object detection, autonomous driving, fog, domain gap, hard example mining, data augmentation, auxiliary domains

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review