Unsupervised Domain Adaptation Object Detection Based on Pseudo-Label Purification in Rainy and Foggy Weather
DOI:
https://doi.org/10.66421/int.j.datasci.intell.syst25Keywords:
Unsupervised Domain Adaptation, Object Detection, Pseudo-Label Purification, Adverse Weather Conditions, Rainy Weather Detection, Foggy Weather Detection, Computer VisionAbstract
The deployment of autonomous driving systems and intelligent surveillance networks relies heavily on robust object detection algorithms. While significant strides have been made using deep learning techniques on clear weather datasets, performance degrades precipitously when these models are applied to adverse weather conditions such as rain and fog. This degradation arises primarily from the domain shift between the training data, which typically consists of high-quality clear images, and the testing data, which suffers from visibility impediments and visual artifacts. Unsupervised Domain Adaptation offers a solution by transferring knowledge from a labeled source domain to an unlabeled target domain; however, standard self-training approaches in this field often suffer from error propagation due to noisy pseudo-labels generated on the target data. This paper proposes a novel framework for Unsupervised Domain Adaptation Object Detection specifically tailored for rainy and foggy weather. We introduce a Pseudo-Label Purification mechanism that refines the self-training process by dynamically filtering unreliable predictions based on uncertainty estimation and temporal consistency. By mitigating the influence of false positives caused by rain streaks and fog haze, our method ensures that the model learns domain-invariant features rather than overfitting to weather-specific noise. Extensive experiments on standard benchmarks demonstrate that our approach achieves superior detection accuracy compared to state-of-the-art adaptation methods
References
Zeng, G., Zhang, X., Wang, Z., Xu, H., Chen, Z., Li, B., & Tu, Z. (2025). Yolo-count: Differentiable object counting for text-to-image generation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 16765-16775).
Huang, Y., Zhang, K., Wang, Y., Du, D., Yuan, Y., & Zhao, Z. (2025, June). Enhancing Open-Vocabulary Panoptic Segmentation with Semantic-Guided Q-Tuning. In 2025 IEEE International Conference on Multimedia and Expo (ICME) (pp. 1-6). IEEE.
Ma, F., Chai, J., & Wang, H. (2019). Two-dimensional compact variational mode decomposition-based low-light image enhancement. IEEE Access, 7, 136299-136309.
Martinez, N. N., Lee, Y., Eck, J. E., & O, S. (2017). Ravenous wolves revisited: A systematic review of offending concentration. Crime Science, 6(1), 10.
Fan, J., Liang, W., & Zhang, W. Q. (2025). SARNet: A Spike-Aware consecutive validation Framework for Accurate Remaining Useful Life Prediction. arXiv preprint arXiv:2510.22955.
Zou, Y., & Yin, Z. (2025). Alignment, mining and fusion: Representation alignment with hard negative mining and selective knowledge fusion for medical visual question answering. In Proceedings of the Computer Vision and Pattern Recognition Conference (pp. 29623-29633).
Zhang, K., Zhao, S., Zeng, H., & Chen, J. (2025). Two-Stage archive evolutionary algorithm for constrained Multi-Objective optimization. Mathematics, 13(3), 470.
Lyu, M. R., Ray, B., Roychoudhury, A., Tan, S. H., & Thongtanunam, P. (2025). Automatic programming: Large language models and beyond. ACM Transactions on Software Engineering and Methodology, 34(5), 1-33.
Zhu, D., Xie, C., Wang, Z., & Zhang, H. (2025). RaX-Crash: A Resource Efficient and Explainable Small Model Pipeline with an Application to City Scale Injury Severity Prediction. arXiv preprint arXiv:2512.07848.
Hu, Q., Peng, Y., KinTak, U., & Chen, J. (2025). MSFusion: A Degradation-Correctable Framework for Robust Infrared and Visible Image Fusion. IEEE Sensors Journal, 26(2), 2749-2766.
Li, J., Chai, W., Fu, X., Xu, H., & Xie, S. (2025). Science-t2i: Addressing scientific illusions in image synthesis. In Proceedings of the Computer Vision and Pattern Recognition Conference (pp. 2734-2744).
Hu, Q., Peng, Y., Zhang, C., Lin, Y., U, K., & Chen, J. (2025). Building Instance Extraction via Multi-Scale Hybrid Dual-Attention Network. Buildings, 15(17), 3102.
Zhou Z, Leng N, Ma H, et al. Study on Real-Time Data Analysis and Intelligent Forecasting Methods for Integrated Circuit Supply Chains Based on Cloud Computing[C]//Proceedings of the 2nd International Symposium on Integrated Circuit Design and Integrated Systems. 2025: 245-250.
Qiu, S., Wang, H., Zhang, Y., Ke, Z., & Li, Z. (2025). Convex optimization of Markov decision processes based on Z transform: A theoretical framework for two-space decomposition and linear programming reconstruction. Mathematics, 13(11), 1765.
Wang, R., Guo, T., Li, Y., Meng, D., & Liang, B. (2025). Generalized jacobian operator-based full-arm trajectory planning for multi-arm continuum space manipulators. Aerospace Science and Technology, 111559.
Liu, F., Wang, J., Tian, J., Zhuang, D., Miranda-Moreno, L., & Sun, L. (2022). A universal framework of spatiotemporal bias block for long-term traffic forecasting. IEEE Transactions on Intelligent Transportation Systems, 23(10), 19064-19075.
Ma, F., Liu, L., & Cheng, H. V. (2024). TIMA: Text-Image Mutual Awareness for Balancing Zero-Shot Adversarial Robustness and Generalization Ability. arXiv preprint arXiv:2405.17678.
Ke, Z., Cao, Y., Chen, Z., Yin, Y., He, S., & Cheng, Y. (2025). Early warning of cryptocurrency reversal risks via multi-source data. Finance Research Letters, 107890.
Wang, Y., Zhang, R., & Liu, J. (2023). RLS-DTS: Reinforcement-learning linguistic steganalysis in distribution-transformed scenario. IEEE Signal Processing Letters, 30, 1232-1236.
Wang, Y., Song, R., Li, L., Tang, Y., Zhang, R., & Liu, J. (2025). User profile constructed by multiple attributes for optimizing linguistic steganalysis in social networks. Expert Systems with Applications, 129311.
Li, Y., Zou, Y., He, X., Xu, Q., Liu, M., Jin, S., ... & Zhang, J. (2025). HFA-UNet: Hybrid and full attention UNet for thyroid nodule segmentation. Knowledge-Based Systems, 114245.
Ma, Y., Qu, D., & Pyrozhenko, M. (2026). Bio-RegNet: A Meta-Homeostatic Bayesian Neural Network Framework Integrating Treg-Inspired Immunoregulation and Autophagic Optimization for Adaptive Community Detection and Stable Intelligence. Biomimetics, 11(1), 48.
Tang, Y., Kojima, K., Gotoda, M., Nishikawa, S., Hayashi, S., Koike-Akino, T., ... & Klamkin, J. (2020). InP grating coupler design for vertical coupling of InP and silicon chips. Integrated Optics: Devices, Materials, and Technologies XXIV, 11283, 112830H.
Liu, F., & Liu, C. (2018, June). Towards accurate and high-speed spiking neuromorphic systems with data quantization-aware deep networks. In Proceedings of the 55th Annual Design Automation Conference (pp. 1-6).
Liang, L., Chen, J., Shi, J., Zhang, K., & Zheng, X. (2025). Noise-Robust image edge detection based on multi-scale automatic anisotropic morphological Gaussian Kernels. PLoS One, 20(5), e0319852.
Hu, Q., Peng, Y., Shao, Z., & Chen, J. (2026). Scene degradation-aware fusion network for robust infrared and visible image synthesis in extreme conditions. The Visual Computer, 42(1), 48.
Li, Z., Zhang, Y., Pan, T., Sun, Y., Duan, Z., Fang, J., ... & Wang, J. (2025, July). FocusLLM: Precise understanding of long context by dynamic condensing. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (pp. 31087-31101).
Ahmad, N. R. (2025). Leadership styles in the 21st century: A comparative analysis of transformational and transactional leadership. Journal for Social Science Archives, 3(1), 576–587. https://doi.org/10.59075/jssa.v3i1.142
Ahmad, N. R. (2025). Corporate social responsibility in the digital age: Navigating ethical marketing and consumer expectations. Research Journal of Psychology, 3(1), 287–297. https://doi.org/10.59075/rjs.v3i1.66
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Paul Wagner, Lucas Dubois

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
All articles are published under the Creative Commons Attribution 4.0 (CC BY 4.0) License. Users may share, copy, adapt, and distribute the work for any purpose, provided proper credit is given to the original author(s) and source. Authors retain copyright.
