Deepa Mane; Sandhya Arora

Cross-Domain Outdoor Scene Understanding Using PC-GAN and TL-SZKSE-DenseNet in Adverse Weather Conditions

PDF (1825KB), PP.140-162

Views: 0 Downloads: 0

Author(s)

Deepa Mane ^1,2,* Sandhya Arora ³

1. Department of Computer Engineering, Smt. Kashibai Navale College of Engineering, SPPU, Pune, India

2. Department of Artificial Intelligence and Data Science, Pune Institute of Computer Technology, SPPU, Pune, India

3. Department of Computer Engineering, MKSSS's Cummins college of Engineering for women, SPPU, Pune, India

* Corresponding author.

DOI: https://doi.org/10.5815/ijieeb.2026.05.09

Received: 1 Nov. 2025 / Revised: 1 Jan. 2026 / Accepted: 29 Jun. 2026 / Published: 8 Oct. 2026

Index Terms

Outdoor Scene, Autonomous Driving, Cross Domain, Multi-Scale Analysis, Object Detection, Transfer Learning, and Deep Learning.

Abstract

In autonomous driving, Outdoor Scene (OS) understanding in unfavorable weather conditions is pivotal for enabling safe navigation. Conventional studies failed to capture the intricate pairwise object relationship within scenes in OS classification. To mitigate these challenges, this research introduces a dual framework bringing together a Panoptic Cycle-Generative Adversarial Network (PC-GAN) with a Transfer Learning-based Squeeze-and-Zero-Knowledge Spatial Excitation DenseNet (TL-SZKSE-DenseNet). The suggested PC-GAN achieves realistic cross-weather image translation with semantically aligned representations in clear and degraded scenes. The TL-SZKSE-DenseNet improves feature learning with the addition of a zero-knowledge spatial excitation mechanism that selectively enhances weather-independent and context-dependent features for enhanced discrimination in crowded scenes. This generative–discriminative hybrid method effectively overcomes structured and unstructured driving domain gaps, as demonstrated on benchmark datasets such as BDD and IndiaScene365. Primarily, the OS images are generalized to adverse conditions using Panoptic Cycle-Generative Adversarial Network (PC-GAN). After that, using saliency mapping, the most significant regions in the generalized output are highlighted. In the meantime, the significant features are extracted utilizing the Orangutan Optimization Algorithm (OOA). In addition, from the detected objects, the Object Frequency and Relationship (OFR) is analyzed. Later, using TL-SZKSE-DenseNet, the OS are classified based on the OFR and the selected features. The performance of our proposed approach was tested on our newly developed IndiaScene365 dataset as well as various benchmark datasets such as BDD100k. Our newly developed dataset, IndiaScene365, includes 3000 images that entail three different weather conditions, which are foggy weather, rainy weather, and poor light conditions along with clear day weather. The proposed approach records 94.87% accuracy, 93.24%, and 94.18% sensitivity on our dataset.

Cite This Paper

Deepa Mane, Sandhya Arora, "Cross-Domain Outdoor Scene Understanding Using PC-GAN and TL-SZKSE-DenseNet in Adverse Weather Conditions", International Journal of Information Engineering and Electronic Business(IJIEEB), Vol.18, No.5, pp. 140-162, 2026. DOI:10.5815/ijieeb.2026.05.09

Reference

[1]Aloufi, N., Alnori, A., & Basuhail, A. (2024). Enhancing Autonomous Vehicle Perception in Adverse Weather: A Multi Objectives Model for Integrated Weather Classification and Object Detection. Electronics (Switzerland), 13, 1–20. https://doi.org/10.3390/electronics13153063
[2]Chow, T. Y., Lee, K. H., & Chan, K. L. (2023). Detection of Targets in Road Scene Images Enhanced Using Conditional GAN-Based Dehazing Model. Applied Sciences (Switzerland), 13, 1–26. https://doi.org/10.3390/app13095326
[3]Dreissig, M., Scheuble, D., Piewak, F., & Boedecker, J. (2023). Survey on LiDAR Perception in Adverse Weather Conditions. Arxiv, 1–8. https://doi.org/10.1109/IV55152.2023.10186539
[4]Guan, D., Huang, J., Xiao, A., Lu, S., & Cao, Y. (2022). Uncertainty-Aware Unsupervised Domain Adaptation in Object Detection. IEEE Transactions on Multimedia, 24, 2022. https://doi.org/10.1109/TMM.2021.3082687
[5]Humayun, M., Ashfaq, F., Jhanjhi, N. Z., & Alsadun, M. K. (2022). Traffic Management: Multi-Scale Vehicle Detection in Varying Weather Conditions Using YOLOv4 and Spatial Pyramid Pooling Network. Electronics (Switzerland), 11, 1–29. https://doi.org/10.3390/electronics11172748
[6]Iqbal, J., Hafiz, R., & Ali, M. (2022). FogAdapt: Self-supervised domain adaptation for semantic segmentation of foggy images. Arxiv, 1–13. https://doi.org/10.1016/j.neucom.2022.05.086
[7]Jhung, J., & Kim, S. (2021). Behind-the-scenes (Bts): Wiper-occlusion canceling for advanced driver assistance systems in adverse rain environments. Sensors, 21, 1–21. https://doi.org/10.3390/s21238081
[8]Kumar, D., & Muhammad, N. (2023). Object Detection in Adverse Weather for Autonomous Driving through Data Merging and YOLOv8. Sensors, 23, 1–29. https://www.mdpi.com/1424-8220/23/20/8471
[9]Li, J., Xu, R., Liu, X., Ma, J., Li, B., Zou, Q., Ma, J., & Yu, H. (2025). Domain Adaptation Based Object Detection for Autonomous Driving in Foggy and Rainy Weather. IEEE Transactions on Intelligent Vehicles, 10(2), 2025. https://doi.org/10.1109/TIV.2024.3419689
[10]Liao, H., Shen, H., Li, Z., Wang, C., Li, G., Bie, Y., & Xu, C. (2024). GPT-4 enhanced multimodal grounding for autonomous driving: Leveraging cross-modal attention with large language models. Communications in Transportation Research, 4, 1–19. https://doi.org/10.1016/j.commtr.2023.100116
[11]Lin, H., Parsi, A., Mullins, D., Horgan, J., Ward, E., Eising, C., Denny, P., Deegan, B., Glavin, M., & Jones, E. (2024). A Study on Data Selection for Object Detection in Various Lighting Conditions for Autonomous Vehicles. Journal of Imaging, 10(7), 1–18. https://doi.org/10.3390/jimaging10070153
[12]Lin, H. Y., Huang, Y. C., Lai, J. X., & You, T. T. (2025). Domain Adaptation for Vehicle Detection Under Adverse Weather. IEEE Open Journal of Intelligent Transportation Systems, 6, 1–11. https://doi.org/10.1109/OJITS.2025.3563373
[13]Liu, B., Jin, J., Zhang, Y., & Sun, C. (2025). WRRT-DETR : Weather-Robust RT-DETR for Drone-View Object Detection in Adverse Weather. Drones, 9, 1–21. https://www.mdpi.com/2504-446X/9/5/369
[14]Mo, T., Zheng, S., Chan, W. Y., & Yang, R. (2025). Review of AI Image Enhancement Techniques for In-Vehicle Vision Systems Under Adverse Weather Conditions. World Electric Vehicle Journal, 16(2), 1–20. https://doi.org/10.3390/wevj16020072
[15]Nawaz, M., Khan, S., Daud, M., Asim, M., Anwar, G. A., Shahid, A. R., Ho, H. P. A., Chan, T., Pak Kong, D., & Yuan, W. (2025). Improving Autonomous Vehicle Cognitive Robustness in Extreme Weather With Deep Learning and Thermal Camera Fusion. IEEE Open Journal of Vehicular Technology, 6, 1–16. https://doi.org/10.1109/OJVT.2025.3529495
[16]Ni, H., & Su, W. (2025). A Self-Attention CycleGAN for Unsupervised Image Hazing. Big Data and Cognitive Computing, 9(4), 1–16. https://doi.org/10.3390/bdcc9040096
[17]Ogunrinde, I., & Bernadin, S. (2023). Deep Camera–Radar Fusion with an Attention Framework for Autonomous Vehicle Vision in Foggy Weather Conditions. Sensors, 23(14), 1–27. https://doi.org/10.3390/s23146255
[18]Osipov, A., Pleshakova, E., Gataullin, S., Finogeev, A., & Yadav, V. (2022). Deep Learning Method for Recognition and Classification of Images from Video Recorders in Difficult Weather Conditions. Sustainability (Switzerland), 14(4), 1–16. https://doi.org/10.3390/su14042420
[19]Patil, S. M., Pawar, S. D., Mhatre, S. N., & Kharade, P. A. (2024). Yolov4-based hybrid feature enhancement network with robust object detection under adverse weather conditions. Signal, Image and Video Processing, 18, 1–31. https://doi.org/10.1007/s11760-024-03068-6
[20]Piroli, A., Dallabetta, V., Kopp, J., Walessa, M., Meissner, D., & Dietmayer, K. (2023). Energy-Based Detection of Adverse Weather Effects in LiDAR Data. Arxiv, 1–8. https://doi.org/10.1109/LRA.2023.3282382
[21]Rao, Z., Guo, J., Tang, L., Huang, Y., Ding, X., & Guo, S. (2025). SRCD: Semantic Reasoning With Compound Domains for Single-Domain Generalized Object Detection. IEEE Transactions on Neural Networks and Learning Systems, 36(7), 2025. https://doi.org/10.1109/TNNLS.2024.3480120
[22]Roy, S. D., & Bhowmik, M. K. (2022). AWDMC-Net: Classification of Adversarial Weather Degraded Multiclass scenes using a Convolution Neural Network. Computer Vision and Image Understanding, 222, 1–19. https://doi.org/10.1016/j.cviu.2022.103498
[23]Sharma, T., Debaque, B., Duclos, N., Chehri, A., Kinder, B., & Fortier, P. (2022). Deep Learning-Based Object Detection and Scene Perception under Bad Weather Conditions. Electronics (Switzerland), 11, 1–11. https://doi.org/10.3390/electronics11040563
[24]Siddiqua, M., Brahim Belhaouari, S., Akhter, N., Zameer, A., & Khurshid, J. (2023). MACGAN: An All-in-One Image Restoration under Adverse Conditions Using Multidomain Attention-Based Conditional GAN. IEEE Access, 11, 70482–70502. https://doi.org/10.1109/ACCESS.2023.3289591
[25]Tahir, N. U. A., Zhang, Z., Asim, M., Chen, J., & ELAffendi, M. (2024). Object Detection in Autonomous Vehicles under Adverse Weather: A Review of Traditional and Deep Learning Approaches. Algorithms, 17, 1–36. https://doi.org/10.3390/a17030103
[26]Thanh, P. T. H., Tran, T. T., Nguyen, T. H., Nguyen, M. H. V., Pham, T. V., Duy, T. V. T., & Nguyen, D. D. (2025). ULDepth: Transform Self-supervised Depth Estimation to Unpaired Multi-domain Learning. IEEE Open Journal of Signal Processing, 1–13. https://doi.org/10.1109/OJSP.2025.3597873
[27]Wang, M., Su, T., Chen, S., Yang, W., Liu, J., & Wang, Z. (2023). Automatic Model-Based Dataset Generation for High-Level Vision Tasks of Autonomous Driving in Haze Weather. IEEE Transactions on Industrial Informatics, 19(8), 9071–9081. https://doi.org/10.1109/TII.2022.3224958
[28]Yang, H., Carballo, A., Zhang, Y., & Takeda, K. (2023). Framework for Generation and Removal of Multiple Types of Adverse Weather from Driving Scene Images. Sensors, 23(3), 1–16. https://doi.org/10.3390/s23031548
[29]Yuan, Y., Wei, Y., Guo, Y., Chen, J., & Jiang, T. (2025). CME-YOLO: A Cross-Modal Enhanced YOLO Algorithm for Adverse Weather Object Detection in Autonomous Driving. Big Data and Cognitive Computing, 9, 1–32. https://doi.org/10.3390/bdcc9040092
[30]H. C. Shin et al., ‘Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning’, IEEE Trans Med Imaging, vol. 35, no. 5, pp. 1285–1298, May 2016, doi: 10.1109/TMI.2016.2528162.
[31]S. Di, H. Zhang, C. G. Li, X. Mei, D. Prokhorov, and H. Ling, ‘Cross-Domain Traffic Scene Understanding: A Dense Correspondence-Based Transfer Learning Approach’, IEEE Transactions on Intelligent Transportation Systems, vol. 19, no. 3, pp. 745–757, Mar. 2018, doi: 10.1109/TITS.2017.2702012.
[32]A. Krizhevsky, I. Sutskever, and G. E. Hinton, ‘ImageNet Classification with Deep Convolutional Neural Networks’, Commun ACM, vol. 60, no. 6, 2017, doi: 10.1145/3065386.
[33]D. Dai, C. Sakaridis, S. Hecker, and L. van Gool, ‘Curriculum Model Adaptation with Synthetic and Real Data for Semantic Foggy Scene Understanding’, Int J Comput Vis, vol. 128, no. 5, pp. 1182–1204, May 2020, doi: 10.1007/s11263-019-01182-4.
[34]J. Lu, M. Xu, R. Yang, and Z. Wang, ‘Understanding and Predicting the Memorability of Outdoor Natural Scenes’, IEEE Transactions on Image Processing, vol. 29, pp. 4927–4941, 2020, doi: 10.1109/TIP.2020.2975957.
[35]H. Zhou, J. Ma, C. C. Tan, Y. Zhang, and H. Ling, ‘Cross-Weather Image Alignment via Latent Generative Model with Intensity Consistency’, IEEE Transactions on Image Processing, vol. 29, pp. 5216–5228, 2020, doi: 10.1109/TIP.2020.2980210.
[36]M. Milford et al., ‘Condition-invariant, top-down visual place recognition’, in 2014 IEEE International Conference on Robotics and Automation (ICRA), May 2014, pp. 5571–5577. doi: 10.1109/ICRA.2014.6907678.
[37]P. Soviany, R. T. Ionescu, P. Rota, and N. Sebe, ‘Curriculum self-paced learning for cross-domain object detection’, Computer Vision and Image Understanding, vol. 204, Mar. 2021, doi: 10.1016/j.cviu.2021.103166.
[38]Y. Yang, H. Dong, G. Liu, L. Zhang, and L. Li, ‘Cross-Domain Traffic Scene Understanding by Integrating Deep Learning and Topic Model’, Comput Intell Neurosci, vol. 2022, 2022, doi: 10.1155/2022/8884669.
[39]C. Song, J. Wu, L. Zhu, M. Zhang, and H. Ling, ‘Nighttime Road Scene Parsing by Unsupervised Domain Adaptation’, IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 4, pp. 3244–3255, Apr. 2022, doi: 10.1109/TITS.2020.3033569.
[40]C. Zhang, W. Ding, G. Peng, F. Fu, and W. Wang, ‘Street View Text Recognition with Deep Learning for Urban Scene Understanding in Intelligent Transportation Systems’, IEEE Transactions on Intelligent Transportation Systems, vol. 22, no. 7, pp. 4727–4743, Jul. 2021, doi: 10.1109/TITS.2020.3017632.
[41]M. Naseer, S. H. Khan, and F. Porikli,‘Indoor Scene Understanding in 2.5/3D for Autonomous Agents: A Survey’, vol. 7, 2019, doi: 10.1109/ACCESS.2018.2886133.
[42]C. Ju, D. Gao, R. Mane, B. Tan, Y. Liu, and C. Guan, Federated Transfer Learning for EEG Signal Classification, in Proc. IEEE EMBC, 2020, pp. 3040–3045. doi: 10.1109/EMBC44109.2020.9175344.
[43]U. Nadeem, S. A. A. Shah, F. Sohel, R. Togneri, and M. Bennamoun, ‘Deep learning for scene understanding’, in Smart Innovation, Systems and Technologies, vol. 136, Springer Science and Business Media Deutschland GmbH, 2019, pp. 21–51. doi: 10.1007/978-3-030-11479-4_2.
[44]X. Li et al., ‘Yolov3-Pruning(transfer): real-time object detection algorithm based on transfer learning’, J Real Time Image Process, Aug. 2022, doi: 10.1007/s11554-022-01227-x.
[45]K. Weiss, T. M. Khoshgoftaar, and D. D. Wang, ‘A survey of transfer learning’, Journal of Big Data, vol. 3, no. 1, Dec. 2016, doi: 10.1186/s40537-016-0043-6.
[46]Jeff Donahue, Yangqing Jia, Oriol Vinyals, Judy Hoffman, Ning Zhang, Eric Tzeng, and Trevor Darrell. ‘DeCAF: a deep convolutional activation feature for generic visual recognition’, In Proceedings of the 31st International Conference on International Conference on Machine Learning - Volume 32 (ICML'14) 2014.
[47]Torrey, L. and Shavlik, J. ‘Transfer Learning’ Handbook of Research on Machine Learning Applications and Trends: Algorithms, Methods, and Techniques, IGI Global, Hershey, 242-264 (2010) https://doi.org/10.4018/978-1-60566-766-9.ch011.
[48]P. Ganesh Pawar and V. Devendran, ‘Scene Understanding: A Survey to See the World at a Single Glance’, 2nd International Conference on Intelligent Communication and Computational Techniques (ICCT), 2019.
[49]M. Long, Y. Cao, Z. Cao, J. Wang, and M. I. Jordan, ‘Transferable Representation Learning with Deep Adaptation Networks’, IEEE Trans Pattern Anal Mach Intell, vol. 41, no. 12, pp. 3071–3085, Dec. 2019, doi: 10.1109/TPAMI.2018.2868685.
[50]A. Chakraborty, Cosmin Anitescu, Xiaoying Zhuang, and T. Rabczuk, Domain adaptation-based transfer learning approach for solving PDEs on complex geometries’, vol. 1, p. 3, doi: 10.1007/s00366-022-01661-2.
[51]S. J. Pan and Q. Yang, ‘A survey on transfer learning’, IEEE Transactions on Knowledge and Data Engineering, vol. 22, no. 10. pp. 1345–1359, 2010. doi: 10.1109/TKDE.2009.191.
[52]D. P. Thesis, ‘Transfer Learning for Object Category Detection’.
[53]M. Afif, R. Ayachi, Y. Said, and M. Atri, ‘Deep Learning Based Application for Indoor Scene Recognition’, Neural Process Lett, vol. 51, no. 3, pp. 2827–2837, Jun. 2020, doi: 10.1007/s11063-020-10231-w.
[54]R. Xiao, Y. Wang, and C. Tao, ‘Fine-Grained Road Scene Understanding from Aerial Images Based on Semisupervised Semantic Segmentation Networks’, IEEE Geoscience and Remote Sensing Letters, vol. 19, 2022, doi: 10.1109/LGRS.2021.3059708.
[55]Ge, F., Zhang, Y., Wang, L., Tan, Y., Coleman, S., & Kerr, D. (2025). MDAG-Net: Multi-Domain Association-Guided Network for Image-based Long-term Visual Localization. IEEE Transactions on Cognitive and Developmental Systems, 1–15. https://doi.org/10.1109/TCDS.2024.3525180.
[56]J. Lu, N. Verma, and N. K. Jha, ‘Convolutional Autoencoder-Based Transfer Learning for Multi-Task Image Inferences’, IEEE Trans Emerg Top Comput, vol. 10, no. 2, pp. 1045–1057, 2022, doi: 10.1109/TETC.2021.3068063.
[57]F. Husain, B. Dellen, and C. Torras, ‘Scene Understanding Using Deep Learning’, Handbook of Neural Computation, pp. 373–382, Jan. 2017, doi: 10.1016/B978-0-12-811318-9.00020-X.
[58]S. Safavi and M. Jalali, ‘DeePOF: A hybrid approach of deep convolutional neural network and friendship to Point-of-Interest (POI) recommendation system in location-based social networks’, Concurr Comput, vol. 34, no. 15, Jul. 2022.
[59]Mane D, Arora S and Shelke S (2026) IndiaScene365: a transfer learning dataset for Indian scene understanding in diverse weather condition. Front. Artif. Intell. 8:1669512. doi: 10.3389/frai.2025.1669512
[60]Mane, D., Arora, S., Shelke, S. (2025). Target-Aware Generative Adversarial Network for Domain Adaptive Yolo in Cross Weather. International conf. Integration of Artificial Intelligence in IoT: Opportunities and Challenges.
[61]BDD Dataset Link: https://www.kaggle.com/datasets/awsaf49/bdd100k-dataset/data
[62]IndiaScene365 Dataset Link: https://data.mendeley.com/datasets/pwffhg6nhz/2

International Journal of Information Engineering and Electronic Business (IJIEEB)