Welcome to the GitHub repository for our study on Deep Semi-Supervised Learning (DSSL) applied to Remote Sensing Scene Classification. This repository contains configuration files for reproducibility and a comprehensive implementation split across four Colab notebooks. These notebooks cover data and software acquisition, model training, accuracy evaluation, and inference, showcasing the effectiveness of DSSL with limited labeled data.
This repository presents a comparative study of various DSSL methods, including FixMatch [1], CoMatch [2], and Class Aware Contrastive Semi-Supervised Learning (CCSSL) [3], on two remote sensing datasets: UCM [4] and AID [5]. By leveraging a small number of labeled examples alongside unlabeled data, these methods demonstrate their capability to significantly improve classification accuracy.
- Configuration files adapted from Classification-SemiCLS GitHub [3] for reproducibility and ease of experimentation. They provide the details of the experiments, including dataset splits, augmentations, and training settings.
- Logs of the experimental results showcasing the performance of DSSL methods compared to supervised benchmarks. The compressed version of all the logs can be downloaded from google drive here.
[1] K. Sohn, D. Berthelot, N. Carlini, Z. Zhang, H. Zhang, C. A. Raffel, E. D. Cubuk, A. Kurakin, and C. Li, "FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence," in Advances in Neural Information Processing Systems, vol. 33, H. Larochelle et al. (eds.), Curran Associates, Inc., 2020, pp. 596-608.
[2] J. Li, C. Xiong, and S. C. H. Hoi, "CoMatch: Semi-supervised Learning with Contrastive Graph Regularization," in Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021, pp. 9455-9464. doi: 10.1109/ICCV48922.2021.00934.
[3] F. Yang, K. Wu, S. Zhang, G. Jiang, Y. Liu, F. Zheng, W. Zhang, C. Wang, and L. Zeng, "Class-Aware Contrastive Semi-Supervised Learning," in Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, pp. 14401-14410. doi: 10.1109/CVPR52688.2022.01402.
[4] Y. Yang and S. Newsam, "Bag-of-Visual-Words and Spatial Extensions for Land-Use Classification," in Proceedings of the 18th SIGSPATIAL International Conference on Advances in Geographic Information Systems (GIS '10), ACM, 2010, pp. 270-279. doi: 10.1145/1869790.1869829. URL: https://doi.org/10.1145/1869790.1869829.
[5] G.-S. Xia, J. Hu, F. Hu, B. Shi, X. Bai, Y. Zhong, L. Zhang, and X. Lu, "AID: A Benchmark Data Set for Performance Evaluation of Aerial Scene Classification," IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 7, pp. 3965-3981, 2017. doi: 10.1109/TGRS.2017.2685945.