SavedAdd your journal

Journals / Muş Alparslan Üniversitesi Fen Bilimleri Dergisi / 2025 / Cilt: 13 Sayı: 2

EfficientNetB4 Based Classification of Mars Surface Images: Explainability Analysis with Transfer Learning, Data Augmentation and Grad-CAM

EfficientNetB4 Tabanlı Mars Yüzeyi Görüntülerinin Sınıflandırılması: Transfer Öğrenme, Veri Artırma ve Grad-CAM ile Açıklanabilirlik Analizi

Ahmet Faruk Pala(FIRAT ÜNİVERSİTESİ)0009-0000-2985-4235
Gülşah Karaduman(FIRAT ÜNİVERSİTESİ)0000-0001-8034-3019
Erdal Özbay(FIRAT ÜNİVERSİTESİ)0000-0002-9004-4802
Pages
300–310
Publication Date
2025-12-24
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Experimental results demonstrate that the proposed model achieves high classification accuracy and enables reliable identification of geological structures through explainability analyses.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Accurate classification of geological structures on the Martian surface is of critical importance for advancing planetary science research and developing autonomous exploration systems.
📊
Key Findings & Data:In this study, a deep learning–based approach is proposed to classify images of eight different Martian geological structures, namely Other, Slope Streak, Spider, Swiss Cheese, Bright Dune, Crater, Dark Dune, and Impact Ejecta.
💡
Conclusion & Impact:Experimental results demonstrate that the proposed model achieves high classification accuracy and enables reliable identification of geological structures through explainability analyses.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:: Derin öğrenme · EfficientNetB4 · Grad-CAM Mars yüzeyi

Abstract

Accurate classification of geological structures on the Martian surface is of critical importance for advancing planetary science research and developing autonomous exploration systems. In this study, a deep learning–based approach is proposed to classify images of eight different Martian geological structures, namely Other, Slope Streak, Spider, Swiss Cheese, Bright Dune, Crater, Dark Dune, and Impact Ejecta. The Mars Terrain Classification dataset obtained from the Kaggle platform is utilized, and a transfer learning model built upon the EfficientNetB4 architecture is developed. To enhance the model performance, various data preprocessing and data augmentation techniques are applied. Furthermore, Grad-CAM (Gradient-weighted Class Activation Mapping)–based visualization methods are employed to improve the transparency and interpretability of the model’s decision-making process. Experimental results demonstrate that the proposed model achieves high classification accuracy and enables reliable identification of geological structures through explainability analyses. The findings indicate that deep learning models that are both data-efficient and interpretable can provide significant contributions to Martian surface classification, addressing an important gap in the existing literature.

Özet

Mars yüzeyinde yer alan jeolojik yapıların doğru biçimde sınıflandırılması, gezegen bilimi çalışmalarının ilerlemesi ve otonom keşif sistemlerinin geliştirilmesi açısından kritik bir öneme sahiptir. Bu çalışmada, Mars yüzeyine ait sekiz farklı jeolojik yapının (Other, Slope Streak, Spider, Swiss Cheese, Bright Dune, Crater, Dark Dune ve Impact Ejecta) görüntü tabanlı sınıflandırılması için derin öğrenme temelli bir yaklaşım önerilmektedir. Çalışmada, Kaggle platformunda sunulan Mars Terrain Classification veri seti kullanılmış ve EfficientNetB4 mimarisi üzerine inşa edilen bir transfer öğrenme modeli geliştirilmiştir. Model performansını artırmak amacıyla çeşitli veri önişleme ve veri artırma teknikleri uygulanmıştır. Ayrıca, modelin karar verme sürecinin daha iyi anlaşılabilmesi için Grad-CAM tabanlı açıklanabilirlik analizleri gerçekleştirilmiştir. Deneysel sonuçlar, önerilen yöntemin yüksek sınıflandırma başarımı sağladığını ve görselleştirme analizleri sayesinde jeolojik yapıların güvenilir biçimde ayırt edilebildiğini göstermektedir. Elde edilen bulgular, veri verimliliği yüksek ve açıklanabilir derin öğrenme modellerinin Mars yüzeyi analizinde etkili bir çözüm sunduğunu ortaya koymaktadır.

🔖

Cite This Article

Pala, A. F., Karaduman, G., Özbay, E. (2025). EfficientNetB4 Based Classification of Mars Surface Images: Explainability Analysis with Transfer Learning, Data Augmentation and Grad-CAM. Muş Alparslan Üniversitesi Fen Bilimleri Dergisi, Cilt: 13 Sayı: 2, 300–310. https://doi.org/10.18586/msufbd.1726149

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (28)

  1. A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks.” [Online]. Available: https://dl.acm.org/doi/10.1145/3065386/
  2. K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” Sep. 2014, [Online]. Available: http://arxiv.org/abs/1409.1556
  3. M. Tan and Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” May 2019, [Online]. Available: http://arxiv.org/abs/1905.11946
  4. R. R. Selvaraju, M. Cogswell, A. Das, and., “Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization,” in Proceedings of IEEE ICCV, 2017, pp. 618–626.
  5. J. Guo, X. Zhang, Y. Dong, Z. Xue, and B. Huang, “Terrain classification using mars raw images based on deep learning algorithms with application to wheeled planetary rovers,” J Terramech, vol. 108, pp. 33–38, 2023, doi: https://doi.org/10.1016/j.jterra.2023.04.002.
  6. K. Wagstaff et al., “Mars Image Content Classification: Three Years of NASA Deployment and Recent Advances,” Feb. 2021, [Online]. Available: http://arxiv.org/abs/2102.05011we
  7. M. Liu, J. Liu, and X. Ma, “MRISNet:Deep-learning-based Martian instance segmentation against blur,” Earth Sci Inform, vol. 16, no. 1, pp. 965–981, 2023, doi: 10.1007/s12145-022-00890-9.
  8. L´eon Bottou, Large-Scale Machine Learning with Stochastic Gradient Descent. Physica-Verlag HD, 2010. doi: 10.1007/978-3-7908-2604-3.
  9. C. Wang and W. Li, “A Hybrid CNN-Transformer Architecture with Frequency Domain Contrastive Learning for Image Deraining.”
  10. F. Mohammad, Y. Gao, S. Kay, R. Field, M. De Benedetti, and E. V. Ntagiou, “Deep Learning based Semantic Segmentation for Mars Rover Terrain Classification,” in 2024 International Conference on Space Robotics, iSpaRo 2024, Institute of Electrical and Electronics Engineers Inc., 2024, pp. 292–298. doi: 10.1109/iSpaRo60631.2024.10687827.
  11. T. Panambur and M. Parente, “Enhancing Martian Terrain Recognition with Deep Constrained Clustering,” Mar. 2025, [Online]. Available: http://arxiv.org/abs/2503.17633
  12. Z. Chen, B. Wu, and W. C. Liu, “Mars3dnet: Cnn‐based high‐resolution 3d reconstruction of the martian surface from single images,” Remote Sens (Basel), vol. 13, no. 5, pp. 1–22, Mar. 2021, doi: 10.3390/rs13050839.
  13. Y. Dai, T. Zheng, C. Xue, and L. Zhou, “SegMarsViT: Lightweight Mars Terrain Segmentation Network for Autonomous Driving in Planetary Exploration,” Remote Sens (Basel), vol. 14, no. 24, Dec. 2022, doi: 10.3390/rs14246297.
  14. F. Lv et al., “Highly Accurate Visual Method of Mars Terrain Classification for Rovers Based on Novel Image Features,” Entropy, vol. 24, no. 9, Sep. 2022, doi: 10.3390/e24091304.
  15. R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization,” Oct. 2016, doi: 10.1007/s11263-019-01228-7.
  16. S. U. Hassan, S. J. Abdulkadir, M. S. M. Zahid, and S. M. Al-Selwi, “Local interpretable model-agnostic explanation approach for medical imaging analysis: A systematic literature review,” Comput Biol Med, vol. 185, p. 109569, 2025, doi: https://doi.org/10.1016/j.compbiomed.2024.109569.
  17. T. Chase and K. Dantu, “MARs: Multi-view Attention Regularizations for Patch-based Feature Recognition of Space Terrain,” Oct. 2024, [Online]. Available: http://arxiv.org/abs/2410.05182
  18. H. Behzadi-Khormouji and H. Rostami, “Fast multi-resolution occlusion: a method for explaining and understanding deep neural networks,” Applied Intelligence, vol. 51, no. 4, pp. 2431–2455, 2021, doi: 10.1007/s10489-020-01946-3.
  19. S. Yang, W. Xiao, M. Zhang, S. Guo, J. Zhao, and F. Shen, “Image Data Augmentation for Deep Learning: A Survey,” Apr. 2022, [Online]. Available: http://arxiv.org/abs/2204.08610
  20. L. Perez and J. Wang, “The Effectiveness of Data Augmentation in Image Classification using Deep Learning,” Dec. 2017, [Online]. Available: http://arxiv.org/abs/1712.04621
  21. K. Wang, X. Zhang, X. Wang, and Z. Qian, “Identification of Martian minerals based on multiscale spatial-spectral fusion network,” Eng Appl Artif Intell, vol. 149, p. 110527, 2025, doi: https://doi.org/10.1016/j.engappai.2025.110527.
  22. J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, “How transferable are features in deep neural networks?,” Nov. 2014, [Online]. Available: http://arxiv.org/abs/1411.1792
  23. L. Van Der Maaten and G. Hinton, “Visualizing Data using t-SNE,” 2008.
  24. I. Jolliffe, “Principal Component Analysis,” in International Encyclopedia of Statistical Science, M. Lovric, Ed., Berlin, Heidelberg: Springer Berlin Heidelberg, 2011, pp. 1094–1096. doi: 10.1007/978-3-642-04898-2_455.
  25. D. P. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” Dec. 2014, [Online]. Available: http://arxiv.org/abs/1412.6980
  26. E. M. Rewhel et al., “Deep Learning Methods Used in Remote Sensing Images: A Review,” Apr. 01, 2023, Bilingual Publishing Group. doi: 10.30564/jees.v5i1.5232.
  27. M. T. Ribeiro, S. Singh, and C. Guestrin, “‘Why should i trust you?’ Explaining the predictions of any classifier,” in Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery, Aug. 2016, pp. 1135–1144. doi: 10.1145/2939672.2939778.
  28. R. R. Selvaraju, M. Cogswell, A. Das, and., “Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization,” in Proceedings of IEEE ICCV, 2017, pp. 618–626.