SquareSwish-Enabled Fuel-Station Risk Mapping from Satellite Imagery
Source abstract
Abstract
This study introduces SquareSwish, a smooth, self-gated activation fx=xσx2, and benchmarks it against ten established activations (ReLU, LeakyReLU, ELU, SELU, GELU, Snake, LearnSnake, Swish, Mish, Hard-Swish) across six CNN architectures (EfficientNet-B1/B4, EfficientNet-V2-M/S, ResNet-50, and Xception) under a uniform transfer-learning protocol. Two geographically grounded datasets are used in this study. FuelRiskMap-TR comprises 7686 satellite images of urban fuel stations in Türkiye, which is semantically enriched with the OpenStreetMap context and YOLOv8-Small rooftop segmentation (mAP@0.50 = 0.724) to support AI-enabled, ICT-integrated risk screening. In a similar fashion, FuelRiskMap-UK is collected, comprising 2374 images. Risk scores are normalized and thresholded to form balanced High/Low-Risk labels for supervised training. Across identical training settings, SquareSwish achieves a top-1 validation accuracy of 0.909 on EfficientNet-B1 for FuelRiskMap-TR and reaches 0.920 when combined with SELU in a simple softmax-probability ensemble, outperforming the other activations under the same protocol. By squaring the sigmoid gate, SquareSwish more strongly attenuates mildly negative activations while preserving smooth, non-vanishing gradients, tightening decision boundaries in noisy, semantically enriched Earth-observation settings. Beyond classification, the resulting city-scale risk layers provide actionable geospatial outputs that can support inspection prioritization and integration with municipal GIS, offering a reproducible and low-cost safety-planning approach built on openly available imagery and volunteered geographic information.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Remote-Sensing Image Classification · Advanced Neural Network Applications · Automated Road and Building Extraction
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Department of Computer Engineering, Eskişehir Osmangazi University, 26480 Eskişehir, Türkiye
References (38)
Show source-provided references
1. Tan, M., and Le, Q. (2019, January 10–15). Efficientnet: Rethinking model scaling for convolutional neural networks. Proceedings of the International Conference on Machine Learning, Long Beach, CA, USA.
2. Tan, M., and Le, Q. (2021, January 18–24). Efficientnetv2: Smaller models and faster training. Proceedings of the International Conference on Machine Learning, Virtual.
3. He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27–30). Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA.10.1109/CVPR.2016.90 ↗
4. Chollet, F. (2017, January 21–26). Xception: Deep Learning With Depthwise Separable Convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.10.1109/CVPR.2017.195 ↗
5. Nair, V., and Hinton, G.E. (2010, January 21–24). Rectified linear units improve restricted boltzmann machines. Proceedings of the 27th International Conference on Machine Learning (ICML-10), Haifa, Israel.
6. Rectifier nonlinearities improve neural network acoustic models
7. Clevert, D.A., Unterthiner, T., and Hochreiter, S. (2015). Fast and accurate deep network learning by exponential linear units (elus). arXiv.
8. Klambauer, G., Unterthiner, T., Mayr, A., and Hochreiter, S. (2017, January 4–9). Self-normalizing neural networks. Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, CA, USA.
9. Hendrycks, D., and Gimpel, K. (2016). Gaussian error linear units (gelus). arXiv.
10. Ziyin, L., Hartwig, T., and Ueda, M. (2020, January 6–12). Neural networks fail to learn periodic functions and how to fix it. Proceedings of the 34th International Conference on Neural Information Processing Systems, Vancouver, BC, Canada.
11. Ramachandran, P., Zoph, B., and Le, Q.V. (2017). Searching for activation functions. arXiv.
12. Misra, D. (2019). Mish: A self regularized non-monotonic activation function. arXiv.10.5244/C.34.191 ↗
13. Howard, A., Sandler, M., Chu, G., Chen, L.C., Chen, B., Tan, M., Chu, G., Vasudevan, V., Zhu, Y., and Pang, R. (November, January 27). Searching for mobilenetv3. Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, Republic of Korea.
14. Efficient activation functions for embedded inference engines10.1016/j.neucom.2021.02.030 ↗
15. Ma, N., Zhang, X., Liu, M., and Sun, J. (2021, January 20–25). Activate or not: Learning customized activation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, TN, USA.10.1109/CVPR46437.2021.00794 ↗
16. Metaformer baselines for vision10.1109/TPAMI.2023.3329173 ↗
17. A survey on modern trainable activation functions10.1016/j.neunet.2021.01.026 ↗
18. Activation functions in deep learning: A comprehensive survey and benchmark10.1016/j.neucom.2022.06.111 ↗
19. Kunc, V., and Kléma, J. (2024). Three decades of activations: A comprehensive survey of 400 activation functions for neural networks. arXiv.
20. Fang, H., Lee, J.U., Moosavi, N.S., and Gurevych, I. (2023, January 2–6). Transformers with learnable activation functions. Proceedings of the Findings of the Association for Computational Linguistics: EACL, Dubrovnik, Croatia.10.18653/v1/2023.findings-eacl.181 ↗
21. Huang, A.H., and Schlag, I. (2024). Deriving Activation Functions Using Integration. arXiv.
22. Pugliese Viloria, A.d.J., Folini, A., Carrion, D., and Brovelli, M.A. (2024). Hazard susceptibility mapping with machine and deep learning: A literature review. Remote Sens., 16.10.3390/rs16183374 ↗
23. Spatial landslide susceptibility assessment using machine learning techniques assisted by additional data created with generative adversarial networks10.1016/j.gsf.2020.09.002 ↗
24. Jokar Arsanjani, J., Zipf, A., Mooney, P., and Helbich, M. (2015). An introduction to OpenStreetMap in Geographic Information Science: Experiences, research, and applications. OpenStreetMap in GIScience: Experiences, Research, and Applications, Springer.10.1007/978-3-319-14280-7 ↗
25. Li, X., Song, L., Liu, L., and Zhou, L. (2021). GSS-RiskAsser: A Multi-Modal Deep-Learning Framework for Urban Gas Supply System Risk Assessment on Business Users. Sensors, 21.10.3390/s21217010 ↗
26. Karakucs, O., and Corcoran, P. (2025). A Multi-Modal Spatial Risk Framework for EV Charging Infrastructure Using Remote Sensing. arXiv.
27. Integrated machine learning methods with resampling algorithms for flood susceptibility prediction10.1016/j.scitotenv.2019.135983 ↗
28. He, K., Gkioxari, G., Dollár, P., and Girshick, R. (2017, January 22–29). Mask r-cnn. Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy.10.1109/ICCV.2017.322 ↗
29. Central attention network for hyperspectral imagery classification10.1109/TNNLS.2022.3155114 ↗
30. Multiarea target attention for hyperspectral image classification
31. Seghsi: Semantic segmentation of hyperspectral images with limited labeled pixels10.1109/TIP.2024.3492724 ↗
32. Bommasani, R., Hudson, D.A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M.S., Bohg, J., Bosselut, A., and Brunskill, E. (2021). On the opportunities and risks of foundation models. arXiv.
33. Long, J., Shelhamer, E., and Darrell, T. (2015, January 7–12). Fully convolutional networks for semantic segmentation. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA.10.1109/CVPR.2015.7298965 ↗
34. Data-driven accident consequence assessment on urban gas method network based on machine learning10.1016/j.ress.2021.108216 ↗
35. Ma, H., Liu, Y., Ren, Y., Wang, D., Yu, L., and Yu, J. (2020). Improved CNN classification method for groups of buildings damaged by earthquake, based on high resolution remote sensing images. Remote Sens., 12.10.3390/rs12020260 ↗
36. OpenStreetMap Contributors (2025, December 21). Copyright and License. Available online: https://www.openstreetmap.org/copyright.
37. (2020). Codes for the Representation of Names of Countries and Their Subdivisions—Part 1: Country Code (Standard No. ISO 3166-1).
38. (2025, December 21). Esri. ArcGIS World Imagery (ArcGIS Online Basemap Service). Available online: https://services.arcgisonline.com/ArcGIS/rest/services/World_Imagery/MapServer.