Performance of Convolutional Neural Network for Classifying Soil Moisture Level based on In-Situ RGB Soil Surface Images
Abstract
Computer vision offers a promising method for soil moisture assessment especially for real-time field monitoring where sensor-based measurements are limited. This study evaluates the performance of a traditional Convolutional Neural Network (CNN) and ResNet-50 architecture for classifying soil moisture levels directly from in-situ surface images. The research involved 200 field-captured images and corresponding moisture data from a rainfed agricultural area. The models were trained with datasets grouped into two, three, and four moisture categories to test performance under varying complexity. The results showed poor model performance, characterized by high instability and severe overfitting across all experiments. Model accuracy for the traditional CNN significantly decreased from 0.513 to 0.256 as the number of classification categories increased and from 0.487 to 0.205 for ResNet-50. High RMSE values from 0.433 to 0.507 further confirmed substantial prediction errors. This finding highlights the limitation of RGB-based in-situ imagery for soil moisture classification, where environmental variability dominates the visual signal. It also suggests that soil moisture-related features are not sufficiently distinguishable under uncontrolled field conditions. The study concludes that the high variability of direct field images due to factors like inconsistent lighting, illumination, and the presence of non-soil objects is a primary obstacle to accurate classification. Future studies should implement advanced pre-processing techniques such as segmentation to reduce illumination noise.
Downloads
References
Alaoui, A., Barão, L., Ferreira, C.S.S., Schwilch, G., Basch, G., Garcia-Orenes, F., Morugan, A., Mataix-Solera, J., Kosmas, C., Glavan, M., Szabó, B., Hermann, T., Vizitiu, O.P., Lipiec, J., Frąc, M., Reintam, E., Xu, M., Di, J., Fan, H., Sukkel, W., Lemesle, J., Geissen, V., & Fleskens, L. (2020). Visual assessment of the impact of agricultural management practices on soil quality. Agronomy Journal, 112(4), 2608–2623. https://doi.org/10.1002/agj2.20216
Ansari, M.S., & Gautam, P.R. (2024). Classification of soil moisture content with the application of deep learning. Proceedings of the 2024 International Conference on Integrated Circuits, Communication, and Computing Systems (ICIC3S), 1–5. https://doi.org/10.1109/ICIC3S61846.2024.10603387
ASTM International. (2017). ASTM D2488-17e1: Standard practice for description and identification of soils (visual-manual procedures). ASTM International. https://doi.org/10.1520/D2488-17E01
Basha, S.H.S., Dubey, S.R., Pulabaigari, V., & Mukherjee, S. (2020). Impact of fully connected layers on performance of convolutional neural networks for image classification. Neurocomputing, 378, 112–119. https://doi.org/10.1016/j.neucom.2019.10.008
Bhadra, S.K., & Bhavanarayana, M. (1997). Estimation of the influence of soil moisture on soil colour. Zeitschrift für Pflanzenernährung und Bodenkunde, 160(3), 401–405. https://doi.org/10.1002/jpln.19971600309
Celano, G.G.A. (2021). A ResNet-50-based convolutional neural network model for language ID identification from speech recordings. Proceedings of the Third Workshop on Computational Typology and Multilingual NLP (SIGTYP 2021), 136–144. https://doi.org/10.18653/v1/2021.sigtyp-1.13
Chen, H.Y., Sang, I.C., Norris, W.R., Soylemezoglu, A., & Nottage, D. (2024). Terrain classification method using an NIR or RGB camera with a CNN-based fusion of vision and a reduced-order proprioception model. Computers and Electronics in Agriculture, 227, 109539. https://doi.org/10.1016/j.compag.2024.109539
ComWinTop Limited. (2023). CWT soil sensor (NPK type) manual. https://cdck-file-uploads-europe1.s3.dualstack.eu-west-1.amazonaws.com/arduino/original/4X/5/c/0/5c01c131ef10541bfc68f9553c669807e9e54451.pdf
Fu, Y., Taneja, P., Lin, S., Ji, W., Adamchuk, V.I., Daggupati, P., & Biswas, A. (2020). Predicting soil organic matter from cellular phone images under varying soil moisture. Geoderma, 361, 114020. https://doi.org/10.1016/j.geoderma.2019.114020
Ghosh, A., Sufian, A., Sultana, F., Chakrabarti, A., & De, D. (2020). Fundamental concepts of convolutional neural network. In D. De, S. Bhattacharyya, & S. Chakraborty (Eds.), Recent trends and advances in artificial intelligence and internet of things, 519–567. https://doi.org/10.1007/978-3-030-32644-9_36
Hegazi, E.H., Yang, L., & Huang, J. (2021). A convolutional neural network algorithm for soil moisture prediction from Sentinel-1 SAR images. Remote Sensing, 13(24), 4964. https://doi.org/10.3390/rs13244964
Hemdan, E.E.-D., & Al-Atroush, M.E. (2024). An efficient soil image classification framework using transfer learning models for intelligent geotechnical applications. In Proceedings of the ICSDI 2024 Volume 1, 440–447. https://doi.org/10.1007/978-981-97-8712-8_55
Jackson, R.S. (2020). Site selection and climate. Wine science (331–374). Brock University. https://doi.org/10.1016/B978-0-12-816118-0.00005-2
Jeong, H.Y., & Na, S.I. (2024). Efficient faba bean leaf disease identification through smart detection using deep convolutional neural networks. Legume Research, 47(8), 1404–1411. https://doi.org/10.18805/LRF-798
Kim, D., Kim, T., Jeon, J., & Son, Y. (2023). Convolutional neural network-based soil water content and density prediction model for agricultural land using soil surface images. Applied Sciences, 13(5), 2936. https://doi.org/10.3390/app13052936
Kulkarni, A., Patel, U., & Cheeran, A. (2025). Deep learning in medical imaging: CNN-based approach for multi-class brain tumor classification. Proceedings of the 2025 International Conference on Sustainable Energy Technologies and Computational Intelligence (SETCOM).
Omar, E.Z. (2020). Investigation and classification of fibre deformation using interferometric and machine learning techniques. Applied Physics B: Lasers and Optics, 126(4), 1–14. https://doi.org/10.1007/s00340-020-7399-1
Pan, M., Zhang, W., Zhong, Z., Jiang, X., Jiang, Y., Lin, C., Qi, L., & Wu, S. (2026). A Hybrid CNN-Transformer Model for Soil Texture Estimation from Microscopic Images. Agronomy, 16(3), 333. https://doi.org/10.3390/AGRONOMY16030333
Song, L., Li, H., Liu, S., Chen, J., Fan, J., Wang, Q., & Chanussot, J. (2024). Attention Network with Outdoor Illumination Variation Prior for Spectral Reconstruction from RGB Images. Remote Sensing, 16(1), 180. https://doi.org/10.3390/RS16010180
Qi, D., Junaidi, A., Howe, C.W., & Zain, A.M. (2024). Improving unbalanced security X-ray image classification using VGG16 and AlexNet with Z-Score normalization and augmentation. Lecture Notes in Electrical Engineering, 1182, 205–217. https://doi.org/10.1007/978-981-97-1463-6_14
Rasheed, M.W., Tang, J., Sarwar, A., Shah, S., Saddique, N., Khan, M. U., Imran Khan, M., Nawaz,S., Shamshiri, R.R., Aziz, M., & Sultan, M. (2022). Soil moisture measuring techniques and factors affecting the moisture dynamics: A comprehensive review. Sustainability, 14(18), 11538. https://doi.org/10.3390/su141811538
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
