A New Approach for Soil Carbon and Nitrogen Assessment through Image Analysis: A Case Study with Oxisols and Inceptisols
DOI:
https://doi.org/10.22456/2175-2745.143392Keywords:
Soil, Carbon, Nitrogen, Transfer Learning, Deep LearningAbstract
This paper presents a new method for predicting carbon and nitrogen content in Oxisols and Inceptisols soils using image analysis. The dataset comprises 40 Oxisols images and 51 Inceptisols images collected in southern Brazil. The images were segmented to isolate the soil area, resulting in the extraction of 12,744 windows. A pre-trained EfficientNetV2-S model was fine-tuned for regression, adding five layers at the end. The model was evaluated using RMSE and R² metrics. Two experiments were conducted: one predicting carbon and nitrogen content for each window and another predicting the median values of the windows for each soil sample. The results show that the median-based prediction significantly improved model performance, achieving an R² of 0.7425 for carbon and 0.7774 for nitrogen. This method offers a faster, non-destructive alternative to laboratory analysis for estimating soil carbon and nitrogen content, contributing to sustainable agricultural practices.
Downloads
References
SILVA, F. M. et al. Using proximal sensors to assess pedogenetic development of inceptisols and oxisols in Brazil. Geoderma Regional, v. 28, p. e00465, 2022. ISSN 2352-0094. Disponível em: ⟨https://bit.ly/3zZKkYL⟩.
SKORUPA, A. L. et al. Similar soils but different soil-forming factors: Converging evolution of inceptisols in Brazil. Pedosphere, v. 27, n. 4, p. 747–757, 2017. ISSN 1002-0160. Disponível em: ⟨https://bit.ly/3Yw8tyr⟩.
GONÇALVES, T. D.; LOHE, C.; CAMPOS, J. E. G. Hydraulic characterization from porous aquifers of the Brazilian Federal District. Brazilian Journal of Geology, SciELO, v. 45, n. 2, p. 259–271, 2015. Disponível em: ⟨https://bit.ly/3BPIlqn⟩.
PINTO, L. C. et al. Role of inceptisols in the hydrology of mountainous catchments in southeastern Brazil. Journal of Hydrologic Engineering, v. 21, n. 2, p. 05015017, 2016. Disponível em: ⟨https://bit.ly/3NvHY6U⟩.
FONTANA, A.; BALIEIRO, F. de C.; PEREIRA, M. G. Classes de solos do Brasil: Sistema Brasileiro de Classificação de Solos. [S.l.]: Embrapa Solos & UFRRJ, 2020.
SCIENTIFICA, V. Carbon and Nitrogen Analysis in Soil for Sustainable Agriculture and Farming. 2021. Disponível em: ⟨https://bit.ly/3LKacdh⟩.
MORAIS, P. A. d. O. et al. Avaliando a fertilidade do solo com imagens digitais. Revista Processos Químicos, v. 14, n. 27, p. 9–16, set. 2020. Disponível em: ⟨https://bit.ly/40sduL1⟩.
AI, X. et al. Prediction of soil organic carbon content using machine learning based fuzzy c-means clustering. In: 2024 Third International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE). [s.n.], 2024. p. 1–4. Disponível em: ⟨https://bit.ly/4dk9HCl⟩.
BARRA, I.; MOATASSEM, T. E.; KEBEDE, F. Soil particle size thresholds in soil spectroscopy and its effect on the multivariate models for the analysis of soil properties. Sensors, v. 23, n. 22, 2023. ISSN 1424-8220. Disponível em: ⟨https://bit.ly/4iP11rt⟩.
TUDOCELULAR. Ficha técnica do Samsung Galaxy J4 Plus. Acessado Dez. 2024. Disponível em: ⟨https://bit.ly/3BB9Kg4⟩.
GOOGLEPLAY. Aplicativo Open Camera. Acessado Dez. 2024. Disponível em: ⟨https://bit.ly/3P5uAao⟩.
GORDON, D. C.; SUTCLIFFE, W. H. A new dry combustion method for the simultaneous determination of total organic carbon and nitrogen in seawater. Marine Chemistry, v. 1, n. 3, p. 231–244, 1973. ISSN 0304-4203. Disponível em: ⟨https://bit.ly/3Ufdee6⟩.
DENG, J. et al. Imagenet: A large-scale hierarchical image database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition. [s.n.], 2009. p. 248–255. Disponível em: ⟨https://bit.ly/4dPew6D⟩.
GUTOSKI, M. et al. A comparative study of transfer learning approaches for video anomaly detection. International Journal of Pattern Recognition and Artificial Intelligence, v. 35, n. 05, p. 2152003, 2021. Disponível em: ⟨https://bit.ly/3YtUJW7⟩.
AQUINO, M. R. et al. A study of the influence of data complexity and similarity on soft biometrics classification performance in a transfer learning scenario. Learning and Nonlinear Models, v. 18, p. 56–65, 06 2021. Disponível em: ⟨https://bit.ly/40aoTPs⟩.
KUMAR, A. Hold-out method for training machine learning models. 2022. Disponível em: ⟨https://bit.ly/3LP0KFm⟩.
CHAI, T.; DRAXLER, R. R. Root mean square error (rmse) or mean absolute error (mae)? – arguments against avoiding rmse in the literature. Geoscientific Model Development, v. 7, p. 1247–1250, 2014. Disponível em: ⟨https://bit.ly/4doeL8O⟩.
KVÅLSETH, T. O. Cautionary note about r 2. The American Statistician, ASA Website, v. 39, n. 4, p. 279–285, 1985. Disponível em: ⟨https://bit.ly/4d1wR0y⟩.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Juliano Tiago Rinaldi, Dalcimar Casanova, Larissa Macedo dos Santos Tonial, Heitor Silvério Lopes, Marcelo Teixeira

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Autorizo aos editores a publicação de meu artigo, caso seja aceito, em meio eletrônico de acordo com as regras do Public Knowledge Project.