Detection of Potato Crops in Advanced Vegetative Stage Using Sentinel 2 with Remote Sensing in the Peruvian Andes

Carrillo P.P. · Proceedings of the 2025 IEEE 32nd International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2025 (2025) · DOI: 10.1109/intercon67304.2025.11244653

Resumen

In Peru, 9,856 hectares of potato crops were lost between planting and harvesting during the 2022/2023 agricultural season. This highlights the limited monitoring that occurs through traditional, labor-intensive methodologies that require significant human resources (surveyors), time (transportation), and financial resources (logistics) for conducting in situ agricultural surveys. Consequently, agricultural observation reports are restricted by the lengthy timelines associated with traditional methodologies, which impairs real-time monitoring capacity. Therefore, the objective of this research is to develop a machine learning classification model capable of detecting potato crops in advanced vegetation stages using Sentinel-2 spectral imagery. The study involved building an agricultural area dataset using georeferenced polygons. These polygons were subsequently used to compile a gallery of Sentinel-2 spectral images for the 2022/2023 season. Pixel processing methods were applied through spectral bands to filter out cloud cover, and vegetation indices were calculated as supplementary attributes. Finally, positive (potato crops) and negative (other crops and agricultural areas) classes were created to train the XGBoost, Random Forest, and Support Vector Machine models. The best-trained model was SVM with an F1-score of 0.85, demonstrating its robust ability to distinguish potato crop vegetation from other agricultural areas. This study represents an initial step toward the detection of georeferenced crops in Peru, aiming to establish a foundation for future research on plot-level productivity assessment and the formulation of precise agricultural policies to strengthen national production.

Instituciones: Universidad del Pacífico