Remotely piloted aircraft system and machine learning for detection of coffee plants subjected to foliar application of chitosan

Considered a biostimulant, chitosan can affect the physiological responses of plants to water deficit, acting as an antitranspirant under agricultural stress. Currently, images obtained by Remotely Piloted Aircraft Systems (RPAS), together with machine learning techniques, aid in resolving agricultural problems, including water issues. Therefore, the objective of this study was to differentiate between coffee plants subjected to the foliar application of chitosan and those not subjected to it, based on spectral data extracted from RPAS-acquired images and classification via machine learning. For this purpose, the random forest (RF) classifier was applied to two coffee cultivars (Catucaí Amarelo 2SL and Catuaí Vermelho IAC 99) over two years of study (2021 and 2022). The images were obtained by a 3DR SOLO aircraft with a Parrot Sequoia sensor, processed in PIX4D Mapper software and analysed in QGIS and RStudio software. The results showed good performance metrics for differentiating between coffee plants subjected and not subjected to the foliar application of chitosan, indicating that this method is a valid approach for modelling the presence of the biostimulant in coffee plants, thus confirming that the model can efficiently support the practices of precision agriculture.