Qualitative Forest Inventory in Eucalyptus Plantations Using Unmanned Aerial Vehicles, Multispectral Sensors, and Deep Learning

• Main Author: Magalhaes Albuquerque, Alexandre
• Publication Date: 2024
• Other Authors: Debiasi, Paula, Lourenco De Lima, Thierry Vinicius, Hirokawa Higa, Gabriel Toshio, Pistori, Hemerson, Ferraco Scolforo, Henrique, Ferreira Silva, Thais Cristina, De Andrade Porto, Joao Vitor, Stape, Jose Luiz [UNESP]
• Format: Article
• Language: eng
• Source: Repositório Institucional da UNESP
• Download full: http://dx.doi.org/10.1109/LGRS.2024.3465892; https://hdl.handle.net/11449/308837
• topic: Artificial intelligence (AI); Eucalyptus; photogrammetry; unmanned aerial vehicles (UAVs); vegetation indices

Forest inventory is an important activity for planning and decision-making in forest management. It is usually carried out in the field using different sampling methods and processes, which are usually limited by its high costs and by the scarcity of manpower. In this work, we evaluate deep learning methods in qualitative forest inventory of Eucalyptus plantations using unmanned aerial vehicles (UAVs), multispectral sensors, and deep learning. For evaluation, we present a dataset collected in two study areas located in different municipalities in the State of Mato Grosso do Sul, including field measurements collected by occasion of the qualitative forest inventory at four months (QFI 4m) and aerophotogrammetric coverage of 36 plots represented by 124 sampling units. State-of-the-art neural networks were then used to predict four variables, collected through traditional QFI 4m and approximated by two models: PB50 and PC50, which are adaptations of the PV50 index, and the total and average biomass in the sampling unit. The results show that the transformer-based architecture multiaxis vision transformer (MaxViT) presented the lowest errors in predicting all the variables. For example, for the PB50 variable, it achieved a root mean square error (RMSE) of 7.5 (±1.85) and a mean absolute percentage error (MAPE) of 0.33 (±0.23).