Modeling and detection of invasive trees using UAV image and machine learning in a subtropical forest in Brazil

Identification of invasive trees in a Brazilian subtropical forest using remotely piloted aircraft systems and machine learning

“…The supervised classification process was carried out according to the methodology proposed by da Silva et al. 18 After conducting an analysis of the features in the experimental area using ArcGIS Pro software, we identified four classes that were of interest for training the classifiers. These classes included "Psidium guajava," "similar species" (visual photointerpretation of some species with a spectral response similar to the target species, such as the native species Psidium cattleianum), "shade," and "other species" (other tree species found in the study area with spectral response different from the target species).…”

“…15,16 In this regard, the use of remotely piloted aircraft systems (RPAS) for detecting invasive species in large and hard-to-reach areas has become more frequent, 7 driven by the development of sensors and positioning systems that enable the collection of high-resolution and accurate images at a low cost. 17 RPAS equipped with RGB, multispectral, and hyperspectral sensors play an essential role in the identification and monitoring of areas invaded by IAS, as reported in the studies by da Silva et al, 18 Koco et al, 19 Liang et al, 20 and Lopatin et al. 21 Furthermore, the processing of highresolution images obtained by RPAS can be handled by machine learning (ML) algorithms such as support vector machine (SVM) 22 and random forest (RF).…”

“…23 Both are considered the most effective and widely used classifiers for remote sensing data in the forest context. 17 Da Silva et al 18 employed the RF and SVM algorithms with the aim of detecting the IAS Hovenia dulcis in a protected area situated in southern Brazil, using sets of RGB images obtained from RPAS. Remarkably, the obtained results exhibited an overall accuracy rate of 91.5% and a kappa coefficient of 0.87.…”

“…RPAS equipped with RGB, multispectral, and hyperspectral sensors play an essential role in the identification and monitoring of areas invaded by IAS, as reported in the studies by da Silva et al., 18 Koco et al., 19 Liang et al., 20 and Lopatin et al. 21 .…”

Improvement of Hexacopter UAVs Attitude Parameters Employing Control and Decision Support Systems

“…Multirotor drones have seen continuous development over the past ten years as the need for this type of platform has grown continuously, and they are employed in a wide range of activities and fields such as inspection of large industrial installations [ 11 , 12 ], large buildings and constructions, oil and gas pipelines [ 13 , 14 ], inspection of continuous-flow machinery in quarries (to monitor temperatures in the area of high-friction pits using thermal imaging cameras), inspection of petrochemical installations (to detect cracks, fissures and leaks that may occur in pressure vessels using thermal imaging cameras [ 15 , 16 ]), etc. They can be equipped with a wide range of electromagnetic spectrum sensors [ 17 , 18 ], gamma ray sensors [ 19 ], biological sensors [ 20 , 21 , 22 ] and chemical sensors [ 23 , 24 ], which provide remote sensing functions.…”