The classification of tree species can significantly benefit from high spatial and spectral information acquired by unmanned aerial vehicles (UAVs) associated with advanced classification methods. This study investigated the following topics concerning the classification of 16 tree species in two subtropical forest fragments of Southern Brazil: i) the potential integration of UAV-borne hyperspectral images with 3D information derived from their photogrammetric point cloud (PPC); ii) the performance of two machine learning methods (support vector machine – SVM and random forest – RF) when employing different datasets at a pixel and individual tree crown (ITC) levels; iii) the potential of two methods for dealing with the imbalanced sample set problem: a new weighted SVM (wSVM) approach, which attributes different weights to each sample and class, and a deep learning classifier (convolutional neural network – CNN), associated with a previous step to balance the sample set; and finally, iv) the potential of this last classifier for tree species classification as compared to the above mentioned machine learning methods. Results showed that the inclusion of the PPC features to the hyperspectral data provided a great accuracy increase in tree species classification results when conventional machine learning methods were applied, between 13 and 17% depending on the classifier and the study area characteristics. When using the PPC features and the canopy height model (CHM), associated with the majority vote (MV) rule, the SVM, wSVM and RF classifiers reached accuracies similar to the CNN, which outperformed these classifiers for both areas when considering the pixel-based classifications (overall accuracy of 84.4% in Area 1, and 74.95% in Area 2). The CNN was between 22% and 26% more accurate than the SVM and RF when only the hyperspectral bands were employed. The wSVM provided a slight increase in accuracy not only for some lesser represented classes, but also some major classes in Area 2. While conventional machine learning methods are faster, they demonstrated to be less stable to changes in datasets, depending on prior segmentation and hand-engineered features to reach similar accuracies to those attained by the CNN. To date, CNNs have been barely explored for the classification of tree species, and CNN-based classifications in the literature have not dealt with hyperspectral data specifically focusing on tropical environments. This paper thus presents innovative strategies for classifying tree species in subtropical forest areas at a refined legend level, integrating UAV-borne 2D hyperspectral and 3D photogrammetric data and relying on both deep and conventional machine learning approaches.

Studies designed to discriminate different successional forest stages play a strategic role in forest management, forest policy and environmental conservation in tropical environments. The discrimination of different successional forest stages is still a challenge due to the spectral similarity among the concerned classes. Considering this, the objective of this paper was to investigate the performance of Sentinel-2 and Landsat-8 data for discriminating different successional forest stages of a patch located in a subtropical portion of the Atlantic Rain Forest in Southern Brazil with the aid of two machine learning algorithms and relying on the use of spectral reflectance data selected over two seasons and attributes thereof derived. Random Forest (RF) and Support Vector Machine (SVM) were used as classifiers with different subsets of predictor variables (multitemporal spectral reflectance, textural metrics and vegetation indices). All the experiments reached satisfactory results, with Kappa indices varying between 0.9, with Landsat-8 spectral reflectance alone and the SVM algorithm, and 0.98, with Sentinel-2 spectral reflectance alone also associated with the SVM algorithm. The Landsat-8 data had a significant increase in accuracy with the inclusion of other predictor variables in the classification process besides the pure spectral reflectance bands. The classification methods SVM and RF had similar performances in general. As to the RF method, the texture mean of the red-edge and SWIR bands were considered the most important ranked attributes for the classification of Sentinel-2 data, while attributes resulting from multitemporal bands, textural metrics of SWIR bands and vegetation indices were the most important ones in the Landsat-8 data classification.

The use of remote sensing data for tree species classification in tropical forests is still a challenging task, due to their high floristic and spectral diversity. In this sense, novel sensors on board of unmanned aerial vehicle (UAV) platforms are a rapidly evolving technology that provides new possibilities for tropical tree species mapping. Besides the acquisition of high spatial and spectral resolution images, UAV-hyperspectral cameras operating in frame format enable to produce 3D hyperspectral point clouds. This study investigated the use of UAV-acquired hyperspectral images and UAV-photogrammetric point cloud (PPC) for classification of 12 major tree species in a subtropical forest fragment in Southern Brazil. Different datasets containing hyperspectral visible/near-infrared (VNIR) bands, PPC features, canopy height model (CHM), and other features extracted from hyperspectral data (i.e., texture, vegetation indices-VIs, and minimum noise fraction-MNF) were tested using a support vector machine (SVM) classifier. The results showed that the use of VNIR hyperspectral bands alone reached an overall accuracy (OA) of 57% (Kappa index of 0.53). Adding PPC features to the VNIR hyperspectral bands increased the OA by 11%. The best result was achieved combining VNIR bands, PPC features, CHM, and VIs (OA of 72.4% and Kappa index of 0.70). When only the CHM was added to VNIR bands, the OA increased by 4.2%. Among the hyperspectral features, besides all the VNIR bands and the two VIs (NDVI and PSSR), the first four MNF features and the textural mean of 565 and 679 nm spectral bands were pointed out as more important to discriminate the tree species according to Jeffries–Matusita (JM) distance. The SVM method proved to be a good classifier for the tree species recognition task, even in the presence of a high number of classes and a small dataset.

In recent years, many agriculture-related problems have been evaluated with the integration of artificial intelligence techniques and remote sensing systems. Specifically, in fruit detection problems, several recent works were developed using Deep Learning (DL) methods applied in images acquired in different acquisition levels. However, the increasing use of anti-hail plastic net cover in commercial orchards highlights the importance of terrestrial remote sensing systems. Apples are one of the most highly-challenging fruits to be detected in images, mainly because of the target occlusion problem occurrence. Additionally, the introduction of high-density apple tree orchards makes the identification of single fruits a real challenge. To support farmers to detect apple fruits efficiently, this paper presents an approach based on the Adaptive Training Sample Selection (ATSS) deep learning method applied to close-range and low-cost terrestrial RGB images. The correct identification supports apple production forecasting and gives local producers a better idea of forthcoming management practices. The main advantage of the ATSS method is that only the center point of the objects is labeled, which is much more practicable and realistic than bounding-box annotations in heavily dense fruit orchards. Additionally, we evaluated other object detection methods such as RetinaNet, Libra Regions with Convolutional Neural Network (R-CNN), Cascade R-CNN, Faster R-CNN, Feature Selective Anchor-Free (FSAF), and High-Resolution Network (HRNet). The study area is a highly-dense apple orchard consisting of Fuji Suprema apple fruits (Malus domestica Borkh) located in a smallholder farm in the state of Santa Catarina (southern Brazil). A total of 398 terrestrial images were taken nearly perpendicularly in front of the trees by a professional camera, assuring both a good vertical coverage of the apple trees in terms of heights and overlapping between picture frames. After, the high-resolution RGB images were divided into several patches for helping the detection of small and/or occluded apples. A total of 3119, 840, and 2010 patches were used for training, validation, and testing, respectively. Moreover, the proposed method’s generalization capability was assessed by applying simulated image corruptions to the test set images with different severity levels, including noise, blurs, weather, and digital processing. Experiments were also conducted by varying the bounding box size (80, 100, 120, 140, 160, and 180 pixels) in the image original for the proposed approach. Our results showed that the ATSS-based method slightly outperformed all other deep learning methods, between 2.4% and 0.3%. Also, we verified that the best result was obtained with a bounding box size of 160 × 160 pixels. The proposed method was robust regarding most of the corruption, except for snow, frost, and fog weather conditions. Finally, a benchmark of the reported dataset is also generated and publicly available.

This work explores the integration of airborne Light Detection and Ranging (LiDAR) data and WorldView-2 (WV2) images to classify the land cover of a subtropical forest area in Southern Brazil. Different deep and machine learning methods were used: one based on convolutional neural network (CNN) and three ensemble methods. We adopted both pixel- (in the case of CNN) and object-based approaches. The results demonstrated that the integration of LiDAR and WV2 data led to a significant increase (7% to 16%) in accuracies for all classifiers, with kappa coefficient (κ) ranging from 0.74 for the random forest (RF) classifier associated with the WV2 dataset, to 0.92 for the forest by penalizing attributes (FPA) with the full (LiDAR + WV2) dataset. Using the WV2 dataset solely, the best κ was 0.81 with CNN classifier, while for the LiDAR dataset, the best κ was 0.8 with the rotation forest (RotF) algorithm. The use of LiDAR data was especially useful for the discrimination of vegetation classes because of the different height properties among them. In its turn, the WV2 data provided better performance for classes with less structure variation, such as field and bare soil. All the classification algorithms had a nearly similar performance: the results vary slightly according to the dataset used and none of the methods achieved the best accuracy for all classes. It was noticed that both datasets (WV2 and LiDAR) even when applied alone achieved good results with deep and machine learning methods. However, the advantages of integrating active and passive sensors were evident. All these methods provided promising results for land cover classification experiments of the study area in this work.

The Santa Catarina Southern Plateau is located in Southern Brazil and is a region that has gained considerable attention due to the rapid conversion of the typical landscape of natural grasslands and wetlands into agriculture, reforestation, pasture, and more recently, wind farms. This study’s main goal was to characterize the polarimetric attributes of the experimental quad-polarization acquisition mode of the Advanced Land Observing Satellite/Phased Array type L-band Synthetic Aperture Radar (ALOS/PALSAR-2) for mapping seven land cover classes. The polarimetric attributes were evaluated alone and combined with SENTINEL-2A using a supervised classification method based on the Support Vector Machine (SVM) algorithm. The results showed that the intensity backscattering alone reached an overall classification accuracy of 37.48% and a Kappa index of 0.26. Interestingly, the addition of polarimetric features increased to 71.35% and 0.66, respectively. It shows that the use of polarimetric decomposition features was relatively efficient in discriminating land cover classes. SENTINEL-2A data alone performed better and achieved a weighted overall accuracy and Kappa index of 85.56% and 0.82. This increase was also significant for the Z-test. However, the addition of ALOS/PALSAR-2 derived features to SENTINEL-2A slightly improved accuracy and was marginally significant at a 95% confidence level only when all features were considered. Possible implications for that performance are the accumulated precipitation prior to SAR data acquisition, which coincides with the rainy season period. The experimental quad-polarization mode of ALOS/PALSAR- 2 shall be evaluated in the near future over different seasonal conditions to confirm results. Alternatively, further studies are then suggested by focusing on additional features derived from SAR data such as texture and interferometric coherence to increase classification accuracy. These measures would be an interesting data source for monitoring specific land cover classes such as the threatened grasslands and wetlands during periods of frequent cloud coverage. Future investigations could also address multitemporal approaches employing either single or multifrequency SAR.