António Ribeiro Sousa

Abstract
Unmanned aerial vehicles have become a popular remote sensing platform for agricultural applications, with an emphasis on crop monitoring. Although there are several methods to detect vegetation through aerial imagery, these remain dependent of manual extraction of vegetation parameters. This article presents an automatic method that allows for individual tree detection and multi-temporal analysis, which is crucial in the detection of missing and new trees and monitoring their health conditions over time. The proposed method is based on the computation of vegetation indices (VIs), while using visible (RGB) and near-infrared (NIR) domain combination bands combined with the canopy height model. An overall segmentation accuracy above 95% was reached, even when RGB-based VIs were used. The proposed method is divided in three major steps: (1) segmentation and first clustering; (2) cluster isolation; and (3) feature extraction. This approach was applied to several chestnut plantations and some parameters—such as the number of trees present in a plantation (accuracy above 97%), the canopy coverage (93% to 99% accuracy), the tree height (RMSE of 0.33 m and R2 = 0.86), and the crown diameter (RMSE of 0.44 m and R2 = 0.96)—were automatically extracted. Therefore, by enabling the substitution of time-consuming and costly field campaigns, the proposed method represents a good contribution in managing chestnut plantations in a quicker and more sustainable way.

Abstract
<p><strong>Abstract.</strong> Today’s climatic proneness to extreme conditions together with human activity have been triggering a series of wildfire-related events that put at risk ecosystems, as well as animal and vegetal patrimony, while threatening dwellers nearby rural or urban areas. When intervention teams - firefighters, civil protection, police - acknowledge these events, usually they have already escalated to proportions hardly controllable mainly due wind gusts, fuel-like solo conditions, among other conditions that propitiate fire spreading.</p> <p>Currently, there is a wide range of camera-capable sensing systems that can be complemented with useful location data - for example, unmanned aerial systems (UAS) integrated cameras and IMU/GPS sensors, stationary surveillance systems - and processing components capable of fostering wildfire events detection and monitoring, thus providing accurate and faithful data for decision support. Precisely in what concerns to detection and monitoring, Deep Learning (DL) has been successfully applied to perform tasks involving classification and/or segmentation of objects of interest in several fields, such as Agriculture, Forestry and other similar areas. Usually, for an effective DL application, more specifically, based on imagery, datasets must rely on heavy and burdensome logistics to gather a representative problem formulation. What if putting together a dataset could be supported in customizable virtual environments, representing faithful situations to train machines, as it already occurs for human training in what regards some particular tasks (rescue operations, surgeries, industry assembling, etc.)?</p> <p>This work intends to propose not only a system to produce faithful virtual environments to complement and/or even supplant the need for dataset gathering logistics while eventually dealing with hypothetical proposals considering climate change events, but also to create tools for synthesizing wildfire environments for DL application. It will therefore enable to extend existing fire datasets with new data generated by human interaction and supervision, viable for training a computational entity. To that end, a study is presented to assess at which extent data virtually generated data can contribute to an effective DL system aiming to identify and segment fire, bearing in mind future developments of active monitoring systems to timely detect fire events and hopefully provide decision support systems to operational teams.</p>

Abstract
<p><strong>Abstract.</strong> In the early 1980's, the European chestnut tree (<i>Castanea sativa, Mill.</i>) assumed an important role in the Portuguese economy. Currently, the Trás-os-Montes region (Northeast of Portugal) concentrates the highest chestnuts production in Portugal, representing the major source of income in the region (€50M-€60M).</p> <p>The recognition of the quality of the Portuguese chestnut varieties has increasing the international demand for both industry and consumer-grade segments. As result, chestnut cultivation intensification has been witnessed, in such a way that widely disseminated monoculture practices are currently increasing environmental disaster risks. Depending on the dynamics of the location of interest, monocultures may lead to desertification and soil degradation even if it encompasses multiple causes and a whole range of consequences or impacts. In Trás-os-Montes, despite the strong increase in the cultivation area, phytosanitary problems, such as the chestnut ink disease (<i>Phytophthora cinnamomi</i>) and the chestnut blight (<i>Cryphonectria parasitica</i>), along with other threats, e.g. chestnut gall wasp (<i>Dryocosmus kuriphilus</i>) and nutritional deficiencies, are responsible for a significant decline of chestnut trees, with a real impact on production. The intensification of inappropriate agricultural practices also favours the onset of phytosanitary problems. Moreover, chestnut trees management and monitoring generally rely on in-field time-consuming and laborious observation campaigns. To mitigate the associated risks, it is crucial to establish an effective management and monitoring process to ensure crop cultivation sustainability, preventing at the same time risks of desertification and land degradation.</p> <p>Therefore, this study presents an automatic method that allows to perform chestnut clusters identification, a key-enabling task towards the achievement of important goals such as production estimation and multi-temporal crop evaluation. The proposed methodology consists in the use of Convolutional Neural Networks (CNNs) to classify and segment the chestnut fruits, considering a small dataset acquired based on digital terrestrial camera.</p>

Abstract
<p><strong>Abstract.</strong> In recent years unmanned aerial vehicles (UAVs) have been used in several applications and research studies related to environmental monitoring. The works performed have demonstrated the suitability of UAVs to be employed in different scenarios, taking advantage of its capacity to acquire high-resolution data from different sensing payloads, in a timely and flexible manner. In forestry ecosystems, UAVs can be used with accuracies comparable with traditional methods to retrieve different forest properties, to monitor forest disturbances and to support disaster monitoring in fire and post-fire scenarios. In this study an area recently affected by a wildfire was surveyed using two UAVs to acquire multi-spectral data and RGB imagery at different resolutions. By analysing the surveyed area, it was possible to detect trees, that were able to survive to the fire. By comparing the ground-truth data and the measurements estimated from the UAV-imagery, it was found a positive correlation between burned height and a high correlation for tree height. The mean NDVI value was extracted used to create a three classes map. Higher NDVI values were mostly located in trees that survived that were not/barely affected by the fire. The results achieved by this study reiterate the effectiveness of UAVs to be used as a timely, efficient and cost-effective data acquisition tool, helping for forestry management planning and for monitoring forest rehabilitation in post-fire scenarios.</p>

Abstract