Unmanned Aerial Vehicle Platform Research Articles

Truthful and performance-optimal computation outsourcing for aerial surveillance platforms via learning-based auction

This paper proposes a novel truthful computing algorithm for learning task outsourcing decision-making strategies in edge-enabled unmanned aerial vehicle (UAV) networks. In our considered scenario, a single UAV performs face identification in a monitored target area. The execution of the identification requires a certain computing power, and its complexity and time are dependent on the number of faces in the recorded images. As a consequence, the task cannot be fully executed by a single UAV under high image arrivals or with images that have a high density of faces. In those conditions, UAV can outsource the task to one of the nearby edges. Importantly, the computing task distribution should be energy-efficient and delay-minimal due to the constraints imposed by the UAV platform characteristics and applications. Based on those fundamental requirements, our proposed algorithm conducts sequential decision-making for image sharing with one selected edge. The edge is selected based on a second price auction for truthfulness. Besides the truthfulness guarantees, deep learning based approximation for the auction solution is used for revenue-optimality. Our evaluation demonstrates that the proposed algorithm achieves the desired performance.

Service Function Chain Scheduling in Heterogeneous Multi-UAV Edge Computing

Supporting Artificial Intelligence (AI)-enhanced intelligent applications on the resource-limited Unmanned Aerial Vehicle (UAV) platform is difficult due to the resource gap between the two. It is promising to partition an AI application into a service function (SF) chain and then dispatch the SFs onto multiple UAVs. However, it is still a challenging task to efficiently schedule the computation and communication resources of multiple UAVs to support a large number of SF chains (SFCs). Under the multi-UAV edge computing paradigm, this paper formulates the SFC scheduling problem as a 0–1 nonlinear integer programming problem. Then, a two-stage heuristic algorithm is put forward to solve this problem. At the first stage, if the resources are surplus, the SFCs are deployed to UAV edge servers in parallel based on our proposed pairing principle between SFCs and UAVs for minimizing the completion time sum of tasks. In contrast, a revenue maximization heuristic method is adopted to deploy the arrived SFCs in a serial service mode when the resource is insufficient. A series of experiments are conducted to evaluate the performance of our proposal. Results show that our algorithm outperforms other benchmark algorithms in the completion time sum of tasks, the overall revenue, and the task execution success ratio

Operation quality control of rapeseed strip aerial seeding system via under-constrained seeding technique

Due to the strong aerodynamic disturbance induced by high-speed rotating rotors and under-constrained seeding condition, it is difficult to control the width of landing seeds in target area during the aerial seeding process. In this study, based on the multi-rotor plant protection unmanned aerial vehicle (UAV) platform, the strategy of under-constrained seeding at a low dropping height is employed to realize the strip aerial seeding of rapeseed. Firstly, the spatial-temporal characteristics of downwash flow field in forward flight state are verified via the high-fidelity computational fluid dynamics method. Afterwards, the migration mechanism of falling seeds and operation parameter optimization are investigated by employing the gas-solid coupling simulation technique. Finally, flight experiments are carried out based on multi-rotor UAV platform to verify the validity of simulation results, where the image processing technique is used to accurately obtain the coordinates and distribution characteristics of landing seeds. The result indicates that the strip effect could be further improved with a higher flight speed and lower dropping height, for instance, the strip width of 95 % landing seeds could be controlled within 81.1 mm when the flight speed and dropping height are 3.5 m/s and 0.5 m respectively. The research results can provide a suggestion for scientific setting of operation parameters of aerial seeding system.

UAV-towed scalar magnetic gradiometry: A case study in relation to iron oxide copper-gold mineralization, Nautanen (Arctic Sweden)

Scalar magnetic surveying using unmanned aerial vehicle (UAV) platforms is slowly gaining momentum within geophysical applications. So far, only a handful of studies have dealt with UAV-towed scalar field measurements, while even fewer have considered towed scalar difference measurements (or gradients). In this paper, we demonstrate the possibilities and benefits of deploying precisely positioned noise-minimized UAV-towed scalar transverse horizontal difference (THD) measurements for mineral exploration purposes. UAV-towed gradiometry bird data are presented from the Nautanen area in northern Sweden and compared with ground magnetic surveys. This area is known for its iron oxide copper-gold mineralizations. The UAV survey spans a total area of 2.5 km2. It was covered using an average line spacing of 30 m and a constant flight altitude above ground level of 30 m. High-quality scalar total-field and THD data were collected with a dynamic noise level of the raw scalar data of about ±0.05 nT. Comparison with the ground magnetic data shows a strong correlation between magnetic anomaly lows and highs across the survey areas. A map with new structural information is presented based on subtle magnetic structures identified in discrete derivatives of the total magnetic intensity anomaly and THD data. Such systems may replace high-quality heliborne systems and reduce costs of the geophysical exploration phase. However, mapping with UAV-towed systems is not straightforward. With typical UAV flight speeds of only 10–12 m/s, the wind often disturbs the 3D attitude of the bird during flights. Hence, advanced processing is required to obtain the intended gradients. Similar challenges are less important in surveying, where the survey speed often greatly exceeds the wind speed.

A Comprehensive Review of Scab Disease Detection on Rosaceae Family Fruits via UAV Imagery

Disease detection in plants is essential for food security and economic stability. Unmanned aerial vehicle (UAV) imagery and artificial intelligence (AI) are valuable tools for it. The purpose of this review is to gather several methods used by our peers recently, hoping to provide some knowledge and assistance for researchers and farmers so that they can employ these technologies more advantageously. The studies reviewed in this paper focused on Scab detection in Rosaceae family fruits. Feature extraction, segmentation, and classification methods for processing the UAV-obtained images and detecting the diseases are discussed briefly. The advantages and limitations of diverse kinds of UAVs and imaging sensors are also explained. The widely applied methods for image analysis are machine learning (ML)-based models, and the extensively used UAV platforms are rotary-wing UAVs. Recent technologies that cope with challenges related to disease detection using UAV imagery are also detailed in this paper. Some challenging issues such as higher costs, limited batteries and flying time, huge and complex data, low resolution, and noisy images, etc., still require future consideration. The prime significance of this paper is to promote automation and user-friendly technologies in Scab detection.

Frequency Domain Electromagnetic System Based on Unmanned Aerial Vehicles Platform for Detecting Shallow Subsurface Targets

Due to the advantages of being nondestructive, rapid, and convenient, the electromagnetic detection method has attracted growing interest in the field of shallow subsurface detection. With the rapid development of unmanned aerial vehicle (UAV) technology, the use of the UAV platform for measurement can not only improve work efficiency but also avoid the significant losses that may be caused by humans working in dangerous areas. Therefore, we propose a broadband frequency domain electromagnetic system AFEM-3 based on a UAV platform for shallow subsurface targets detection (within less than 2 m). The sensor head adopts a concentric planar coil structure with a high spatial resolution, and a bucking coil connected in reverse series with the transmitting coil is used to suppress the primary field at the receiving coil. We designed a transmitting module based on unipolar frequency multiplication sinusoidal pulse width modulation technology that can generate multi-frequency arbitrary combination transmitting waveforms with low total harmonic distortion. It can also be matched to a variety of different transmitter coils by using the same hardware circuit. In addition, the global navigation satellite system and inertial measurement unit are integrated on the sensor head. The measurement response value, position, and attitude information can be displayed in real-time through the host computer. Through the static experiment of a standard coil, we verified the consistency between the AFEM-3 system with the theory. The performance of the system was evaluated through field experiments. The experimental results show that the system can effectively detect multiple metal targets in shallow subsurface areas. For different metal targets, the AFEM-3 system can provide obvious frequency domain characteristics.

Development of multi-channel whole-air sampling equipment onboard an unmanned aerial vehicle for investigating volatile organic compounds' vertical distribution in the planetary boundary layer

Abstract. To achieve near-continuous vertical observations of volatile organic compounds (VOCs) in the planetary boundary layer (PBL), multi-channel whole-air sampling equipment onboard an unmanned aerial vehicle (UAV) platform was developed in this study. The equipment consists of a multi-position solenoid valve and specially designed lightweight quartz sampling canisters. The canisters have little adsorption loss of VOCs and good inter-canister reproducibility. The 7 d recovery test shows that most VOC species (97 %) had a 1-week decay within 20 %. Online instruments for measuring O3, NO2, CO, SO2, and meteorological parameters are also integrated into the UAV platform. During one take-off and landing, the UAV platform can reach 800 m above the ground within 40 min and take whole-air samples at six heights. Vertical profiles of VOCs and trace gases during the evolution of the PBL in south-western China are successfully obtained by deploying the newly developed UAV system.

Local Adaptive Illumination-Driven Input-Level Fusion for Infrared and Visible Object Detection

Remote sensing object detection based on the combination of infrared and visible images can effectively adapt to the around-the-clock and changeable illumination conditions. However, most of the existing infrared and visible object detection networks need two backbone networks to extract the features of two modalities, respectively. Compared with the single modality detection network, this greatly increases the amount of calculation, which limits its real-time processing on the vehicle and unmanned aerial vehicle (UAV) platforms. Therefore, this paper proposes a local adaptive illumination-driven input-level fusion module (LAIIFusion). The previous methods for illumination perception only focus on the global illumination, ignoring the local differences. In this regard, we design a new illumination perception submodule, and newly define the value of illumination. With more accurate area selection and label design, the module can more effectively perceive the scene illumination condition. In addition, aiming at the problem of incomplete alignment between infrared and visible images, a submodule is designed for the rapid estimation of slight shifts. The experimental results show that the single modality detection algorithm based on LAIIFusion can ensure a large improvement in accuracy with a small loss of speed. On the DroneVehicle dataset, our module combined with YOLOv5L could achieve the best performance.

A Review of Estimation Methods for Aboveground Biomass in Grasslands Using UAV

Grasslands are one of the world’s largest ecosystems, accounting for 30% of total terrestrial biomass. Considering that aboveground biomass (AGB) is one of the most essential ecosystem services in grasslands, an accurate and faster method for estimating AGB is critical for managing, protecting, and promoting ecosystem sustainability. Unmanned aerial vehicles (UAVs) have emerged as a useful and practical tool for achieving this goal. Here, we review recent research studies that employ UAVs to estimate AGB in grassland ecosystems. We summarize different methods to establish a comprehensive workflow, from data collection in the field to data processing. For this purpose, 64 research articles were reviewed, focusing on several features including study site, grassland species composition, UAV platforms, flight parameters, sensors, field measurement, biomass indices, data processing, and analysis methods. The results demonstrate that there has been an increase in scientific research evaluating the use of UAVs in AGB estimation in grasslands during the period 2018–2022. Most of the studies were carried out in three countries (Germany, China, and USA), which indicates an urgent need for research in other locations where grassland ecosystems are abundant. We found RGB imaging was the most commonly used and is the most suitable for estimating AGB in grasslands at the moment, in terms of cost–benefit and data processing simplicity. In 50% of the studies, at least one vegetation index was used to estimate AGB; the Normalized Difference Vegetation Index (NDVI) was the most common. The most popular methods for data analysis were linear regression, partial least squares regression (PLSR), and random forest. Studies that used spectral and structural data showed that models incorporating both data types outperformed models utilizing only one. We also observed that research in this field has been limited both spatially and temporally. For example, only a small number of papers conducted studies over a number of years and in multiple places, suggesting that the protocols are not transferable to other locations and time points. Despite these limitations, and in the light of the rapid advances, we anticipate that UAV methods for AGB estimation in grasslands will continue improving and may become commercialized for farming applications in the near future.

Application of Precision Agriculture for the Sustainable Management of Fertilization in Olive Groves

Olive tree growing (Olea europaea L.) has considerably increased in the last decades, as has the consumption of extra virgin olive oil in the world. Precision agriculture is increasingly being applied in olive orchards as a new method to manage agronomic variability with the aim of providing individual plants with the right input amount, limiting waste or excess. The objective of this study was to develop a methodology on a GIS platform using GEOBIA algorithms in order to build prescription maps for variable rate (VRT) nitrogen fertilizers application in an olive orchard. The fertilization plan was determined for each tree by applying its own nitrogen balance, taking into account the variability of nitrogen in soil, leaf, production, and actual biometric and spectral conditions. Each olive tree was georeferenced using the S7-G Stonex instrument with real-time kinematic RTK positioning correction and the trunk cross section area (TCSA) was measured. Soil and leaves were sampled to study nutrient variability. Soil and plant samples were analyzed for all major physical and chemical properties. Spectral data were obtained using a multispectral camera (DJI multispectral) carried by an unmanned aerial vehicle (UAV) platform (DJI Phantom4). The biometric characteristics of the plants were extracted from the achieved normalized vegetation index (NDVI) map. The obtained prescription map can be used for variable rate fertilization with a tractor and fertilizer spreader connected via the ISOBUS system. Using the proposed methodology, the variable rate application of nitrogen fertilizer resulted in a 31% reduction in the amount to be applied in the olive orchard compared to the standard dose.

Evaluation of Multispectral Data Acquired from UAV Platform in Olive Orchard

Precision agriculture is a management strategy to improve resource efficiency, production, quality, profitability and sustainability of the crops. In recent years, olive tree management is increasingly focused on determining the correct health status of the plants in order to distribute the main resource using different technologies. In the olive grove, the focus is often on the use of multispectral information from UAVs (Unmanned Aerial Vehicle), but it is not known how important spectral and biometric information actually is for the agronomic management of the olive grove. The aim of this study was to investigate the ability of multispectral data acquired from a UAV platform to predict nutritional status, biometric characteristics, vegetative condition and production of olive orchard as tool to DSS. Data were collected on vegetative characteristics closely related to vigour such as trunk cross-sectional area (TCSA), Nitrogen concentration of the leaves, canopy area and canopy volume. The production was collected for each plant to create an accurate yield map. The flight was carried out with a UAV equipped with a multispectral camera, at an altitude of 50 m and with RTK correction. The flight made it possible to determine the biometric condition and the spectral features through the normalized difference vegetation index (NDVI). The NDVI map allowed to determine the canopy area. The Structure for Motion (SfM) algorithm allow to determine the 3D canopy volume. The experiment showed that the NDVI was able to determine with high accuracy the vegetative characteristic as canopy area (r = 0.87 ***), TCSA (r = 0.58 ***) and production (r = 0.63 ***). The vegetative parameters are closely correlated with the production, especially the canopy area (r = 0.75 ***). Data clustering showed that the production of individual plants is closely dependent on leaf nitrogen concentration and vigour status.

UAV-Correlated MIMO Channels: 3-D Geometrical-Based Polarized Model and Capacity Analysis

The unmanned aerial vehicle (UAV) communication with the multiple-input–multiple-output (MIMO) system has attracted lots of attention to improve spectral efficiency and channel capacity. The large antenna spacing is often required to reduce the spatial correlation between subchannels. However, this requirement cannot be well satisfied in practical scenarios due to the limited dimension of UAV platforms. Therefore, polarization diversity is a promising approach for UAV MIMO systems. In this work, we propose a three-dimensional (3-D) geometrical-based polarized model for UAV-correlated MIMO channels. By utilizing the geometrical theory of polarization, we describe the channel depolarization caused by the terrestrial scattering environment based on the multicylinder geometrical model and then acquire the channel polarization function and polarized channel impulse response (CIR). Furthermore, we investigate the impact of the key factors, such as the UAV tilt rotation, cross-polarization, and limited antenna spacing on the UAV MIMO channel capacity. To validate the proposed UAV channel model, we compare the spatial correlations of the numerical simulation and UAV field measurement results for the co- and cross-polarized channels, and the close agreements between them are observed. This work can provide useful guidance and support for the design and performance evaluation of UAV communication networks.

Design and Application of an Airborne Radioactivity Survey System Based on Unmanned Aerial Vehicle

In this study, an airborne radioactivity survey (ARS) system based on an unmanned helicopter platform was designed to satisfy the requirements of geological exploration and environmental radiation monitoring under complex environmental conditions. The system is composed of an unmanned aerial vehicle platform, airborne equipment, and a ground control station, with both NaI(Tl) and CeBr3 crystals as detectors to meet the energy resolution and detection efficiency requirements of airborne radioactivity measurement. The designed ARS system is stable and reliable through the field flight test, and the measurement results are consistent with the existing data and can be widely used in fields such as large-scale uranium resource exploration, nuclear accident emergency monitoring, and environmental radiation investigation.

Array-Based Ground Penetrating Synthetic Aperture Radar on Board an Unmanned Aerial Vehicle for Enhanced Buried Threats Detection

The usage of unmanned aerial vehicles (UAVs)-based ground penetrating radar (GPR) systems has gained interest over the last years thanks to advantages over ground-based systems, such as contactless inspection and capability, to reach difficult-to-access areas. The former is of paramount importance concerning the detection of buried threats, such as improvised explosive devices (IEDs) and landmines. Current state-of-the-art UAV-based GPR systems are able to provide centimeter-level resolution thanks to the use of GPR-synthetic aperture radar (SAR) processing techniques. One of the challenges to keep improving these systems is the scanning throughput, that is, the area that can be scanned in a given time. This contribution presents an array-based GPR-SAR system for subsurface imaging, aiming at maximizing the scanning throughput without jeopardizing the imaging capabilities of the system. First, the antenna array is mounted on a portable setup to evaluate its performance and imaging capabilities. Next, the antenna array is integrated into the UAV platform, and the UAV-based GPR-SAR system with the array is tested in realistic scenarios with different kinds of buried targets. Results show that the scanning throughput is significantly improved and, furthermore, the coherent combination of all transmitting–receiving channels of the array provides enhanced detection capabilities.

Moving target detection based on background modeling and frame difference

The integration of Unmanned Aerial Vehicle (UAV) technology with moving target detection has diverse applications in military reconnaissance, space remote sensing, and smart cities. Traditional motion-based target detection algorithms offer fast processing speeds but lack accuracy. Deep learning-based algorithms, while accurate for specific targets only, are complex and not suitable for resource-limited UAV platforms and lack real-time performance. Therefore, this study proposes a real-time moving target detection algorithm for UAV platforms based on traditional frame difference algorithm. The purpose of this algorithm is to improve detection accuracy, which has been hindered by the limitations of traditional algorithms caused by camera shake, background changes, and fast-moving targets. The algorithm involves rough background modeling, background updating during subsequent video image sequences, image morphology processing, and background compensation. Experimental results from multiple sets of UAV-borne video data demonstrate the algorithm's high target detection rate, low false alarm rate, and ability to detect moving targets stably in complex environments. The proposed algorithm achieves a speed of 25 FPS and a detection accuracy of 91.8%, meeting real-time and accurate detection requirements.

Improved YOLOv5 for Aerial Images Based on Attention Mechanism

Object detection based on unmanned aerial vehicle(UAV) platforms is essential for both engineering and research. Complex scale problems in UAV application scenarios require strong regression localization capabilities from target detection algorithms. Nonetheless, due to the constraints of UAV platform, it is difficult to increase accuracy by deepening the network. Therefore, this paper presents an improved YOLOv5 with an attention mechanism, consisting a Convolution-Swin Transformer Block(CSTB) utilizing Swin Transformer as well as a Convolution-block Attention Module(CBAM) to improve network positioning accuracy. In addition, this paper incorporates Bidirectional Feature Pyramid Network(BiFPN) [1], Spatial Pyramid Pooling-Fast(SPPF) and some network components to increase the average precision while maintaining the limited size of the model. Experiments on Visdrone2019 dataset show that the proposed approach can raise the mean Average Precision(mAP) by 5.4% compared to YOLOv5, with only 18% increase in model size.

Monitoring Thosea sinensis Walker in Tea Plantations Based on UAV Multi-Spectral Image

Thosea sinensis Walker (TSW) rapidly spreads and severely damages the tea plants. Therefore, finding a reliable operational method for identifying the TSW-damaged areas via remote sensing has been a focus of a research community. Such methods also enable us to calculate the precise application of pesticides and prevent the subsequent spread of the pests. In this work, based on the unmanned aerial vehicle (UAV) platform, five band images of multispectral red-edge camera were obtained and used for monitoring the TSW in tea plantations. By combining the minimum redundancy maximum relevance (mRMR) with the selected spectral features, a comprehensive spectral selection strategy was proposed. Then, based on the selected spectral features, three classic machine learning algorithms, including random forest (RF), support vector machine (SVM), and k-nearest neighbors (KNN) were used to construct the pest monitoring model and were evaluated and compared. The results showed that the strategy proposed in this work obtained ideal monitoring accuracy by only using the combination of a few optimized features (2 or 4). In order to differentiate the healthy and TSW-damaged areas (2-class model), the monitoring accuracies of all the three models were computed, which were above 96%. The RF model used the least number of features, including only SAVI and Bandred. In order to further discriminate the pest incidence levels (3-class model), the monitoring accuracies of all the three models were computed, which were above 80%, among which the RF algorithm based on SAVI, Bandred, VARI_green, and Bandred_edge features achieve the highest accuracy (OAA of 87%, and Kappa of 0.79). Considering the computational cost and model accuracy, this work recommends the RF model based on a few optimal feature combinations to monitor and distinguish the severity of TSW in tea plantations. According to the UAV remote sensing mapping results, the TSW infestation exhibited an aggregated distribution pattern. The spatial information of occurrence and severity can offer effective guidance for precise control of the pest. In addition, the relevant methods provide a reference for monitoring other leaf-eating pests, effectively improving the management level of plant protection in tea plantations, and guaranting the yield and quality of tea plantations.

ПОДЗАРЯДКА БЕСПИЛОТНЫХ ЛЕТАТЕЛЬНЫХ АППАРАТОВ С ВЕРТИКАЛЬНЫМ ВЗЛЕТОМ-ПОСАДКОЙ НА КОНТАКТНЫХ ПЛАТФОРМАХ С АДАПТИРУЕМОЙ ШИРИНОЙ КОНТАКТНЫХ ПОЛОС

Relevance The article describes one of the modifications of ground charging platforms for electric unmanned aerial vehicles (UAVs) with vertical takeoff and landing (VTOL). It belongs to the class of charging platforms with open pads. The previously described platforms with coplanar band parallel electrodes (CBPE) have multiple parallel strip electrodes with alternating polarity, isolated from each other by narrow dielectric separators. The corresponding on-board electrodes of the multi-rotor UAV with a symmetrical frame are proposed to be arranged at the ends of the support legs under the bearing beams of the engine-propeller groups. Thus, the contact points for the quadrocopter will be at the vertices of the square. With certain geometric relationships between the width of the ground electrodes and the interelectrode distances of the on-board electrodes for any UAV position on the platform, almost 100% different polarities of the on-board electrodes is ensured after landing, and as a consequence, the ability to bring the voltage from the ground DC source to the on-board battery charger. The advantages of such platforms are: relative simplicity of design, independence of maintenance from the accuracy of landing and the ability to charge several UAVs at the same time. However, UAVs may have various sizes and therefore various interelectrode distances, which may violate the condition of different polarities of on-board electrodes after landing. Therefore, it is important to find a solution in which reliable service of the CBPE platform for UAVs of different sizes is possible. Aim of research Finding a technical solution for adaptive band width change of a CBPE landing platform that allows charging UAVs of different sizes. Research methods Prototyping the adaptable CBPE platform and pilot study of its operation. Results The proposed modification of the platform consists in arranging the switching of bands in such a way that alternating groups of bands of one polarity are formed. Bands can be grouped into 1, 2, 3, 4, etc. Such a group works as a single band of appropriate width, which can be changed discretely. In this way, it is possible to adapt the width of the platform stripes depending on the type of vehicle coming for landing. An experimental landing platform for multicopters and a variant of implementing a commuting device for platform adaptive configuration are described.

Scale-Aware Pomegranate Yield Prediction Using UAV Imagery and Machine Learning

Highlights UAV-based remote sensing platform and ML methods for individual tree level yield prediction. Decision tree ML model had an accuracy of 85% for yield prediction. Vegetation indices derived from UAV high-resolution imagery were used for yield prediction. Scale-aware prediction model has great potential for efficient field management. Abstract. Accurately estimating the yield is one of the most important research projects for orchard management. Growers need to estimate the yield of trees at an early stage to make smart decisions for field management. However, methods to predict the yield at the individual tree level are currently not available because of the complexity and variability of each tree. To improve the accuracy of yield prediction, the authors evaluate the performance of an unmanned aerial vehicle (UAV)-based remote sensing system and machine learning (ML) approaches for tree-level pomegranate yield estimation. A lightweight multispectral camera was mounted on the UAV platform to acquire high-resolution images. Eight features were extracted from the UAV images, including the normalized difference vegetation index (NDVI), the green normalized vegetation index (GNDVI), the red-edge normalized difference vegetation index (NDVIre), red-edge triangulated vegetation index (RTVIcore), individual tree canopy size, the modified triangular vegetation index (MTVI2), the chlorophyll index-green (CIg), and the chlorophyll index-rededge (CIre). Correlation coefficients (R2) were calculated between these vegetation indices and tree yield. Classic machine learning approaches were applied with the extracted features to predict the yield at the individual tree level. Results showed that the decision tree classifier had the best prediction performance, with an accuracy of 85%. The study demonstrated the potential of using UAV-based remote sensing methods, coupled with ML algorithms, for pomegranate yield estimation. Predicting the yield at the individual tree level will enable the stakeholders to manage the orchard at different scales, thus improving field management efficiency. Keywords: Machine learning, Remote sensing, UAV, Vegetation index, Yield prediction.

Thermal UAV Image Super-Resolution Guided by Multiple Visible Cues

Unmanned aerial vehicle (UAV) thermal-imaging has received much attention, but the insufficient image resolution caused by thermal imaging systems is still a crucial problem that limits the understanding of thermal UAV images. However, high-resolution visible images are relatively easy to access, and it is thus valuable for exploring useful information from visible image to assist thermal UAV image super-resolution (SR). In this article, we propose a novel multiconditioned guidance network (MGNet) to effectively mine the information of visible images for thermal UAV image SR. High-resolution visible UAV images usually contain salient appearance, semantic, and edge information, which plays a critical role in boosting the performance of thermal UAV image SR. Therefore, we design an effective multicue guidance module (MGM) to leverage the appearance, edge, and semantic cues from visible images to guide thermal UAV image SR. In addition, we build the first benchmark dataset for the task of thermal UAV image SR guided by visible images. It is collected by a multimodal UAV platform and composes of 1025 pairs of manually aligned visible and thermal images. Extensive experiments on the built dataset show that our MGNet can effectively leverage useful information from visible images to improve the performance of thermal UAV image SR and perform well against several state-of-the-art methods. The dataset is available at: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/mmic-lcl/Datasets-and-benchmark-code</uri> .