Search Results (280)

Effective monitoring of maize phenology under stress conditions is crucial for optimizing agricultural management and mitigating yield losses. Crop prediction models constructed from Convolutional Neural Network (CNN) have been widely applied. However, CNNs often struggle to capture long-range temporal dependencies in phenological data, which are crucial for modeling seasonal and cyclic patterns. The Transformer model complements this by leveraging self-attention mechanisms to effectively handle global contexts and extended sequences in phenology-related tasks. The Transformer model has the global understanding ability that CNN does not have due to its multi-head attention. This study, proposes a synergistic framework, in combining CNN with Transformer model to realize global-local feature synergy using two models, proposes an innovative phenological monitoring model utilizing near-ground remote sensing technology. High-resolution imagery of maize fields was collected using unmanned aerial vehicles (UAVs) equipped with multispectral and thermal infrared cameras. By integrating this data with CNN and Transformer architectures, the proposed model enables accurate inversion and quantitative analysis of maize phenological traits. In the experiment, a network was constructed adopting multispectral and thermal infrared images from maize fields, and the model was validated using the collected experimental data. The results showed that the integration of multispectral imagery and accumulated temperature achieved an accuracy of 92.9%, while the inclusion of thermal infrared imagery further improved the accuracy to 97.5%. This study highlights the potential of UAV-based remote sensing, combined with CNN and Transformer as a transformative approach for precision agriculture. Full article

The scale and magnitude of urban heating are often assessed using Satellite-Derived Land Surface Temperature (SD-LST). Yet, discrepancies in spatial resolution limit SD-LST’s ability to reflect pedestrian thermal experience, potentially leading to ineffective mitigation strategies. Hyper-local measurements of urban heat, defined as surface temperatures (T_S) at the scale of pedestrian activity (e.g., bus stops or street segments), may provide more accurate insights into thermal comfort. This study compares hyper-local ~0.01 m resolution T_S collected via consumer-grade Forward-Looking Infrared (FLIR) thermography with resampled 30 m resolution SD-LST from Landsat 8 and 9 images to evaluate their utility in predicting thermal comfort indices across 60 bus stops in Denver, Colorado. During the summer of 2023, 270 FLIR measurements were collected over 19 dates, with a four-day subset (n = 33) coinciding with Landsat imagery. FLIR T_S averaged 25.12 ± 5.39 °C, while SD-LST averaged 35.90 ± 12.56 °C, a significant 10.77 °C difference (95% CI: 6.81–14.73; p < 0.001). FLIR T_S strongly correlated with biometeorological metrics such as air temperature and mean radiant temperature (r > 0.8; p < 0.001), while SD-LST correlations were weak (r < 0.3). Linear mixed-effects models using FLIR T_S explained 50–66% of the variance in thermal comfort indices and met ISO 7726 standards. Each 1 °C increase in FLIR TS predicted a 0.75 °C rise in mean radiant temperature. These results highlight hyper-local thermography as a reliable, low-cost tool for urban heat resilience planning. Full article

The growing global reliance on solar photovoltaic (PV) systems requires robust, automated inspection techniques to ensure reliability and efficiency. Thermal infrared (IR) imaging is widely used for detecting PV faults; however, accurate classification remains challenging due to severe class imbalance, low thermal contrast, and high inter-class visual similarity among fault types. This study proposes a hierarchical deep learning framework for thermal PV fault classification, integrating a multi-class dataset-balancing strategy to enhance representational efficiency. The proposed framework consists of two major components: (i) a hierarchical two-stage classification scheme that mitigates data imbalance and leverages limited labeled data for improved fault discrimination; and (ii) a contrast-preserving MixUp augmentation technique designed explicitly for low-contrast thermal imagery, improving minority fault class recognition and overall robustness. Comprehensive experiments were conducted on benchmark 8-class thermal PV datasets using nine deep network architectures. Dataset refactoring decisions are validated through quantitative inter-class distance analysis using multiple complementary metrics. Results demonstrate that the proposed hierarchical SlantNet model achieves the best trade-off between accuracy and computational efficiency, achieving an F1-Efficiency Index of 337.6 and processing 42,072 images per second on a GPU, over twice the efficiency of conventional approaches. Comparatively, the Swin-T Transformer attained the highest classification accuracy of 89.48% and F1 score of 80.50%, while SlantNet achieved 86.15% accuracy and 73.03% F1 score with substantially higher inference speed, highlighting its real-time potential. Ablation studies on augmentation and regularization strategies confirm that the proposed techniques significantly improve minority class detection without compromising overall performance, with detailed per-class precision, recall, and F1 analysis. The proposed framework delivers a high-accuracy, low-latency, and edge-deployable solution for automated PV inspection, facilitating seamless integration into operational PV plants for real-time fault diagnosis. Full article

High-resolution TIR missions require sustained and well-characterized radiometric accuracy to support applications such as land surface temperature retrieval, drought monitoring, and surface energy budget analysis. To address this need, we develop an operational and automated ground-based vicarious radiometric calibration framework for TIR sensors and demonstrate its performance using the Wide-swath Thermal Infrared Imager (WTI) onboard Gaofen-5 01A (GF-5A). Three arid Gobi calibration sites were selected by integrating Moderate Resolution Imaging Spectroradiometer (MODIS) cloud products, Shuttle Radar Topography Mission (SRTM)-derived topography, and WTI-based radiometric uniformity metrics to ensure low cloud cover, flat terrain, and high spatial homogeneity. Automated ground stations deployed at Golmud, Dachaidan, and Dunhuang have continuously recorded 1 min contact surface temperature since October 2023. Field-measured emissivity spectra, Integrated Global Radiosonde Archive (IGRA) radiosonde profiles, and MODTRAN (MODerate resolution atmospheric TRANsmission) v5.2 simulations were combined to compute top-of-atmosphere (TOA) radiances, which were subsequently collocated with WTI imagery. After data screening and gain-stratified regression, linear calibration coefficients were derived for each TIR band. Based on 189 scenes from February–July 2024, all four bands exhibit strong linearity (R-squared greater than 0.979). Validation using 45 independent scenes yields a mean brightness–temperature root-mean-square error (RMSE) of 0.67 K. A full radiometric-chain uncertainty budget—including contact temperature, emissivity, atmospheric profiles, and radiative transfer modeling—results in a combined standard uncertainty of 1.41 K. The proposed framework provides a low-maintenance, traceable, and high-frequency solution for the long-term on-orbit radiometric calibration of GF-5A WTI and establishes a reproducible pathway for future TIR missions requiring sustained calibration stability. Full article

Peatlands play a critical role in global and regional climate regulation by functioning as long-term carbon sinks, regulating hydrology, and modulating land–atmosphere energy exchange. Intact peat ecosystems store large amounts of organic carbon and stabilize local climate through high water retention and evapotranspiration, whereas peatland degradation disrupts these functions and can transform peatlands into significant sources of greenhouse gas emissions and climate extremes such as drought and fire. Indonesia contains approximately 13.6–40.5 Gt of carbon, around 40% of which is stored on the island of Sumatra. However, tropical peatlands in this region are highly vulnerable to climate anomalies and land-use change. This study investigates the impacts of major climate anomalies—specifically El Niño and positive Indian Ocean Dipole (pIOD) events in 1997/1998, 2015/2016, and 2019—on peatland cover change across South Sumatra, Jambi, Riau, and the Riau Islands. Landsat 5 Thematic Mapper and Landsat 8 Operational Land Imager/Thermal Infrared Sensor imagery were analyzed using a Random Forest machine learning classification approach. Climate anomaly periods were identified using El Niño-Southern Oscillation (ENSO) and IOD indices from the National Oceanic and Atmospheric Administration. To enhance classification accuracy and detect vegetation and hydrological stress, spectral indices including the Normalized Difference Vegetation Index (NDVI), Modified Soil Adjusted Vegetation Index (MSAVI), Normalized Difference Water Index (NDWI), and Normalized Difference Drought Index (NDDI) were integrated. The results show classification accuracies of 89–92%, with kappa values of 0.85–0.90. The 2015/2016 El Niño caused the most severe peatland degradation (>51%), followed by the 1997/1998 El Niño (23–38%), while impacts from the 2019 pIOD were comparatively limited. These findings emphasize the importance of peatlands in climate regulation and highlight the need for climate-informed monitoring and management strategies to mitigate peatland degradation and associated climate risks. Full article

The built environment is responsible for 40% of global energy demand, and, in line with urbanisation and population growth, this demand is expected to increase steadily. Urban areas are mostly composed of materials that can absorb energy from solar radiation and dissipate the accumulated energy in the form of heat. This study integrates a UAV-based Zenmuse XT S IR camera and handheld FLIR C5 thermal camera with ENVI-met microclimate simulation, providing quantitative insights for sustainable urban planning. From the 24 h experiment results, the characteristics of building surface materials are profiled for lowering energy use for internal thermal control during the operation stage of buildings. This study shows that building surface materials with the lowest solar reflectance and highest specific heat capacity reached a peak surface temperature of 73.5 °C in Jakarta (tropical hot climate) and 44.3 °C in Xiamen (subtropical late winter climate). In contrast, materials with the highest solar reflectance and lowest specific heat only reach a peak surface temperature of 58.1 °C in Jakarta and 27.9 °C in Xiamen. The peak surface temperature occurs at 2 PM in the afternoon. Moreover, we demonstrate the capability of an infrared drone to identify the peak surface temperatures of 55.8 °C at 2 PM in the study area in Xiamen. In addition, the ENVI-met validated model shows satisfactory correlation values of R > 0.9 and R2 > 0.8. This result demonstrates UAV-IR and ENVI-met simulation integration as a scalable method for city-level UHI diagnostics and monitoring. Full article

Effectively leveraging multi-source unmanned aerial vehicle (UAV) observations for reliable object recognition is often compromised by environmental extremes (e.g., occlusion and low illumination) and the inherent physical discrepancies between modalities. To overcome these limitations, we propose DLANet, a lightweight, rotation-aware multimodal object detection framework that introduces a dual-level attention relearning strategy to maximize complementary information from visible (RGB) and thermal infrared (TIR) imagery. DLANet integrates two novel components: the Implicit Fine-Grained Fusion Module (IF²M), which facilitates deep cross-modal interaction by jointly modeling channel and spatial dependencies at intermediate stages, and the Adaptive Branch Feature Weighting (ABFW) module, which dynamically recalibrates modality contributions at higher levels to suppress noise and pseudo-targets. This synergistic approach allows the network to relearn feature importance based on real-time scene conditions. To support industrial applications, we construct the OilLeak dataset, a dedicated benchmark for onshore oil-spill detection. The experimental results demonstrate that DLANet achieves state-of-the-art performance, recording an mAP_0.5 of 0.858 on the public DroneVehicle dataset while maintaining high efficiency, with 39.04 M parameters and 72.69 GFLOPs, making it suitable for real-time edge deployment. Full article

Photovoltaic systems (PV) are increasingly recognized as fundamental to the worldwide adoption of renewable energy technologies. Nonetheless, the efficiency and longevity of solar panels can be compromised by various anomalies, ranging from physical defects to environmental impacts. Early and accurate detection of these anomalies is crucial for maintaining optimal performance and preventing significant energy losses. This study presents SolarAttnNet, a novel convolutional neural network (CNN) architecture with integrated channel and spatial attention mechanisms for solar panel anomaly detection. The proposed model addresses the critical need for automated detection systems, which are crucial for maintaining energy production efficiency and optimizing maintenance. This approach leverages attention mechanisms that emphasize the most relevant features within thermal and visual imagery, improving detection accuracy across multiple anomaly types. SolarAttnNet is evaluated on three distinct solar panel datasets, demonstrating its effectiveness through comprehensive ablation studies that isolate the contribution of each architectural component. Experimental results show that SolarAttnNet achieves superior performance compared to state-of-the-art methods, with accuracy improvements of 3.9% on the PV Systems-AD dataset (94.2% vs. 90.3%), 3.6% on the InfraredSolarModules dataset (92.1% vs. 88.5%), and 3.5% on the RoboflowAnomalies dataset (89.7% vs. 86.2%) compared to baseline ResNet-50. For challenging subtle anomalies like cell cracks and PID, the proposed model demonstrates even more significant improvements with F1-score gains of 4.8% and 5.4%, respectively. Ablation studies reveal that the channel attention mechanism contributes a 2.6% accuracy improvement while spatial attention adds 2.3% across datasets. This work contributes to advancing automated inspection technologies for renewable energy infrastructure, supporting more efficient maintenance protocols and ultimately enhancing solar energy production. Full article

Tracking small, fast-moving unmanned aerial vehicles (UAVs) in thermal infrared (TIR) imagery is a significant challenge due to low-resolution targets, Dynamic Background Clutter, and frequent occlusions. To address this, we introduce MemLoTrack, a novel onestream Vision Transformer tracker that integrates a memory mechanism into a parameterefficient LoRA framework. MemLoTrack enhances a baseline tracker (LoRAT) with two key components: (i) a gated First-In, First-Out (FIFO) memory bank (MB) for temporal context aggregation and (ii) a lightweight Memory Attention Layer (MAL) for effective information retrieval. A key component of our method is a selective memory update policy, which commits a frame to the memory bank only when it satisfies both a classification confidence threshold ($\tau$) and a Kalman filter-based motion consistency check. This gating mechanism robustly prevents memory contamination due to distractors, occlusions, and reappearance events. Our training is highly efficient, updating only the LoRA adapters, MAL, and prediction head while the pretrained DINOv2 backbone remains frozen. Evaluated on the challenging Anti-UAV410 benchmark, MemLoTrack (L_mem = 7, $\tau$ = 0.8) achieves an AUC of 63.6 and a State Accuracy (SA) of 64.0, representing a significant improvement over the LoRAT baseline by +1.4 AUC and +1.5 SA. Compared to the state-of-the-art method FocusTrack, MemLoTrack demonstrates superior robustness with higher AUC (63.6 vs. 62.8) and SA (64.0 vs. 63.9), while trading lower precision (P/P-Norm) scores. Furthermore, MemLoTrack operates at 153 FPS on a single RTX 4070 Ti SUPER, demonstrating that parameter-efficient fine-tuning with a selective memory mechanism is a powerful and deployable strategy for real-time Anti-UAV tracking in demanding TIR environments. Full article

To mitigate the limitations of low measurement accuracy and substantial environmental interference in UAV-based TIR imaging for coal fire monitoring, this study presents an integrated temperature correction approach, termed NRBO-XGBoost. The proposed method applies temperature correction to TIR imagery and subsequently investigates coal fire detection using the corrected TIR data. By leveraging multi-source data (thermal infrared measurements, UAV flight altitude, and meteorological parameters), the NRBO optimizes XGBoost hyperparameters to improve model convergence speed and global search capability, effectively overcoming the limitations of traditional methods, such as local optima entrapment and poor generalization. Experimental results demonstrate that the NRBO-XGBoost model achieves superior performance in temperature correction, with a coefficient of determination (R²) of 0.9993, while reducing RMSE and MAE by 85.6% and 86.6%, respectively. Notably, the model exhibits enhanced stability in high-temperature regions (>300 °C). The 3D reconstruction results demonstrate a nearly 6-fold expansion in high-temperature area coverage (from 0.43% to 2.60%), coupled with a morphological transformation of fragmented hotspots into continuous, belt-shaped distributions. Integrating visible-light textures further improves boundary clarity and spatial semantic representation of thermal anomalies. This study provides a high-precision temperature correction and 3D visualization solution for coal fire monitoring, offering critical technical support for early warning systems and firefighting strategies. Full article

Infrared–visible (IR–VIS) image fusion is becoming central to unmanned aerial vehicle (UAV) perception, enabling robust operation across day–night cycles, backlighting, haze or smoke, and large viewpoint or scale changes. However, for practical applications some challenges still remain: visible images are illumination-sensitive; infrared imagery suffers thermal crossover and weak texture; motion and parallax cause cross-modal misalignment; UAV scenes contain many small or fast targets; and onboard platforms face strict latency, power, and bandwidth budgets. Given these UAV-specific challenges and constraints, we provide a UAV-centric synthesis of IR–VIS fusion. We: (i) propose a taxonomy linking data compatibility, fusion mechanisms, and task adaptivity; (ii) critically review learning-based methods—including autoencoders, CNNs, GANs, Transformers, and emerging paradigms; (iii) compare explicit/implicit registration strategies and general-purpose fusion frameworks; and (iv) consolidate datasets and evaluation metrics to reveal UAV-specific gaps. We further identify open challenges in benchmarking, metrics, lightweight design, and integration with downstream detection, segmentation, and tracking, offering guidance for real-world deployment. A continuously updated bibliography and resources are provided and discussed in the main text. Full article

This study addresses the observation of floating plastic debris in freshwater environments using an Unmanned Aerial Vehicle (UAV) multi-sensor strategy. An experimental campaign is described where an heterogeneous plastic assemblage, namely a plastic target, and a naturally occurring leaf-litter mat are observed by a UAV platform in the Lake Calore (Avellino, Southern Italy) within the framework of the “multi-layEr approaCh to detect and analyze cOastal aggregation of MAcRo-plastic littEr” (ECOMARE) Italian Ministry of Research (MUR)-funded project. Three UAV platforms, equipped with optical, multispectral, and thermal sensors, are adopted, which overpass the two targets with the objective of analyzing the sensitivity of optical radiation to plastic and the possibility of discriminating the plastic target from the natural one. Georeferenced orthomosaics are generated across the visible, multispectral (Green, Red, Red Edge, Near-Infrared—NIR), and thermal bands. Two novel indices, the Plastic Detection Index (PDI) and the Heterogeneity Plastic Index (HPI), are proposed to discriminate between the detection of plastic litter and natural targets. The experimental results highlight that plastics exhibit heterogeneous spectral and thermal responses, whereas natural debris showed more homogeneous signatures. Green and Red bands outperform NIR for plastic detection under freshwater conditions, while thermal imagery reveals distinct emissivity variations among plastic items. This outcome is mainly explained by the strong NIR absorption of water, the wetting of plastic surfaces, and the lower sensitivity of the Mavic 3′s NIR sensor under high-irradiance conditions. The integration of optical, multispectral, and thermal data demonstrate the robustness of UAV-based approaches for distinguishing anthropogenic litter from natural materials. Overall, the findings underscore the potential of UAV-mounted remote sensing as a cost-effective and scalable tool for the high-resolution monitoring of plastic pollution over inland waters. Full article

Embankment piping and leakage are primary causes of flood control infrastructure failure, accounting for more than 90% of embankment failures worldwide and posing significant threats to public safety and economic stability. Current manual inspection methods are labor-intensive, hazardous, and inadequate for emergency flood season monitoring, while existing automated approaches using thermal infrared imaging face limitations in cost, weather dependency, and deployment flexibility. This study addresses the critical scientific challenge of developing reliable, cost-effective automated detection systems for embankment safety monitoring using Unmanned Aerial Vehicle (UAV)-based visible light imagery. The fundamental problem lies in extracting subtle textural signatures of piping and leakage from complex embankment surface patterns under varying environmental conditions. To solve this challenge, we propose the Embankment-Frequency Network (EmbFreq-Net), a frequency-enhanced deep learning framework that leverages frequency-domain analysis to amplify hazard-related features while suppressing environmental noise. The architecture integrates dynamic frequency-domain feature extraction, multi-scale attention mechanisms, and lightweight design principles to achieve real-time detection capabilities suitable for emergency deployment and edge computing applications. This approach transforms traditional post-processing workflows into an efficient real-time edge computing solution, significantly improving computational efficiency and enabling immediate on-site hazard assessment. Comprehensive evaluations on a specialized embankment hazard dataset demonstrate that EmbFreq-Net achieves 77.68% mAP@0.5, representing a 4.19 percentage point improvement over state-of-the-art methods, while reducing computational requirements by 27.0% (4.6 vs. 6.3 Giga Floating-Point Operations (GFLOPs)) and model parameters by 21.7% (2.02M vs. 2.58M). These results demonstrate the method’s potential for transforming embankment safety monitoring from reactive manual inspection to proactive automated surveillance, thereby contributing to enhanced flood risk management and infrastructure resilience. Full article

Fine-grained identification of crop planting structures provides key data for precision agriculture, thereby supporting scientific production and evidence-based policy making. This study selected a representative experimental farmland in Qingyang, Gansu Province, and acquired Unmanned Aerial Vehicle (UAV) multi-temporal data (six epochs) from multiple sensors (multispectral [visible–NIR], thermal infrared, and LiDAR). By fusing 59 feature indices, we achieved high-accuracy extraction of cropland and planting structures and identified the key feature combinations that discriminate among crops. The results show that (1) multi-source UAV data from April + June can effectively delineate cropland and enable accurate plot segmentation; (2) July is the optimal time window for fine-scale extraction of all planting-structure types in the area (legumes, millet, maize, buckwheat, wheat, sorghum, maize–legume intercropping, and vegetables), with a cumulative importance of 72.26% for the top ten features, while the April + June combination retains most of the separability (67.36%), enabling earlier but slightly less precise mapping; and (3) under July imagery, the SAM (Segment Anything Model) segmentation + RF (Random Forest) classification approach—using the RF-selected top 10 of the 59 features—achieved an overall accuracy of 92.66% with a Kappa of 0.9163, representing a 7.57% improvement over the contemporaneous SAM + CNN (Convolutional Neural Network) method. This work establishes a basis for UAV-based recognition of typical crops in the Qingyang sector of the Loess Plateau and, by deriving optimal recognition timelines and feature combinations from multi-epoch data, offers useful guidance for satellite-based mapping of planting structures across the Loess Plateau following multi-scale data fusion. Full article

Severe water scarcity has become a critical constraint to global agricultural development. Enhancing both the timeliness and accuracy of crop evapotranspiration (ET_c) retrieval is essential for optimizing irrigation scheduling. Addressing the limitations of conventional ground-based point-source measurements in rapidly acquiring two-dimensional ET_c information at the field scale, this study employed unmanned aerial vehicle (UAV) remote sensing equipped with multispectral and thermal infrared sensors to obtain high spatiotemporal resolution imagery of a representative citrus orchard (Citrus reticulata Blanco cv. ‘Yichangmiju’) in western Hubei at different phenological stages. In conjunction with meteorological data (air temperature, daily net radiation, etc.), ET_c was retrieved using two established approaches: the Seguin-Itier (S-I) model, which relates canopy–air temperature differences to ET_c, and the multispectral-driven single crop coefficient method, which estimates ET_c by combining vegetation indices with reference evapotranspiration. The thermal-infrared-driven S-I model, which relates canopy–air temperature differences to ET_c, and the multispectral-driven single crop coefficient method, which estimates ET_c by combining vegetation indices with reference evapotranspiration. The findings indicate that: (1) both the S-I model and the single crop coefficient method achieved satisfactory ET_c estimation accuracy, with the latter performing slightly better (accuracy of 80% and 85%, respectively); (2) the proposed multi-source fusion model consistently demonstrated high accuracy and stability across all phenological stages (R² = 0.9104, 0.9851, and 0.9313 for the fruit-setting, fruit-enlargement, and coloration–sugar-accumulation stages, respectively; all significant at p < 0.01), significantly enhancing the precision and timeliness of ET_c retrieval; and (3) the model was successfully applied to ET_c retrieval during the main growth stages in the Cangwubang citrus-producing area of Yichang, providing practical support for irrigation scheduling and water resource management at the regional scale. This multi-source fusion approach offers effective technical support for precision irrigation control in agriculture and holds broad application prospects. Full article