Remote Sensing Applications-Society and Environment最新文献

{"title":"Estimating vegetation temperature from UAV multispectral imagery-based vegetation indices","authors":"Andres Montes de Oca ,&nbsp;Gerardo Flores","doi":"10.1016/j.rsase.2025.101711","DOIUrl":"10.1016/j.rsase.2025.101711","url":null,"abstract":"<div><div>In agricultural research, the computation of temperature and water content indicators, such as the Crop Water Stress Index (CWSI), relies on expensive and specialized thermal imaging devices. To overcome this limitation, this study presents a novel and cost-effective methodology for precise temperature estimation. By using an affordable multispectral imaging system, the objective is to provide growers with low-cost Unmanned Aerial Systems (UAS) capable of estimating vegetation temperature without the need for thermal imagery. This investigation delves into the relationship between multispectral imagery-based vegetation indices and temperature derived from thermal imagery. After correcting and calibrating these data sources, an estimation model is established to compute vegetation temperature using only visible and near-infrared (NIR) radiation, effectively eliminating the need for thermal imagery. Among various vegetation indices tested, the <em>Green Chlorophyll Index</em> (GCI) demonstrates the highest correlation with ground truth temperature (R² = 0.71) in vegetation regions including a park and a cornfield. Consequently, GCI is used to compute the temperature estimate map and derive a CWSI estimate, which entirely foregoes thermal imagery. Rigorous quantitative comparisons are made between ground truth and estimated temperature to validate the accuracy of the results. Although the proposed approach is currently in the early stages, it appears promising as a practical tool for growers to assess water content features at low and high resolutions without compromising accuracy compared to the traditional thermal-based method. The open-source software developed for this research is available online as supplementary material, fostering transparency and repeatability.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101711"},"PeriodicalIF":4.5,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145110172","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A lightweight Dual-Stream Attention Network for real-time landslide monitoring in multi-modal remote sensing imagery","authors":"Pooja Dhayal ,&nbsp;Pradeep Singh ,&nbsp;Kanishk Sharma ,&nbsp;Samarpita Sarkar ,&nbsp;Dhani Ram Rajak ,&nbsp;Alok Bhardwaj ,&nbsp;Balasubramanian Raman","doi":"10.1016/j.rsase.2025.101732","DOIUrl":"10.1016/j.rsase.2025.101732","url":null,"abstract":"<div><div>Rapid delineation of landslides from post-event remote-sensing imagery demands models that can be <em>deployed in the field</em>, often on battery-powered edge devices with strict memory and latency budgets. Most state-of-the-art detectors break those constraints, shipping tens of millions of parameters to gain marginal accuracy. We therefore introduce a <em>Dual-Stream Attention Network</em> whose total trainable footprint is just <em>2.03</em> <!-->M parameters – roughly the size of a single JPEG image – yet still exploits the complementary physics of co-registered RGB orthomosaics and digital-elevation models. Two light encoders process the modalities independently; a provably isometric late-fusion operator merges their hierarchies without information loss, and a <span><math><mrow><mo>(</mo><mo>≤</mo><mn>1</mn><mo>)</mo></mrow></math></span>-Lipschitz spatial-channel gate discards redundant features while preserving gradient stability. Despite its size, the network attains an IoU of 0.835 and an mIoU of 0.818 on the high-resolution Bijie benchmark, coming within 4–5 percentage points of heavyweight baselines such as DeepLabv3^{protect relax special {t4ht=＋}}(R-101) while using <span><math><mrow><mo>≈</mo><mspace></mspace><mn>95</mn><mtext>%</mtext></mrow></math></span> fewer weights. Ablation studies show that each architectural choice – dual-stream processing, late fusion, and attentional gating – contributes at least <span><math><mrow><mo>+</mo><mn>1</mn><mo>.</mo><mn>5</mn></mrow></math></span> pp IoU. A single Jetson Xavier AGX segments a 256 × 256 tile in 1.6 s (<span><math><mrow><mo>&lt;</mo><mn>10</mn></mrow></math></span> W envelope), confirming real-time suitability for rapid landslide mapping missions. By reconciling DEM-derived information with extreme parameter efficiency, the proposed architecture offers a practical foundation for next-generation, on-device geohazard monitoring systems.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101732"},"PeriodicalIF":4.5,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145109072","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Upscaling CO2 fluxes from agricultural drained lowland peatlands in England using remote sensing and machine learning","authors":"Asima Khan ,&nbsp;Muhammad Ali ,&nbsp;Joerg Kaduk ,&nbsp;Ashiq Anjum ,&nbsp;Heiko Balzter","doi":"10.1016/j.rsase.2025.101728","DOIUrl":"10.1016/j.rsase.2025.101728","url":null,"abstract":"<div><div>Drained lowland peatlands in the UK are used as prime agricultural areas but are significant sources of CO₂ emissions. Monitoring and quantifying CO₂ dynamics in these ecosystems is critical to achieving the UK’s legal net-zero target by 2050. This study pioneers the upscaling of carbon fluxes (Gross Ecosystem Productivity (GEP), Total Ecosystem Respiration (TER), and Net Ecosystem Exchange of CO₂ (NEE)) in East Anglia’s agricultural peatlands (England) using remote sensing (RS) and machine learning (ML). A Random Forest model, trained with Landsat and Sentinel-2 imagery, meteorological data, and soil carbon information, predicts field-scale CO₂ fluxes with 77% overall accuracy. TER prediction was the strongest (R² = 0.84; RMSE = 1.18 gC/m²/d; NRMSE = 8%), followed by NEE (R² = 0.77; RMSE = 1.37 gC/m²/d; NRMSE = 8.13%), and GEP (R² = 0.76, RMSE = 1.97 gC/m²/d; NRMSE = 9.87%). The average predictive uncertainty for 14-day fluxes was <span><math><mrow><mo>±</mo><mn>1</mn><mo>.</mo><mn>69</mn></mrow></math></span> gC/m²/d, which scaled with magnitude. The model was more accurate in grasslands compared to croplands. We validated the model with spatial cross-validation, finding it accurately predicts NEE seasonality at an unseen grassland site but deviates from observed mean values in winter and spring. We demonstrate the applicability of the model by upscaling annual and seasonal fluxes across the Fens, where the annual NEE in 2023 ranged from 1.04 to -2.52 kgC/m², depicting high spatial variability. This study establishes a baseline NEE scenario for the Fens and lays the groundwork for refining CO₂ flux modelling in drained peatlands, highlighting the potential of RS and ML for supporting the UK’s GHG reduction strategies in peatland ecosystems.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101728"},"PeriodicalIF":4.5,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145120952","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Farmland parcel extraction and area calculation from UAV images based on semantic segmentation","authors":"Zhongzhou Su,&nbsp;Kai Chen,&nbsp;Mengmeng Liu","doi":"10.1016/j.rsase.2025.101734","DOIUrl":"10.1016/j.rsase.2025.101734","url":null,"abstract":"<div><div>In irrigation management of smart farming, the accurate and efficient calculation of farmland parcel area is regarded as a critical component, by which the optimization of water resource allocation and the enhancement of agricultural production efficiency are significantly promoted. This study proposed a method for farmland parcel extraction and area calculation from Unmanned Aerial Vehicle (UAV) images based on semantic segmentation. First, a theoretical dynamic pixel adjustment model was established based on camera imaging principles to improve the calculation method for farmland parcel area. Then, farmland parcel extraction from UAV images was performed by semantic segmentation, through which the efficiency of area calculation was enhanced. Finally, verification of the proposed method and its improved algorithm’s accuracy and applicability was conducted by utilizing self-built farmland parcel datasets and multi-altitude aerial image datasets. Experimental results indicated that the accuracy of parcel segmentation and that of area calculation methods exert a mutual influence on each other, with their relative importance varying across different stages of the workflow and the calculation accuracy and efficiency of farmland parcel area was significantly improved by the proposed method. In the task of farmland parcel extraction, the semantic segmentation model based on Deeplabv3+ resulted excellent performance, achieving a test-set F1 score, Miou, OA, precision and recall of the test set are 96.60 %, 95.67 %, 97.83 %, 97.95 % and 98.41 %. An average relative error of 1.2 % is maintained by the improved algorithm across the aerial altitude range of 22–49m. In the range of 84–121m altitude, reductions of 84.56 % in mean squared error (MSE) and 64.46 % in mean absolute error (MAE) were achieved when compared with traditional methods. Comparative analysis with measured area data demonstrated that the area calculation error of the proposed method is strictly constrained within a 4 % threshold, satisfying with the precision standards of agricultural engineering.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101734"},"PeriodicalIF":4.5,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145110171","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Warm-dry climate narrowed the difference between vegetation greenness and photosynthetic phenology periods in the Yellow River Basin, China","authors":"Yan Liu ,&nbsp;Jialin Bi ,&nbsp;Xiaoyu Cai ,&nbsp;Liping Wang ,&nbsp;Guoqing Li ,&nbsp;Xiaohui Wang ,&nbsp;Xianbin Liu ,&nbsp;Zhanpeng Xu ,&nbsp;Xiang Yu ,&nbsp;Chao Zhan","doi":"10.1016/j.rsase.2025.101735","DOIUrl":"10.1016/j.rsase.2025.101735","url":null,"abstract":"<div><div>China has got a great achievement on ecological restoration in the Yellow River basin (YRB), which owned an extremely frail ecosystem in history. Nowadays, warm and dry climate has threatened the ecosystem there. Change of vegetation phenology act as an effective indicator for ecosystem balance against climate change. Thus, the periods of two vegetation phenology, vegetation green phenology periods (VGPPs) and vegetation photosynthetic phenology periods (VPPPs), were gained and compared here to identify the response of vegetation ecosystem to warm and dry climate in YRB. Sun-induced chlorophyll fluorescence (SIF) and NDVI were used to gain the two VGPPs with SG-filtering and dynamic thresholding method. The results showed that: 1) the differences between VGPPs and VPPPs gradually reduced with the warm-dry climate in the YRB from 2001 to 2022. Specifically, the growing season length (GSL) of VGPPs was shortened by 5 days, while VPPPs was lengthened by 11 days from 2001 to 2022. Thus, the difference between the duration of VGPPs and that of VPPPs was reduced from 40 d to 24 d, with an obviously declined trend (<em>k</em> = −0.74); 2) the start of growing season (SOS) appeared earlier in the southern and southeastern regions, while it occurred later in the western and northern regions. The end of growing season (EOS) showed the opposite pattern compared to SOS; 3) for VGPPs and VPPPs, temperature was the dominated factor for SOS, soil water was the dominated factor for EOS. Besides that, VPPPs were more sensitive than VGPPs to temperature, soil water (SW), precipitation and photosynthetically active radiation (PAR).</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101735"},"PeriodicalIF":4.5,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145110174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Vessel detection leveraging satellite imagery and YOLO in maritime surveillance","authors":"Rita Magalhães ,&nbsp;Ana Paula Falcão ,&nbsp;Alberto Barbosa","doi":"10.1016/j.rsase.2025.101730","DOIUrl":"10.1016/j.rsase.2025.101730","url":null,"abstract":"<div><div>Vessel detection is essential for maritime surveillance, supporting the monitoring of fishing, commercial, and transportation activities, as well as detecting suspicious or illegal vessels and aiding search and rescue operations. Satellite imagery has become increasingly valuable for Earth Observation applications due to its wide range of spatio-temporal cover. The European Space Agency offers free optical imagery from the Sentinel-2 satellite, providing a cost-effective solution for large-scale maritime surveillance. Deep learning models, particularly You Only Look Once (YOLO), have demonstrated impressive object detection performance and are widely regarded as state-of-the-art for real-time applications. This article addresses an existing research gap by comparing YOLO versions 8 and 10 for vessel and wake detection using a dataset provided by CEiiA. The models were tested to determine the optimal configuration for this task and after evaluation, and a YOLOv8 configuration was selected. Following this selection, the dataset was expanded leading to further performance improvements. The final YOLOv8 model achieved an F1-score of 88.69 % (with 90.47 % precision and 86.98 % recall), an IoU of 72.91 %, a mAP50 of 87.53 % and mAP50-95 of 61.20 %. Single class training improved bounding box localization and delineation, however it slightly reduced performance, possibly due to a loss of contextual information. Additionally, testing on high-resolution images confirmed that models trained on lower-resolution data can still perform relatively well on higher-resolution images, with potential for further improvement through fine-tuning. These results support that YOLO models are highly effective for real-time vessel detection and can be reliably applied in maritime surveillance.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101730"},"PeriodicalIF":4.5,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222199","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Contribution of high-resolution remote sensing to spatial ecology of forest ecosystems at the single tree level: A systematic review","authors":"Yousef Erfanifard ,&nbsp;Matteo Garbarino ,&nbsp;Krzysztof Stereńczak","doi":"10.1016/j.rsase.2025.101733","DOIUrl":"10.1016/j.rsase.2025.101733","url":null,"abstract":"<div><div>Forest spatial ecology investigates the complex relationships between spatial patterns and ecological processes, offering critical insights into forest ecosystem dynamics. This review synthesizes findings from 66 studies, highlighting the growing significance of high-resolution remote sensing (HR-RS) technologies in the field. HR-RS is particularly valuable for capturing tree–tree interactions and tree–environment relationships that are difficult to detect using traditional field methods, especially in large or densely vegetated forests. HR-RS datasets, including imagery and point clouds, enable spatially explicit measurements of individual trees, capturing both quantitative attributes (e.g., height, crown size) and qualitative characteristics (e.g., species, health status). Among the reviewed studies, 35 % employed aerial imagery to detect features such as canopy gaps, snags, and pest outbreaks, while 40 % utilized point pattern analysis to assess tree–tree ecological interactions. LiDAR was widely used for its ability to represent forest 3D structure and biophysical attributes. Notably, 45.5 % of the studies focused on tree–environment relationships, using HR-RS to map environmental variables such as soil moisture and microclimate conditions. However, advanced technologies such as multispectral and hyperspectral LiDAR remain underutilized, revealing a gap in current research. To advance forest spatial ecology, future studies should prioritize multisensor data fusion, longitudinal UAV–LiDAR monitoring, and advanced 3D spatial analyses. The integration of machine learning and deep learning techniques will also be essential for improving tree classification and detecting spatial patterns, ultimately deepening our understanding of forest ecological processes.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101733"},"PeriodicalIF":4.5,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145110175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of a litho-structural map for the Upper Mereb area, Eritrea, using multi-source remote sensing data and machine learning models","authors":"Kabral Mogos Asghede ,&nbsp;Abazar M.A. Daoud ,&nbsp;Musaab A.A. Mohammed ,&nbsp;Woldegabriel Genzebu ,&nbsp;Kefela Beyene Kiflay ,&nbsp;Péter Pecsmány ,&nbsp;János Vágó","doi":"10.1016/j.rsase.2025.101722","DOIUrl":"10.1016/j.rsase.2025.101722","url":null,"abstract":"<div><div>The Upper Mereb catchment area, located on the Southern zone of Eritrea, is a geologically complex region within the Arabian-Nubian Shield (ANS). The area's intricate litho-structural framework presents significant challenges for mineral exploration and groundwater investigations. Traditional geological mapping techniques often struggle to capture the fine-scale structural details necessary for resource assessments in such complex terrains. This study introduces a novel, high-resolution litho-structural mapping approach, integrating Landsat 9 (L9) multispectral data and gravity data with advanced machine learning algorithms, specifically Artificial Neural Networks (ANN) and Support Vector Machines (SVM). The classification results indicate that ANN outperforms SVM, achieving an accuracy exceeding 79 %, demonstrating the effectiveness of machine learning in distinguishing lithological units. Furthermore, detailed field investigations validate the accuracy of the litho-structural map, showing strong correlations with ground-truth data. A key component of this study is the structural analysis of lineament orientations, which provides critical insights into the tectonic evolution of the region. The Pan-African orogeny has significantly influenced the structural framework, with dominant NE-SW compressional forces creations the fracture patterns. The identified lineaments fall into three primary sets: NW-SE extensional fractures, indicative of crustal stretching; NE-SW release fractures, reflecting zones of stress relaxation; and N-S shear fractures, formed under oblique stress conditions. These structural features highlight the region's complex deformation history and provide essential information for understanding subsurface fluid flow and resource potential. This study represents the first comprehensive application of an integrated remote sensing and geophysical machine learning approach to geological mapping in the Upper Mereb area. The results emphasize the potential of hybrid remote sensing and geophysical data fusion for enhancing structural interpretations, offering a powerful tool for mineral exploration, groundwater assessments, and tectonic studies in the Arabian-Nubian Shield.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101722"},"PeriodicalIF":4.5,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimised multi-hierarchical feature fusion with multi-kernel CNN and spectral-spatial convolutions for remote sensing image classification","authors":"Chiagoziem C. Ukwuoma ,&nbsp;Dongsheng Cai ,&nbsp;Oluwatoyosi Bamisile ,&nbsp;Chibueze D. Ukwuoma ,&nbsp;Chinedu I. Otuka ,&nbsp;Nnadozie O. Anyanwu ,&nbsp;Chidera O. Ukwuoma ,&nbsp;Qi Huang","doi":"10.1016/j.rsase.2025.101727","DOIUrl":"10.1016/j.rsase.2025.101727","url":null,"abstract":"<div><div>Remote sensing image classification is a central process in the interpretation of Earth observation data for applications such as land use mapping, environmental monitoring, and urban planning. Despite significant progress brought about by deep learning, particularly Convolutional Neural Networks (CNNs), existing models are prone to failing to represent complex spatial patterns, multi-scale object variation, and rich spectral dependencies characteristic of high-resolution remote sensing images. In addition, most models are suffering from greater computational complexity and poor generalisation on diverse datasets. To address these issues, we propose an Optimised Multi-Hierarchical Feature Fusion Framework, a deep learning model that integrates multi-kernel convolution, spectral-spatial depthwise convolution, and residual learning into a ResNet-50 backbone. The model learns a rich set of spatial textures and spectral features at different hierarchical layers, leading to enhanced feature representation and classification robustness. We evaluated the proposed method on six remote sensing datasets: AID, EUROSAT, NWPU-RESISC45, RSSCN7, UC Merced, and WHU-RS19 with standard performance metrics like accuracy, precision, recall, specificity, and F1-score. The proposed method achieved an average accuracy of 99.00 % for AID, 99.51 % for EUROSAT, 99.56 % for RESISC45, 95.71 % for RSSCN7, 99.14 % for UC Merced, and 95.79 % for WHU-RS19. Moreover, qualitative visualisation techniques such as Class Activation Mapping (CAM) and LIME were employed to provide explanations of model decisions by identifying the most influential image regions to the predictions. These visual explanations confirmed that the model is based on semantically meaningful regions, further improving its interpretability and trustworthiness. The superior accuracy and robust performance on multiple datasets verify the efficiency and generalisation ability of the designed model, indicating it is a strong candidate for practical remote sensing classification tasks. <span><span>https://github.com/chiagoziemchima/Multi-Hierarchical-Feature-Fusion-/tree/main</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101727"},"PeriodicalIF":4.5,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145110173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Acquisition of oilseed rape seedling population based on visible light imagery from unmanned aerial vehicles","authors":"Xinbei Wei,&nbsp;Dongyang Zhen,&nbsp;Yang Yang,&nbsp;Yilin Ren","doi":"10.1016/j.rsase.2025.101717","DOIUrl":"10.1016/j.rsase.2025.101717","url":null,"abstract":"<div><div>Seedling density of oilseed rape has a significant effect on seed yield and quality, and accurate and timely estimation of seedling density in the field can guide later field management to ensure high yield and oil quality is of great significance. Currently, object detection is the most commonly used seedling counting method, which counts seedlings by accurately identifying them, but it will lead to unavoidable errors when the target overlap rate is large. This study aimed to reduce this error by obtaining seedling counts through regression prediction constructing a model of shape characteristics and seedling counts of oilseed rape seedlings in the field. Ultra-high resolution RGB images were captured using an unmanned aerial vehicle (UAV) during the growth stage when the oilseed rape seedlings had at least two leaves. The study used three regression modeling methods, namely, fully connected neural network (DNN), random forest algorithm (RF), and multiple linear regression (MLR), to construct a regression prediction model, and the threshold segmentation method (OTSU) was used to segment oilseed rape seedling targets from the processed images of the ExG color vegetation index, and the segmented oilseed rape seedling objects were extracted and selected as the total area of the connective domain, tal_are, the total outer connective rectangle The total area of the connected domain tal_are, the total perimeter of the outer rectangle S, and the total height of the outer rectangle h-all were extracted and selected as model inputs. The results show that: (1) the relative error of prediction of regression modeling is 8.68 %, which is much lower than the 17.75 % relative error achieved by the SSD object detection model; (2) there is a strong correlation between the shape features of oilseed rape seedlings and the actual number of seedlings measured, and the DNN fully-connected neural network performs the best among the different modeling methods, and in the five-fold cross validation, its average R can reach 0.81, and the average MAPE of 7.22 %, RMSE of 11.61, and prediction error of 8.68 % determined that the DNN prediction model can better estimate the number of rape seedlings. In summary, this study establishes a practical and fast remote sensing counting method for the early growth of oilseed rape seedlings, reduces the error of object detection due to high overlap, and uses regression prediction method to invert the model in oilseed rape land to verify the validity of the model. This method provides a scalable solution for large-scale field monitoring, enabling precision agriculture practices such as optimized planting density and early yield prediction.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101717"},"PeriodicalIF":4.5,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145223399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}