Unmanned Aerial Vehicle Remote Sensing (UAV-RS) has emerged as a transformative technology in high-resolution Earth observation, with widespread applications in precision agriculture, ecological monitoring, and disaster response. However, a systematic understanding of its scientific evolution and structural bottlenecks remains lacking. This study collected 4985 peer-reviewed articles from the Web of Science Core Collection and conducted a comprehensive scientometric analysis using CiteSpace v.6.2.R4, Origin 2022, and Excel. We examined publication trends, country/institutional collaboration networks, keyword co-occurrence clusters, and emerging research fronts. Results reveal an exponential growth in UAV-RS research since 2015, dominated by application-driven studies. Hotspots include vegetation indices, structure from motion modeling, and deep learning integration. However, foundational challenges—such as platform endurance, sensor coordination, and data standardization—remain underexplored. The global collaboration network exhibits a “strong hubs, weak bridges” pattern, limiting transnational knowledge integration. This review highlights the imbalance between surface-level innovation and deep technological maturity and calls for a paradigm shift from fragmented application responses to integrated systems development. Our findings provide strategic insights for researchers, policymakers, and funding agencies to guide the next stage of UAV-RS evolution.

Rapidly and accurately extracting tobacco plant information can facilitate tobacco planting man-agement, precise fertilization and yield prediction. In karst plateau of southern China, tobacco plant identification is affected by large ground undulations, fragmented planting areas, complex and diverse habitats and uneven plant growth. This study took a tobacco planting area in Gui-zhou Province as the research object and used DJI UAVs to collect UAV visible light images. Considering plot fragmentation, plant size, presence of weeds and shadow masking, this area was classified into eight habitats. The datasets of different habitats were constructed to train the U-Net model. The results show that (1) the overall precision, recall, F1-score and IOU of tobacco plant information extraction were 0.68, 0.85, 0.75 and 0.60, respectively. (2) The precision was the highest for the subsurface-fragmented and weed-free habitat and the lowest for the smooth-tectonics and weed-infested habitat. (3) The weed-infested habitat with smaller tobacco plants can cause blurred images, reducing the plant identification accuracy. This study verified the feasibility of the U-Net model for tobacco single-plant identification in complex habitats. Decomposing complex habitats to establish the sample set method is a new attempt to improve crop identification in complex habitats in karst mountainous areas.

Karst hillslope land systems are characterized by strong microtopographic heterogeneity, exposed rock–soil patches, and management-related disturbance, resulting in pronounced fine-scale variation in ecological responses. This study investigated three contrasting karst hillslopes within the Guanling–Zhenfeng Huajiang Rocky Desertification Comprehensive Control Demonstration Area in Guizhou Province, China, designated as High, Medium, and Natural sites according to their observed land cover and management characteristics, using UAV multispectral imagery and LiDAR-derived DEM data. A terrain–ecology coupling strength indicator (TECSI) framework was developed to assess the spatially transferable predictability of ecological response patterns by terrain variables. At the 5 m block scale, FVC, OSAVI, NDRE750, rock–soil exposure index (REI), and GLCM contrast were linked with LiDAR-derived microtopographic predictors using generalized additive models, residual Moran’s I, random cross-validation, and spatial block cross-validation. FVC, OSAVI, NDRE750, and GLCM contrast differed significantly among the three sites (p < 0.001), whereas REI mainly reflected patchy non-vegetated rock–soil substrate exposure within hillslopes. Across the 15 site–indicator models, deviance explained ranged from 5.5% to 35.8%. Under 25 m spatial block cross-validation, integrated TECSI ranked Medium (0.143) > High (0.086) > Natural (0.036), and this ranking remained stable across sensitivity analyses. Steep-slope units at or above 25° in the High and Medium sites were associated with poorer vegetation condition and increased substrate exposure. TECSI provides a reproducible spatial validation framework for identifying terrain-sensitive ecological response units and supporting fine-scale monitoring, soil–water conservation screening, and vulnerable land unit management in karst hillslopes.