Small-target detection in Unmanned Aerial Vehicle (UAV) aerial images remains a significant and unresolved challenge in aerial image analysis, hampered by low target resolution, dense object clustering, and complex, cluttered backgrounds. In order to cope with these problems, we present AeroLight, a novel and efficient detection architecture that achieves high-fidelity performance in resource-constrained environments. AeroLight is built upon three key innovations. First, we have optimized the feature pyramid at the architectural level by integrating a high-resolution head specifically designed for minute object detection. This design enhances sensitivity to fine-grained spatial details while streamlining redundant and computationally expensive network layers. Second, a Dynamic Feature Fusion (DFF) module is proposed to adaptively recalibrate and merge multi-scale feature maps, mitigating information loss during integration and strengthening object representation across diverse scales. Finally, we enhance the localization precision of irregular-shaped objects by refining bounding box regression using a Shape-IoU loss function. AeroLight is shown to improve mAP50 and mAP50-95 by 7.5% and 3.3%, respectively, on the VisDrone2019 dataset, while reducing the parameter count by 28.8% when compared with the baseline model. Further validation on the RSOD dataset and Huaxing Farm Drone dataset confirms its superior performance and generalization capabilities. AeroLight provides a powerful and efficient solution for real-world UAV applications, setting a new standard for lightweight, high-precision object recognition in aerial imaging scenarios.

The identification of cotton diseases and pests is crucial for maintaining cotton yield and quality. However, conventional manual methods are inefficient and prone to high error rates, limiting their practicality in real-world agricultural scenarios. Furthermore, Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) models are insufficient in generating fine-grained and semantically rich image captions, particularly for complex disease and pest features. To overcome these challenges, we introduce CottonCapT6, a novel multi-task image captioning framework based on the Cross Vision Transformer (CrossViT-18-Dagger-408) and Text-to-Text Transfer Transformer (T5). We also construct a new dataset containing annotated images of seven common cotton diseases and pests to support this work. Experimental results demonstrate that CottonCapT6 achieves a Consensus-based Image Captioning Evaluation (CIDEr) score of 197.2% on the captioning task, demonstrating outstanding performance. Notably, the framework excels in providing more descriptive, coherent, and contextually accurate captions. This approach has strong potential to be deployed in cotton farms in the future, helping pest control personnel and farmers make precise judgments on cotton diseases and pests. However, its generalizability to other crops and environmental conditions remains an area for future exploration.

Canopy 3D architecture is a critical determinant of light interception, photosynthetic efficiency, and final yield in cotton, yet its rapid and accurate characterisation remains challenging in field conditions. To achieve efficient, non-destructive, and quantitative monitoring of the canopy structure of field-grown cotton, this study proposes a 3D structure-based technology stack for high-efficiency, low-cost, and high-precision phenotyping extraction. This stack directly addresses the technical bottlenecks of traditional 3D data acquisition, namely high cost, long processing time, and low operational efficiency, which have hindered large-scale application. We developed a pipeline that integrates a fast reconstruction algorithm with a scale-recovery mechanism using ground control points (GCPs), enabling the generation of true-scale 3D point clouds from UAV aerial images in a cost- and time-effective manner. Using only 141 UAV images and with a reconstruction time of approximately 20 min, we efficiently reconstructed high-quality, scale-accurate point clouds of two 5.5 m × 5.5 m cotton plots, significantly outperforming SfM-MVS and Instant-NGP in terms of both reconstruction efficiency and point cloud completeness. This method, whose current validation is confined to a single season, one growth stage, and two experimental plots, not only achieves a breakthrough by using fewer input images with high efficiency, but also ensures point cloud accuracy and completeness, showing strong potential for rapid field monitoring and real-time management. Based on the high-quality reconstructed point clouds, we further quantitatively evaluated canopy characteristics at harvest, analyzing the coefficient of variation of canopy height, porosity distribution, and canopy volume fraction. The core shortcomings and optimization strategies for the existing canopy structure were identified, providing scientific data support and practical technical references for precision cultivation management and mechanization-compatible planting in cotton.

Detector-assisted promptable segmentation can reuse box annotations, but prompt design creates an accuracy–efficiency trade-off in dense scenes. We study a training-free interface in which a detector trained with bounding-box annotations localizes stacked cartons and a frozen segmenter generates masks. Four interfaces are evaluated: Point-Single, Box, ambiguity-aware Point-Max, and Grid-Iter with sparse positive points and one mask-logit feedback pass. On a leakage-audited split, COCO mask AP, fixed-threshold Hungarian-matched metrics, paired bootstrap uncertainty, and synchronized latency were measured. On 773 test images, Box and Grid-Iter achieved mask AP values of 0.8984 and 0.8988. Their paired AP difference was 0.0007 (95% CI, $-0.0035$ to 0.0047), while matched-mIoU and F1 intervals also included zero. Box required 604.67 ms/image versus 677.33 ms/image for Grid-Iter. A YOLO26s–SAM 3 Box reference achieved a mask AP value of 0.9115 at 320.74 ms/image, whereas mask-supervised YOLOv9c-seg achieved a value of 0.9235 AP on the same cleaned split. The results indicate a metric-dependent Pareto trade-off rather than overall Grid-Iter superiority and position detector-to-segmenter prompting as a reproducible annotation–accuracy option for densely stacked cartons.

Accurate point-cloud segmentation of cotton organs is essential for precise phenotypic characterization. However, reliable instance segmentation of cotton bolls in field-derived point clouds remains challenging because foliage occlusion, contact between adjacent bolls and incomplete reconstruction obscure instance boundaries. Here we present RQ-PointNeXt, an end-to-end framework that directly maps input point clouds to boll instance masks within a unified trainable network. Built on PointNeXt, it incorporates relative-elevation geometric channel attention in the shallow encoder to fuse global channel context with elevation and surface-normal cues. Its Query–mask branch integrates semantic guidance, center-seeded queries and a center-aware mask prior to suppress background responses, localize instances and constrain mask extent. Hungarian matching and multitask optimization establish one-to-one query–instance assignments, whereas query-based decoding produces instance masks without external geometric clustering. We evaluated the framework on 226 field-grown cotton plants containing 720 annotated boll instances reconstructed from UAV multi-view imagery using neural radiance fields. On the held-out test set, overall accuracy, mean class accuracy and mean intersection over union reached 0.8942, 0.8980 and 0.8076, respectively. AP25, AP50 and AP75 were 0.7359, 0.5585 and 0.2777, yielding an mAP25/50/75 of 0.5240. The framework provides instance-level outputs for boll counting and spatial analysis in high-throughput field phenotyping.