To address the challenges of large-scale variations in human targets,the loss of spatial details,and the inconsistency between prediction confidence and localization quality in complex scenarios,this study proposes a ...To address the challenges of large-scale variations in human targets,the loss of spatial details,and the inconsistency between prediction confidence and localization quality in complex scenarios,this study proposes a high-quality localization-aware action recognition method based on YOLOv11d.An SPDConv downsampling structure is introduced into the backbone network and the feature fusion stage to enhance the representation capability of small-scale target features.In addition,a localization quality estimation branch is incorporated into the detection head to explicitly model the Intersection over Union(IoU)of bounding boxes,and the confidence score is reweighted by combining the estimated localization quality with class probability.Experimental results demonstrate that the proposed method achieves an mAP@50 of 96.0%and an mAP@50-95 of 72.3%,representing improvements of 0.3%and 2.8%,respectively,compared with YOLOv11.展开更多
To address the dual challenges of excessive energy consumption and operational inefficiency inherent in the reliance of current agricultural machinery on direct supervision,this study developed an enhanced YOLOv8n-SS ...To address the dual challenges of excessive energy consumption and operational inefficiency inherent in the reliance of current agricultural machinery on direct supervision,this study developed an enhanced YOLOv8n-SS pedestrian detection algorithm through architectural modifications to the baseline YOLOv8n framework.The proposed method had superior performance in dense agricultural contexts while improving detection capabilities for pedestrian distribution patterns under complex farmland conditions,including variable lighting and mechanical occlusions.The main innovations were:(1)integration of spatial pyramid dilated(SPD)operations with conventional convolution layers to construct SPD-Conv modules,which effectively mitigated feature information loss while enhancing small-target detection accuracy;(2)incorporation of selective kernel attention mechanisms to enable context-aware feature selection and adaptive feature extraction.Experimental validation revealed significant performance improvements over the original YOLOv8n model.This enhanced architecture achieved 7.2% and 9.2% increases in m AP0.5 and m AP0.5:0.95 metrics respectively for dense pedestrian detection,with corresponding improvements of 7.6% and 8.7% observed in actual farmland working environments.The proposed method ultimately provides a computationally efficient and robust intelligent monitoring solution for agricultural mechanization,facilitating the transition from conventional agricultural practices toward sustainable,low-carbon production paradigms through algorithmic optimization.展开更多
文摘To address the challenges of large-scale variations in human targets,the loss of spatial details,and the inconsistency between prediction confidence and localization quality in complex scenarios,this study proposes a high-quality localization-aware action recognition method based on YOLOv11d.An SPDConv downsampling structure is introduced into the backbone network and the feature fusion stage to enhance the representation capability of small-scale target features.In addition,a localization quality estimation branch is incorporated into the detection head to explicitly model the Intersection over Union(IoU)of bounding boxes,and the confidence score is reweighted by combining the estimated localization quality with class probability.Experimental results demonstrate that the proposed method achieves an mAP@50 of 96.0%and an mAP@50-95 of 72.3%,representing improvements of 0.3%and 2.8%,respectively,compared with YOLOv11.
基金supported by the General Program of the Natural Science Foundation of Hunan Province of China(2021JJ30359)。
文摘To address the dual challenges of excessive energy consumption and operational inefficiency inherent in the reliance of current agricultural machinery on direct supervision,this study developed an enhanced YOLOv8n-SS pedestrian detection algorithm through architectural modifications to the baseline YOLOv8n framework.The proposed method had superior performance in dense agricultural contexts while improving detection capabilities for pedestrian distribution patterns under complex farmland conditions,including variable lighting and mechanical occlusions.The main innovations were:(1)integration of spatial pyramid dilated(SPD)operations with conventional convolution layers to construct SPD-Conv modules,which effectively mitigated feature information loss while enhancing small-target detection accuracy;(2)incorporation of selective kernel attention mechanisms to enable context-aware feature selection and adaptive feature extraction.Experimental validation revealed significant performance improvements over the original YOLOv8n model.This enhanced architecture achieved 7.2% and 9.2% increases in m AP0.5 and m AP0.5:0.95 metrics respectively for dense pedestrian detection,with corresponding improvements of 7.6% and 8.7% observed in actual farmland working environments.The proposed method ultimately provides a computationally efficient and robust intelligent monitoring solution for agricultural mechanization,facilitating the transition from conventional agricultural practices toward sustainable,low-carbon production paradigms through algorithmic optimization.
