【摘要】 针对目前复杂环境下奶牛在采食过程中挑食行为与采食行为差异不大、识别精度较低、人工识别劳动强度大等问题，本文提出了一种基于巡检机器人和改进RT-DETR模型的奶牛挑食行为识别方法。根据奶牛采食特性设计巡检机器人采集奶牛采食过程数据，分中午、下午和晚上3个时间段分别在3个牛棚进行采集，最终构建包含3个时间段共计10 280幅奶牛采食数据集。对RT-DETR模型进行改进，在RT-DETR模型浅层中引入DAttention（DAT）模块和Bi-Level Routing Attention（BRA）模块融合的DBRA结构，建立了新的图像特征提取结构，提升输入图像局部和全局特征深度融合能力；在RT-DETR模型编码器中融合Efficient Multi-Scale Attention（EMA）模块，增强了提取高层次语义信息能力，更好地联系上下文信息。试验结果表明，改进后模型在奶牛采食视频数据集平均精度均值（mAP@0.5）为99.1%,模型内存占用量为39.6 MB,浮点计算量为4.67×10¹⁰,相较于原模型平均精度均值提高7.4个百分点，模型内存占用量降低0.9 MB,浮点计算量减少2%。与Sparse R-CNN、YOLO v7-L、YOLO v8n、DINO、Swin Transformer和DETR模型相比，平均精度均值（mAP@50）分别提高8.5、9.8、7.8、6.6、11.4、9.5个百分点。研究结果可以为实现畜牧养殖智能化提供技术支持。更多还原

【Abstract】 Aiming to address the challenges of low recognition accuracy, high labor intensity in manual identification, and minimal behavioral differences between selective feeding and normal feeding in dairy cows under complex environmental conditions, a method for identifying selective feeding behavior was proposed based on inspection robots and an improved RT-DETR model. An inspection robot was designed according to dairy cows’ feeding characteristics to collect feeding process data. Data collection was conducted in three barns during three time periods（noon, afternoon, and night）, ultimately establishing a dataset containing 10 280 feeding behavior images across these periods. The RT-DETR model was enhanced by integrating a DBRA structure, which combined the DAttention（DAT） module and Bi-Level Routing Attention（BRA） module into the shallow layers, creating a novel image feature extraction architecture to improve the deep fusion capability of local and global features. Additionally, the Efficient Multi-Scale Attention（EMA） module was incorporated into the model encoder to strengthen high-level semantic information extraction and contextual correlation. Experimental results demonstrated that the improved model achieved a mean average precision（mAP@0.5） of 99.1% on the dairy cow feeding video dataset, with a model memory occupancy of 39.6 MB and floating-point operations（FLOPs） of 4.67×10¹⁰. Compared with the original model, the mAP@0.5 was increased by 7.4 percentage points, memory occupancy was reduced by 0.9 MB, and FLOPs was decreased by 2%. When compared with Sparse R-CNN, YOLO v7-L, YOLO v8n, DINO, Swin Transformer, and DETR models, the proposed model exhibited mAP@50 improvements of 8.5, 9.8, 7.8, 6.6, 11.4 and 9.5 percentage points, respectively. The findings enabled accurate differentiation between normal feeding and selective feeding behaviors, providing technical support for intelligent livestock farming.更多还原