DC-YOLO：应对遮挡和光照变化的湖羊行为检测

doi:10.12133/j.smartag.SA202601021

摘要/Abstract

摘要：

【目的/意义】 畜牧业中，动物行为是评估动物生理健康以及提供环境预警的重要指标。针对集约化养殖中行为检测面临的多目标严重遮挡、光照条件多变、类别分布不均等挑战，本研究旨在开发一种能够实现舍饲湖羊24小时日常行为自动监测的高效算法DCAttention-YOLO（DC-YOLO），以支撑羊群的精准化管理。 【方法】 针对多目标严重遮挡，引入自适应细节-上下文注意力（DCAttention，Adaptive Detail-Contextual Attention）机制构建DCAC3K2模块，并采用边界框相似度Soft-NMS（Bounding Box Similarity Soft-NMS，BS‑NMS）防止错误去除遮挡目标的边界框。针对光照条件多变，预训练光照编码器并将其参数迁移至DC-YOLO。针对类别分布不均，设计综合分类质量焦点损失（CQFL），对遮挡及少数类样本自适应增大损失权重。本研究数据集通过湖羊养殖场全天候监控采集，包含505张标注图像，涵盖饮水、采食、躺卧、舔舐和站立5种行为。 【结果和讨论】 DC-YOLO的重叠度阈值为 0.5 时的平均精度均值mAP@50达到91.4%，相比YOLOv12提高了7.8个百分点。此外，DC-YOLO的参数量为2.29 M，相较YOLOv12减少了8.74%。在CPU环境下，DC-YOLO的推理时间缩短至115.5 ms，帧率提升至8.50帧/s，分别提升了33.2%和48.9%。 【结论】 DC-YOLO在保持低参数量与高推理速度的同时，有效缓解了舍饲环境下多目标遮挡、光照条件多变及行为类别分布不均带来的检测难题，能够为集约化湖羊养殖的行为跟踪与分析提供有效的技术支撑。

关键词: 集约化养殖, 行为检测, 湖羊行为, 注意力机制, YOLO, BS-NMS

Abstract:

[Objective] In the livestock industry, animal behavior serves as a critical indicator for evaluating physiological health and providing environmental early warning. To address the challenges of severe multi-object occlusion, complex illumination, and class imbalance in intensive farming, a detail-contextual attention-YOLO (DC-YOLO) model was developed for the 24/7 automatic detection of daily behaviors in housed Hu sheep, thereby facilitating precision management. [Methods] For severe multi-object occlusion, an adaptive detail-contextual attention (DCAttention) mechanism was introduced to construct the DCAC3K2 module, and bounding box similarity soft-NMS (BS-NMS) was adopted to prevent incorrect removal of occluded bounding boxes. For complex illumination, a Light-Encoder was pre-trained on illumination conditions and its parameters were transferred to DC-YOLO. For class imbalance, a comprehensive classification-quality focal loss was designed to adaptively increase loss weights for occluded and minority-class samples. The dataset, collected through 24/7 surveillance at Hu sheep farms, comprised 505 annotated images depicting 5 daily behaviors: drinking, eating, lying, licking, and standing. [Results and Discussions] DC-YOLO achieved mAP@50 of 91.4%, improving by 7.8 percentage points over YOLOv12. Moreover, DC-YOLO had 2.29 M parameters, an 8.74% reduction compared to YOLOv12. On CPU, the inference time of DC-YOLO was reduced to 115.5 ms and the frame rate increased to 8.50 f/s, corresponding to improvements of 33.2% and 48.9%, respectively. [Conclusions] Experimental results demonstrate that DC-YOLO effectively mitigates detection challenges caused by severe occlusion, complex illumination, and class imbalance while maintaining high inference efficiency. Consequently, it provides effective technical support for tracking and analyzing behaviors in intensive Hu sheep farming.

Key words: Intensive livestock farming, Behavior detection, Hu sheep behavior, Attention Mechanism, YOLO, BS-NMS

中图分类号:

TP391.4

李夏溪, 冀荣华, 常宏瑞, 张所向. DC-YOLO：应对遮挡和光照变化的湖羊行为检测[J]. 智慧农业(中英文), 2026, 8(4): 127-147.

LI Xiaxi, JI Ronghua, CHANG Hongrui, ZHANG Suoxiang. DC-YOLO: A Behavior Detection Model for Hu Sheep Addressing Occlusion and Illumination Variations[J]. Smart Agriculture, 2026, 8(4): 127-147.

图/表 22

参考文献 40

[1]	OECD, FAO. OECD-FAO Agricultural Outlook 2025-2034[R/OL]. Paris: OECD Publishing; Rome: FAO, 2025[2025-11-05].
[2]	THARWAT M, ALKHERAIF A A, OIKAWA S. Production diseases in farm animals: A comprehensive and illustrated clinical, laboratory, and pathological overview[J]. Open Veterinary Journal, 2025, 15(1): 18-34.
[3]	HAYEK M N. The infectious disease trap of animal agriculture[J]. Science Advances, 2022, 8(44): eadd6681.
[4]	RICHMOND S E, WEMELSFELDER F, DE HEREDIA I B, et al. Evaluation of animal-based indicators to be used in a welfare assessment protocol for sheep[J]. Frontiers in Veterinary Science, 2017, 4: 210.
[5]	NECULAI-VALEANU A S, SANDULEANU C, POROSNICU I. From tradition to precision: Leveraging digital tools to improve cattle health and welfare[J]. Frontiers in Veterinary Science, 2025, 12: 1549512.
[6]	RIABOFF L, SHALLOO L, SMEATON A F, et al. Predicting livestock behaviour using accelerometers: A systematic review of processing techniques for ruminant behaviour prediction from raw accelerometer data[J]. Computers and Electronics in Agriculture, 2022, 192: 106610.
[7]	LAMANNA M, BOVO M, CAVALLINI D. Wearable collar technologies for dairy cows: A systematized review of the current applications and future innovations in precision livestock farming[J]. Animals, 2025, 15(3): 458.
[8]	HERLIN A, BRUNBERG E, HULTGREN J, et al. Animal welfare implications of digital tools for monitoring and management of cattle and sheep on pasture[J]. Animals, 2021, 11(3): 829.
[9]	BEZEN R, EDAN Y, HALACHMI I. Computer vision system for measuring individual cow feed intake using RGB-D camera and deep learning algorithms[J]. Computers and Electronics in Agriculture, 2020, 172: 105345.
[10]	WANG A, SUN Y, KORTYLEWSKI A, et al. Robust object detection under occlusion with context-aware CompositionalNets[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2020: 12642-12651.
[11]	KAUR R, SINGH S. A comprehensive review of object detection with deep learning[J]. Digital Signal Processing, 2023, 132: 103812.
[12]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2016: 779-788.
[13]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: Single shot MultiBox detector[C]// Computer Vision – ECCV 2016. Cham, Switzerland: Springer, 2016: 21-37.
[14]	BORGES OLIVEIRA D A, RIBEIRO PEREIRA L G, BRESOLIN T, et al. A review of deep learning algorithms for computer vision systems in livestock[J]. Livestock Science, 2021, 253: 104700.
[15]	REN S, HE K, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
[16]	CAI Z, VASCONCELOS N. Cascade R-CNN: Delving into high quality object detection[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2018: 6154-6162.
[17]	HE K, GKIOXARI G, DOLLÁR P, et al. Mask R-CNN[C]// 2017 IEEE International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2017: 2980-2988.
[18]	BOLYA D, ZHOU C, XIAO F, et al. YOLACT: Real-time instance segmentation[C]// 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2019: 9156-9165.
[19]	CAO Z, HIDALGO G, SIMON T, et al. OpenPose: Realtime multi-person 2D pose estimation using part affinity fields[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(1): 172-186.
[20]	GONG C, ZHANG Y, WEI Y, et al. Multicow pose estimation based on keypoint extraction[J]. PLoS One, 2022, 17(6): e0269259.
[21]	GUO X, MA C, WANG C, et al. A sheep behavior recognition approach based on improved FESS-YOLOv8n neural network[J]. Animals, 2025, 15(6): 893.
[22]	WANG J, ZHAI Y, ZHU L, et al. PD-YOLO: A study of daily behavioural detection in housed sheep[J]. PLoS One, 2024, 19(11): e0313412.
[23]	HU T, YAN R, JIANG C, et al. Grazing sheep behaviour recognition based on improved YOLOV5[J]. Sensors, 2023, 23(10): 4752.
[24]	GU Z, ZHANG H, HE Z, et al. A two-stage recognition method based on deep learning for sheep behavior[J]. Computers and Electronics in Agriculture, 2023, 212: 108143.
[25]	YU Y, ZHU W, MA X, et al. Recognition of sheep feeding behavior in sheepfolds using fusion spectrogram depth features and acoustic features[J]. Animals, 2024, 14(22): 3267.
[26]	ZHANG J, QIAN S, CHEN J, et al. Calm Hu ram lambs assigned by temperament classification are healthier and have better meat quality than nervous Hu ram lambs[J]. Meat Science, 2021, 175: 108436.
[27]	JOCHER G, CHAURASIA A, QIU J. Ultralytics YOLOv8[CP/OL]. [2025-11-5].
[28]	LI X, WANG W, WU L, et al. Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection[C]// Advances in Neural Information Processing Systems 33 (NeurIPS 2020. Vancouver, BC, Canada: Neural Information Processing Systems Foundation, Inc. (NeurIPS), 2020.
[29]	JOCHER G, QIU J. Ultralytics YOLO11[CP/OL]. [2025-11-5].
[30]	TIAN Y, YE Q, DOERMANN D. YOLOv12: Attention-Centric Real-Time Object Detectors[PP/OL]. arXiv: 2502.12524, 2025.
[31]	SHI D. TransNeXt: Robust foveal visual perception for vision transformers[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2024: 17773-17783.
[32]	BODLA N, SINGH B, CHELLAPPA R, et al. Soft-NMS—improving object detection with one line of code[C]// 2017 IEEE International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2017: 5562-5570.
[33]	WANG C Y, YEH I H, MARK LIAO H Y. YOLOv9: Learning what you want to learn using programmable gradient information[C]// Computer Vision – ECCV 2024. Cham, Germany: Springer, 2025: 1-21.
[34]	WANG A, CHEN H, LIU L, et al. YOLOv10: Real-Time End-to-End Object Detection[C]// Advances in Neural Information Processing Systems 37 (NeurIPS 2024). Red Hook, NY, USA: Curran Associates, Inc., 2024: 107984-108011.
[35]	LEI M Q, LI S Q, WU Y H, et al. YOLOv13: Real-time object detection with hypergraph-enhanced adaptive visual perception[PP/OL]. arXiv: 2506.17733, 2025.
[36]	ZHAO Y A, LV W Y, XU S L, et al. DETRs beat YOLOs on real-time object detection[C]// 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2024: 16965-16974.
[37]	LIU S L, LI F, ZHANG H, et al. DAB-DETR: Dynamic anchor boxes are better queries for DETR[PP/OL]. arXiv: 2201.12329, 2022.
[38]	ZHANG H, LI F, LIU S L, et al. DINO: DETR with improved DeNoising anchor boxes for end-to-end object detection[PP/OL]. arXiv: 2203.03605, 2022.
[39]	ZHANG Y F, SUN P Z, JIANG Y, et al. ByteTrack: Multi-object tracking by Associating every detection box[C]// Computer Vision-ECCV 2022: 17th European Conference. Cham, Switzerland: Springer, 2022: 1-21.
[40]	AHARON N, ORFAIG R, BOBROVSKY B Z. BoT-SORT: Robust associations multi-pedestrian tracking[PP/OL]. arXiv: 2206.14651, 2022.

Behavior ID	Typical behavior	Description
0	Drinking	Standing with limbs outstretched and drinking water while looking down at the pipe
1	Eating	Standing with limbs extended and head lowered to eat in front of the feeding trough
2	Lying	Body resting on the ground， with limbs bent or tucked
3	Licking	Standing with limbs outstretched and licking the salt brick
4	Standing	Standing on all limbs with the body supported， and not actively engaged in eating， drinking， or licking

Configuration Item	Value
CPU	Intel（R） Xeon（R） Platinum 8352 V CPU @ 2.10 GHz
GPU	NVIDIA GeForce RTX 4090
OS	Ubuntu 20.04
Programming language	Python 3.8
Deep learning framework	PyTorch 2.0.0
Version of CUDA	CUDA 11.8
System RAM	120 GB

Model	DCAC3K2	CQFL	BS-NMS	A2C2f	P/%	R/%	mAP@50/%	mAP@50：95/%	Parameters/M	t _infer/ms	t _post/ms
YOLOv12	×	×	×	√	81.6	78.5	83.6	64.9	2.51	172.8	0.37
1	√	×	×	√	79.0	87.7	86.7	66.0	2.17	130.3	0.31
2	×	√	×	√	85.9	74.5	85.5	66.9	2.51	165.3	0.39
3	×	×	√	√	83.7	82.8	87.6	68.6	2.51	168.2	1.60
4	√	√	×	√	86.5	86.2	88.0	71.5	2.17	143.2	0.37
5	√	√	√	√	91.0	86.8	90.1	73.8	2.17	133.5	1.74
6	√	√	×	×	85.9	86.1	88.9	73.5	2.29	163.7	1.71
DC-YOLO	√	√	√	×	91.0	86.6	91.4	75.9	2.29	115.5	1.69

Model	Pre-training method	mAP/%				P/%	R/%
Model	Pre-training method	daytime natural	nighttime	abnormal lighting	All	P/%	R/%
YOLOv11	YOLOv11n	93.0	82.9	84.5	87.3	88.1	83.9
	Random initial weights	91.9	82.1	83.9	86.8	83.4	81.8
	Light-Encoder	92.8	83.4	88.4	88.9	84.9	77.6
YOLOv12	YOLOv12n	92.5	80.8	78.7	84.6	79.5	82.7
	Random initial weights	91.8	80.1	77.3	83.6	81.6	78.5
	Light-Encoder	92.4	82.9	80.1	85.7	77.4	85.9
DC-YOLO （ours）	YOLOv11n	94.1	84.3	87.4	90.6	90.0	86.1
	YOLOv12n	94.0	84.4	86.9	90.5	83.7	85.1
	Random initial weights	93.1	83.7	85.9	89.9	83.4	87.6
	Light-Encoder	93.9	85.1	89.3	91.4	91.0	86.6

Model	P /%	R /%	mAP /%	Parameters/M	Computational/GFLOPs	FPS/（f/s）
YOLOv8	81.4	81.6	85.8	2.69	6.8	8.56
YOLOv9	83.5	78.0	84.3	2.43	6.4	7.03
YOLOv10	79.4	76.4	82.2	2.27	6.5	7.51
YOLOv11	83.4	81.8	86.8	2.59	6.4	7.58
YOLOv12	81.6	78.5	83.6	2.51	6.9	5.71
YOLOv13	77.9	83.8	83.9	2.45	6.1	4.29
DINO	62.7	67.6	68.7	47.00	279.0	2.58
DAB-DETR	43.2	51.7	50.4	43.00	195.0	2.52
RT-DETR	63.1	73.6	71.4	42.77	130.5	1.21
DC-YOLO	91.0	86.6	91.4	2.29	6.2	8.50