A Regional Farming Pig Counting System Based on Improved Instance Segmentation Algorithm

doi:10.12133/j.smartag.SA202310001

Abstract

Abstract:

[Objective] Currently, pig farming facilities mainly rely on manual counting for tracking slaughtered and stored pigs. This is not only time-consuming and labor-intensive, but also prone to counting errors due to pig movement and potential cheating. As breeding operations expand, the periodic live asset inventories put significant strain on human, material and financial resources. Although methods based on electronic ear tags can assist in pig counting, these ear tags are easy to break and fall off in group housing environments. Most of the existing methods for counting pigs based on computer vision require capturing images from a top-down perspective, necessitating the installation of cameras above each hogpen or even the use of drones, resulting in high installation and maintenance costs. To address the above challenges faced in the group pig counting task, a high-efficiency and low-cost pig counting method was proposed based on improved instance segmentation algorithm and WeChat public platform. [Methods] Firstly, a smartphone was used to collect pig image data in the area from a human view perspective, and each pig's outline in the image was annotated to establish a pig count dataset. The training set contains 606 images and the test set contains 65 images. Secondly, an efficient global attention module was proposed by improving convolutional block attention module (CBAM). The efficient global attention module first performed a dimension permutation operation on the input feature map to obtain the interaction between its channels and spatial dimensions. The permuted features were aggregated using global average pooling (GAP). One-dimensional convolution replaced the fully connected operation in CBAM, eliminating dimensionality reduction and significantly reducing the model's parameter number. This module was integrated into the YOLOv8 single-stage instance segmentation network to build the pig counting model YOLOv8x-Ours. By adding an efficient global attention module into each C2f layer of the YOLOv8 backbone network, the dimensional dependencies and feature information in the image could be extracted more effectively, thereby achieving high-accuracy pig counting. Lastly, with a focus on user experience and outreach, a pig counting WeChat mini program was developed based on the WeChat public platform and Django Web framework. The counting model was deployed to count pigs using images captured by smartphones. [Results and Discussions] Compared with existing methods of Mask R-CNN, YOLACT(Real-time Instance Segmentation), PolarMask, SOLO and YOLOv5x, the proposed pig counting model YOLOv8x-Ours exhibited superior performance in terms of accuracy and stability. Notably, YOLOv8x-Ours achieved the highest accuracy in counting, with errors of less than 2 and 3 pigs on the test set. Specifically, 93.8% of the total test images had counting errors of less than 3 pigs. Compared with the two-stage instance segmentation algorithm Mask R-CNN and the YOLOv8x model that applies the CBAM attention mechanism, YOLOv8x-Ours showed performance improvements of 7.6% and 3%, respectively. And due to the single-stage design and anchor-free architecture of the YOLOv8 model, the processing speed of a single image was only 64 ms, 1/8 of Mask R-CNN. By embedding the model into the WeChat mini program platform, pig counting was conducted using smartphone images. In cases where the model incorrectly detected pigs, users were given the option to click on the erroneous location in the result image to adjust the statistical outcomes, thereby enhancing the accuracy of pig counting. [Conclusions] The feasibility of deep learning technology in the task of pig counting was demonstrated. The proposed method eliminates the need for installing hardware equipment in the breeding area of the pig farm, enabling pig counting to be carried out effortlessly using just a smartphone. Users can promptly spot any errors in the counting results through image segmentation visualization and easily rectify any inaccuracies. This collaborative human-machine model not only reduces the need for extensive manpower but also guarantees the precision and user-friendliness of the counting outcomes.

Key words: pig counting, deep learning, WeChat mini program, YOLOv8, instance segmentation

ZHANG Yanqi, ZHOU Shuo, ZHANG Ning, CHAI Xiujuan, SUN Tan. A Regional Farming Pig Counting System Based on Improved Instance Segmentation Algorithm[J]. Smart Agriculture, 2024, 6(4): 53-63.

Figures/Tables 12

Fig. 1

Table 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Table 2

Table 3

Table 4

Fig. 6

Table 5

Fig. 7

References 21

1	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]// Proceedings of the 25th International Conference on Neural Information Processing Systems - Volume 1. New York, USA: ACM, 2012: 1097-1105.
2	DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An image is worth 16x16 words: Transformers for image recognition at scale[EB/OL]. arXiv: 2010.11929, 2020.
3	CHEN G, SHEN S W, WEN L Y, et al. Efficient Pig Counting in Crowds with Keypoints Tracking and Spatial-aware Temporal Response Filtering[C]// 2020 IEEE International Conference on Robotics and Automation (ICRA). Piscataway, New Jersey, USA: IEEE, 2020: 10052-10058.
4	TIAN M X, GUO H, CHEN H, et al. Automated pig counting using deep learning[J]. Computers and electronics in agriculture, 2019, 163: ID 104840.
5	XIE S N, GIRSHICK R, DOLLAR P, et al. Aggregated residual transformations for deep neural networks[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2017: 5987-5995.
6	高云, 李静, 余梅, 等. 基于多尺度感知的高密度猪只计数网络研究[J]. 农业机械学报, 2021, 52(9): 172-178.
	GAO Y, LI J, YU M, et al. High-density pig counting net based on multi-scale aware[J]. Transactions of the Chinese society for agricultural machinery, 2021, 52(9): 172-178.
7	LI Y H, ZHANG X F, CHEN D M. CSRNet: dilated convolutional neural networks for understanding the highly congested scenes[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2018: 1091-1100.
8	KIM J, SUH Y, LEE J, et al. EmbeddedPigCount: Pig counting with video object detection and tracking on an embedded board[J]. Sensors, 2022, 22(7): ID 2689.
9	JU M, CHOI Y, SEO J, et al. A kinect-based segmentation of touching-pigs for real-time monitoring[J]. Sensors, 2018, 18(6): ID 1746.
10	王荣, 高荣华, 李奇峰, 等. 融合特征金字塔与可变形卷积的高密度群养猪计数方法[J]. 农业机械学报, 2022, 53(10): 252-260.
	WANG R, GAO R H, LI Q F, et al. High-density pig herd counting method combined with feature pyramid and deformable convolution[J]. Transactions of the Chinese society for agricultural machinery, 2022, 53(10): 252-260.
11	ZHU X Z, HU H, LIN S, et al. Deformable ConvNets V2: more deformable, better results[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2019: 9300-9308.
12	XU B B, WANG W S, FALZON G, et al. Livestock classification and counting in quadcopter aerial images using Mask R-CNN[J]. International journal of remote sensing, 2020, 41(21): 8121-8142.
13	HE K M, GKIOXARI G, DOLLAR P, et al. Mask R-CNN[J]. IEEE transactions on pattern analysis and machine intelligence, 2020, 42(2): 386-397.
14	SARWAR F, GRIFFIN A, REHMAN S U, et al. Detecting sheep in UAV images[J]. Computers and electronics in agriculture, 2021, 187: ID 106219.
15	胡云鸽, 苍岩, 乔玉龙. 基于改进实例分割算法的智能猪只盘点系统设计[J]. 农业工程学报, 2020, 36(19): 177-183.
	HU Y G, CANG Y, QIAO Y L. Design of intelligent pig counting system based on improved instance segmentation algorithm[J]. Transactions of the Chinese society of agricultural engineering, 2020, 36(19): 177-183.
16	ULTRALYTICS. YOLOv8[EB/OL]. [2023-12-02].
17	HE K M, ZHANG X Y, REN S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE transactions on pattern analysis and machine intelligence, 2015, 37(9): 1904-1916.
18	ZHENG Z H, WANG P, LIU W, et al. Distance-IoU loss: Faster and better learning for bounding box regression[J]. Proceedings of the AAAI conference on artificial intelligence, 2020, 34(7): 12993-13000.
19	BOLYA D, ZHOU C, XIAO F Y, et al. YOLACT: real-time instance segmentation[C]// 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2019: 9157-9166.
20	XIE E Z, SUN P Z, SONG X G, et al. PolarMask: single shot instance segmentation with polar representation[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2020: 12193-12202.
21	WANG X L, KONG T, SHEN C H, et al. SOLO: segmenting objects by locations[M]// VEDALDI A, BISCHOF H, BROX T, et al., Eds. Computer vision-ECCV 2020. Cham: Springer International Publishing, 2020: 649-665.

数据集	图像数量	猪只总数/头	单幅图像最小猪只数/头	单幅图像最大猪只数/头
训练集	606	11 609	4	63
测试集	65	2 046	10	46

网络模型	AP50	AP_{（50：95）}	参数量/MB	单幅图像平均检测时间/ms
YOLACT	0.689	0.399	29.2	63
PolarMask	0.742	0.422	36.3	151
SOLO	0.811	0.537	46.2	113
Mask R-CNN	0.900	0.644	43.7	500
YOLOv5x	0.865	0.502	73.9	26
YOLOv8x	0.897	0.636	68.2	44
YOLOv8x-CBAM	0.901	0.646	125.3	109
YOLOv8x-Ours	0.901	0.660	71.7	64

模型	评价指标	得分阈值
模型	评价指标	0.05	0.1	0.15	0.2	0.25	0.3	0.35	0.4
YOLACT	MAE					7.125	7.125	7.125	7.125
	RMSE		—			8.369	8.369	8.369	8.369
	R ²					0.327	0.327	0.327	0.327
PolarMask	MAE					5.512	5.578	2.853	4.157
	RMSE		—			7.132	6.596	3.775	5.018
	R ²					0.519	0.585	0.868	0.763
SOLO	MAE					3.637	2.913	3.141	4.085
	RMSE		—			4.792	3.634	4.066	4.978
	R ²					0.785	0.879	0.843	0.767
Mask R-CNN	MAE					1.969	1.769	1.754	1.769
	RMSE		—			2.649	2.421	2.366	2.376
	R ²					0.932	0.943	0.946	0.945
YOLOv8x	MAE	2.169	1.985	2.462	2.892	3.077
	RMSE	2.826	2.715	3.153	3.526	3.713		—
	R ²	0.922	0.928	0.903	0.879	0.866
YOLOv8x -CBAM	MAE	1.985	1.831	1.923	1.835	2.431
	RMSE	2.720	2.434	2.508	2.440	3.103		—
	R ²	0.928	0.942	0.939	0.939	0.906
YOLOv8x -Ours	MAE	1.969	1.798	1.727	1.799	2.316
	RMSE	2.651	2.375	2.168	2.380	3.005		—
	R ²	0.931	0.945	0.949	0.944	0.938

模型	准确计数的图像		误差小于2头猪的图像		误差小于3头猪的图像		误差大于2头猪的图像
模型	数量	占比/%	数量	占比/%	数量	占比/%	数量	占比/%
YOLCAT	22	33.8	40	61.5	52	80.0	13	20.0
YOLOv5x	29	44.6	48	73.8	56	86.2	9	13.8
PolarMask	33	50.8	47	72.3	56	86.2	9	13.8
SOLO	35	53.8	50	76.9	58	89.2	7	10.8
YOLOv8x	35	53.8	49	75.4	57	87.7	8	12.3
Mask R-CNN	37	56.9	49	75.4	56	86.2	9	13.8
YOLOv8x-CBAM	41	63.1	51	78.5	59	90.8	6	9.2
YOLOv8x-Ours	43	66.2	57	87.7	61	93.8	4	6.2

[1]	MA Liu, MAO Kebiao, GUO Zhonghua. Defogging Remote Sensing Images Method Based on a Hybrid Attention-Based Generative Adversarial Network [J]. Smart Agriculture, 2025, 7(2): 172-182.
[2]	XU Shiwei, LI Qianchuan, LUAN Rupeng, ZHUANG Jiayu, LIU Jiajia, XIONG Lu. Agricultural Market Monitoring and Early Warning: An Integrated Forecasting Approach Based on Deep Learning [J]. Smart Agriculture, 2025, 7(1): 57-69.
[3]	GONG Yu, WANG Ling, ZHAO Rongqiang, YOU Haibo, ZHOU Mo, LIU Jie. Tomato Growth Height Prediction Method by Phenotypic Feature Extraction Using Multi-modal Data [J]. Smart Agriculture, 2025, 7(1): 97-110.
[4]	QI Zijun, NIU Dangdang, WU Huarui, ZHANG Lilin, WANG Lunfeng, ZHANG Hongming. Chinese Kiwifruit Text Named Entity Recognition Method Based on Dual-Dimensional Information and Pruning [J]. Smart Agriculture, 2025, 7(1): 44-56.
[5]	ZHANG Hui, HU Jun, SHI Hang, LIU Changxi, WU Miao. Precision Target Spraying System Integrated with Remote Deep Learning Recognition Model for Cabbage Plant Centers [J]. Smart Agriculture, 2024, 6(6): 85-95.
[6]	LU Bibo, LIANG Di, YANG Jie, SONG Aiqing, HUANGFU Shangwei. Image Segmentation Method of Chinese Yam Leaves in Complex Background Based on Improved ENet [J]. Smart Agriculture, 2024, 6(6): 109-120.
[7]	CHEN Junlin, ZHAO Peng, CAO Xianlin, NING Jifeng, YANG Shuqin. Lightweight YOLOv8s-Based Strawberry Plug Seedling Grading Detection and Localization via Channel Pruning [J]. Smart Agriculture, 2024, 6(6): 132-143.
[8]	LI Hongbo, TIAN Xin, RUAN Zhiwen, LIU Shaowen, REN Weiqi, SU Zhongbin, GAO Rui, KONG Qingming. Seedling Stage Corn Line Detection Method Based on Improved YOLOv8 [J]. Smart Agriculture, 2024, 6(6): 72-84.
[9]	LIU Chang, SUN Yu, YANG Jing, WANG Fengchao, CHEN Jin. Grape Recognition and Localization Method Based on 3C-YOLOv8n and Depth Camera [J]. Smart Agriculture, 2024, 6(6): 121-131.
[10]	LUO Youlu, PAN Yonghao, XIA Shunxing, TAO Youzhi. Lightweight Apple Leaf Disease Detection Algorithm Based on Improved YOLOv8 [J]. Smart Agriculture, 2024, 6(5): 128-138.
[11]	CUI Jiale, ZENG Xiangfeng, REN Zhengwei, SUN Jian, TANG Chen, YANG Wanneng, SONG Peng. Detection Method of Effective Tillering of Rice in Field Based on Lightweight Ghost-YOLOv8 and Smart Phone [J]. Smart Agriculture, 2024, 6(5): 98-107.
[12]	LIU Yi, ZHANG Yanjun. ReluformerN: Lightweight High-Low Frequency Enhanced for Hyperspectral Agricultural Lancover Classification [J]. Smart Agriculture, 2024, 6(5): 74-87.
[13]	NIAN Yue, ZHAO Kaixuan, JI Jiangtao. Cow Hoof Slippage Detecting Method Based on Enhanced DeepLabCut Model [J]. Smart Agriculture, 2024, 6(5): 153-163.
[14]	LI Minghuang, SU Lide, ZHANG Yong, ZONG Zheying, ZHANG Shun. Automatic Measurement of Mongolian Horse Body Based on Improved YOLOv8n-pose and 3D Point Cloud Analysis [J]. Smart Agriculture, 2024, 6(4): 91-102.
[15]	WENG Zhi, FAN Qi, ZHENG Zhiqiang. Automatic Measurement Method of Beef Cattle Body Size Based on Multimodal Image Information and Improved Instance Segmentation Network [J]. Smart Agriculture, 2024, 6(4): 64-75.