SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition
Source abstract
Abstract
To quickly realize facial identity recognition in sheep, this paper proposes a lightweight detection algorithm based on SSD with a self-constructed dataset. Firstly, the algorithm replaces the VGG16 backbone of SSD with the lightweight neural network SqueezeNet, creating a lightweight hybrid network model. Secondly, an ECA mechanism is introduced at the front end of the pooling layer with a parameter volume of 12 × 1000 into the feature extraction network. Lastly, the smoothL1 loss function is replaced with the BalancedL1 loss function. The optimal model size has been reduced from the original SSD’s 132 MB to 35.8 MB. The average precision is 82.39%, and the mean frame rate is 66.11 frames per second. Compared to the baseline SSD model, the average precision has improved by 2.17%, the model volume has decreased by 96.2 MB, and the detection speed has increased by 7.13 frames per second. Using the same dataset on different target detection models for comparison tests, the average accuracy mean values are improved by 2.17%, 3.63%, and 1.30% compared to the SSD model, Faster R-CNN model, and Retinanet model, respectively, which substantiates a better overall performance compared to the pre-improvement model. This paper proposes an improved model that significantly reduces the model size and its computation while keeping the model performance at a high level, providing a methodological reference for the digitization of livestock farming.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Animal Behavior and Welfare Studies · Image Processing and 3D Reconstruction · Industrial Vision Systems and Defect Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Department of Mechanical and Electrical Engineering, Inner Mongolia Agricultural University, Hohhot 010010, China
Funding
basic research operating expenses of colleges and universities directly in the Inner Mongolia Autonomous Region
References (26)
Show source-provided references
1. Visual identification of individual Holstein-Friesian cattle via deep metric learning10.1016/j.compag.2021.106133 ↗
2. A review of the latest research results on individual cow recognition
3. Jiang, H. (2019). Research on Natural Grassland Utilization Assessment Method Based on 3S Technology. [Master’s Thesis, Shihezi University].
4. Application of geovisual analytics to modelling the movements of ruminants in the rural landscape using satellite tracking data10.1080/17538947.2013.872703 ↗
5. Biometric facial identification using attention module optimized YOLOv4 for sheep10.1016/j.compag.2022.107452 ↗
6. Zhang, X., Xuan, C., Xue, J., Chen, B., and Ma, Y. (2023). LSR-YOLO: A High-Precision, Lightweight Model for Sheep Face Recognition on the Mobile End. Animals, 13.10.3390/ani13111824 ↗
7. Nighttime hare monitoring method based on infrared thermal imaging and improved YOLOV3
8. Yan, H. (2019). Research on Pig Face Recognition under Unrestricted Conditions Based on Machine Learning. [Ph.D. Thesis, Shanxi Agricultural University].
9. Feng, M., Cao, D., and Li, J. (2020). Research on the design of pig face recognition APP based on ResNet50 model. Comput. Age, 46–50.
10. Multi-objective pig detection based on feature pyramid attention and deep convolutional network
11. Detection of group-reared hogs using dual-attention feature pyramid network
12. Two-stage method based on triplet margin loss for pig face recognition10.1016/j.compag.2022.106737 ↗
13. A lightweight deep learning model for cattle face recognition10.1016/j.compag.2022.106848 ↗
14. Yang, J. (2022). Research and Implementation of Lightweight Sheep Face Recognition Method Based on Attention Mechanism. [Master’s Thesis, Northwest Agriculture and Forestry University of Science and Technology].10.1109/ITAIC54216.2022.9836724 ↗
15. Zhou, L. (2022). Research on Sheep Face Recognition Method Based on Lightweight Neural Network. [Master’s Thesis, Northwest Agriculture and Forestry University of Science and Technology].
16. Suitability of feeding and chewing time for estimation of feed intake in dairy cows10.1017/S1751731115001366 ↗
17. Evaluation of eating and rumination behaviour using a noseband pressure sensor in cows during the peripartum period10.1186/s12917-014-0195-6 ↗
18. Assessment of retinal recognition technology as a biometric method for sheep identification10.1016/j.compag.2007.07.010 ↗
19. A regularity-based algorithm for identifying grazing and rumination bouts from acoustic signals in grazing cattle10.1016/j.compag.2018.06.021 ↗
20. Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.Y., and Berg, A.C. (2015). SSD: Single Shot MultiBox Detector, Springer International Publishing.10.1007/978-3-319-46448-0_2 ↗
21. Tian, Z. (2021). Research on Garbage Classification Algorithm Based on Deep Learning. [Master’s Thesis, Harbin Institute of Technology].10.23919/CCC52363.2021.9549336 ↗
22. Iandola, F.N., Han, S., Moskewicz, M.W., Ashraf, K., Dally, W.J., and Keutzer, K. (2016). SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and 1MB model size. arXiv.
23. Wang, Q., Wu, B., Zhu, P., Li, P., Zuo, W., and Hu, Q. (2020, January 14–19). ECA-Net: Efficient channel attention for deep convolutional netural networks. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, WA, USA.10.1109/CVPR42600.2020.01155 ↗
24. Transmission line insulator identification based on improved Libra-RCNN
25. Improved YOLO V4 model based on fused coordinate information to recognize cow face
26. Song, S., Liu, T., Wang, H., Hasi, B., Yuan, C., Gao, F., and Shi, H. (2022). Using Pruning-Based YOLOv3 Deep Learning Algorithm for Accurate Detection of Sheep Face. Animals, 12.10.3390/ani12111465 ↗