SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition
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...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-02-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/14/4/1399 |
| _version_ | 1797299043636871168 |
|---|---|
| author | Min Hao Quan Sun Chuanzhong Xuan Xiwen Zhang Minghui Zhao |
| author_facet | Min Hao Quan Sun Chuanzhong Xuan Xiwen Zhang Minghui Zhao |
| author_sort | Min Hao |
| collection | DOAJ |
| description | 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 1<sup>2</sup> × 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. |
| first_indexed | 2024-03-07T22:44:51Z |
| format | Article |
| id | doaj.art-b1e6749e2ae04270b6fbe209b047b746 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-07T22:44:51Z |
| publishDate | 2024-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-b1e6749e2ae04270b6fbe209b047b7462024-02-23T15:05:54ZengMDPI AGApplied Sciences2076-34172024-02-01144139910.3390/app14041399SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial RecognitionMin Hao0Quan Sun1Chuanzhong Xuan2Xiwen Zhang3Minghui Zhao4Department of Mechanical and Electrical Engineering, Inner Mongolia Agricultural University, Hohhot 010010, ChinaDepartment of Mechanical and Electrical Engineering, Inner Mongolia Agricultural University, Hohhot 010010, ChinaDepartment of Mechanical and Electrical Engineering, Inner Mongolia Agricultural University, Hohhot 010010, ChinaDepartment of Mechanical and Electrical Engineering, Inner Mongolia Agricultural University, Hohhot 010010, ChinaDepartment of Mechanical and Electrical Engineering, Inner Mongolia Agricultural University, Hohhot 010010, ChinaTo 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 1<sup>2</sup> × 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.https://www.mdpi.com/2076-3417/14/4/1399sheep face recognitionSSDSqueezeNetattention mechanismloss function |
| spellingShingle | Min Hao Quan Sun Chuanzhong Xuan Xiwen Zhang Minghui Zhao SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition Applied Sciences sheep face recognition SSD SqueezeNet attention mechanism loss function |
| title | SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition |
| title_full | SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition |
| title_fullStr | SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition |
| title_full_unstemmed | SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition |
| title_short | SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition |
| title_sort | squeezenet an improved lightweight neural network for sheep facial recognition |
| topic | sheep face recognition SSD SqueezeNet attention mechanism loss function |
| url | https://www.mdpi.com/2076-3417/14/4/1399 |
| work_keys_str_mv | AT minhao squeezenetanimprovedlightweightneuralnetworkforsheepfacialrecognition AT quansun squeezenetanimprovedlightweightneuralnetworkforsheepfacialrecognition AT chuanzhongxuan squeezenetanimprovedlightweightneuralnetworkforsheepfacialrecognition AT xiwenzhang squeezenetanimprovedlightweightneuralnetworkforsheepfacialrecognition AT minghuizhao squeezenetanimprovedlightweightneuralnetworkforsheepfacialrecognition |