Laser & Optoelectronics Progress, Volume. 56, Issue 24, 241502(2019)
Facial Expression Recognition Algorithm Based on Cosine Distance Loss Function
Figures & Tables(8)
Fig. 2. Depth feature diagrams learned by three loss functions. (a) Softmax loss; (b) Center loss; (c) Island loss
Fig. 3. Comparison of traditional Softmax loss function and AM-Softmax loss function
Fig. 4. Shematic of classification boundaries of Island loss function and loss function based on cosine distance
Fig. 5. Confusion matrix of experimental results of network model based on cosine distance loss function
Table 1. Accuracy of facial expression recognition when
s =10 andm takes different valuesView tableTable 1. Accuracy of facial expression recognition when
s =10 andm takes different valuesm Accuracy /% 0.25 80.771 0.30 80.893 0.35 81.578 0.40 79.985 0.45 79.341 0.50 78.432
Table 2. Accuracy of facial expression recognition when
m =0.35 ands takes different valuesView tableTable 2. Accuracy of facial expression recognition when
m =0.35 ands takes different valuess Accuracy /% 5 80.443 10 81.578 15 82.811 20 83.196 25 81.239 30 80.025
Table 3. Accuracy of facial expression recognition under different loss functions
View tableTable 3. Accuracy of facial expression recognition under different loss functions
Method Accuracy /% DLP-CNN 80.897 GAN-Inpainting 81.874 Softmax loss 78.725 Island loss 81.813 Cosine distance loss 83.196
Tools
Get Citation
Copy Citation Text
Huihua Wu, Hansong Su, Gaohua Liu, Shen Li, Xiao Su. Facial Expression Recognition Algorithm Based on Cosine Distance Loss Function[J]. Laser & Optoelectronics Progress, 2019, 56(24): 241502
Paper Information
Category: Machine Vision
Received: Apr. 25, 2019
Accepted: Jun. 5, 2019
Published Online: Nov. 26, 2019
The Author Email: Liu Gaohua (suppig@126.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20