Facial Expression Recognition via ResNet-18

A facial expression recognizer using modified ResNet-152

“…OA HWT [9] 78.37+1.50 CSO [10] 89.49+0.76 BBO [42] 93.79+1.24 Modified ResNet-152 96.44+0.56 [11] and ResNet-50 [12] as facial expression classifiers to achieve the accuracy of 94.80+1.43%, 95.39+1.41%, respectively. Based on Table 4 and Figure 11, we can see that the "Modified ResNet-152" achieves higher accuracy.…”

“…OA ResNet-18 [11] 94.80+ 1.43 ResNet-50 [12] 95.39+ 1.41 Modified ResNet-152 96.44+0.56 Wenle Xu, Rayan S Cloutier (SDN) [48] is an approach to networking that uses softwarebased controllers or application programming interfaces. In the future, we shall combine SDNs with this task.…”

“…Yang [10] introduced cat swarm optimization (CSO) and achieved 89.49% accuracy. Li used ResNet-18 [11] and ResNet-50 [12] for feature extraction and classification. The methods mentioned above further improve the accuracy of model.…”

An Analysis Model of Learners’ Online Learning Status Based on Deep Neural Network and Multi-Dimensional Information Fusion

“…The extracted eye images are uniformly resized to 24 × 24 pixels in size. In this experiment, the ResNet18 network model structure was built in the Pytorch environment [28], the network contains 17 convolutional layers and 1 fully connected layer (not including pooling layer and BN layer), it is divided into 6 modules. The first module is the first convolutional layer (Conv1), it uses 64 convolution kernels with a convolution size of 7 × 7 to perform convolution operations, and performs maximum pooling after completion; Modules Conv2_x, Conv3_x, Conv4_x, Conv5_x all contain two consecutive convolution layers, using 64, 128, 256, 512 convolution kernels for convolution operations, and each convolution layer uses a 3 × 3 size convolution kernel, and also performs shortcut processing on the output of the previous layer.…”