Research Article
Real-Time Analysis of Youth Emotion Based on Python Language and Smart Sensor Network
Table 4
Constructed convolutional neural networks without fully connected layers.
| | Layers | Output size | Patch size/stride/padding |
| | Conv1 | 32@ | /4 | | (BatchNorm) | | | | Relu1 | | | | Norm1 | | | | Maxpool1 | 32@ | /2 | | Conv2 | 64@ | /1 | | (BatchNorm) | | | | Relu2 | | | | Norm2 | | | | Maxpool2 | 64@ | /2 | | Conv3 | 128@ | /1 | | (BatchNorm) | | | | Relu3 | | | | Conv4 | 256@ | /1 | | Relu4 | | | | Maxpool4 | 256@ | /2 | | Conv5 | 6@ | /1 | | Global Avg.Pooling | 6 | |
|
|
