1 Introduction
2 Variational autoencoder
2.1 Overview of the VAE algorithm
2.2 Construction of the model of daily facial skin temperature for anomaly detection using VAE
3 Experiment
3.1 Experiment system
3.2 Procedure and condition
4 Analysis methods
4.1 Definition of normal and anomaly
4.2 The model construction
Layers | Function | Filters | Size | Stride |
---|---|---|---|---|
Conv1 | – | 16 | 2\(\times\)2 | 2 |
BatchNorm1 | ReLU | – | – | – |
Conv2 | – | 32 | 2\(\times\)2 | 2 |
BatchNorm2 | ReLU | – | – | – |
FC | – | – | – | – |
Layers | Function | Filters | Size | Stride |
---|---|---|---|---|
FC | – | – | – | – |
BatchNorm1 | ReLU | – | – | – |
ConvTrans1 | – | 32 | 2\(\times\)2 | 2 |
BatchNorm2 | ReLU | – | – | – |
ConvTrans2 | – | 16 | 2\(\times\)2 | 2 |
BatchNorm3 | ReLU | – | – | – |
ConvTrans3 | – | 1 | 4\(\times\)4 | 1 |
ConvTrans4 | – | 1 | 4\(\times\)4 | 1 |
BatchNorm4 | ReLU | – | – | – |
BatchNorm5 | ReLU | – | – | – |