Anomaly Detection in Videos Using Two-Stream Autoencoder with Post Hoc Interpretability

Figure 2

Spatial model architecture. The rightmost number indicates the output size of each layer.