Variational Autoencoder for Detecting Abnormal Behaviors in Video

This repository is a PyTorch version of our awesome NIPS paper. The following models are implemented:

• AE: Deep convolutional autoencoder
• VAE: Deep convolutional variational autoencoder

Different models can be chosen using python train.py --model <modelName>. You can visualize inter-training results with visdom package.

Requirements

The following python packages are required:

• PyTorch
• visdom

()()
(' ') HAANJU.YOO