This is my (unofficial) implementation of the paper "One-Class Learned Encoder-Decoder Network with Adverserial Contex Msking for Novelty Detection" in Pytorch. This approche consists of a mask module and a reconstructor that are trained in adverserial manner.
J. T. Jewell, V. Reza Khazaie and Y. Mohsenzadeh, "One-Class Learned Encoder-Decoder Network with Adversarial Context Masking for Novelty Detection," 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2022, pp. 2856-2866, doi: 10.1109/WACV51458.2022.00291.