Uninformed Students

Introduction - 介绍

A simple and incomplete implementation of paper:
MVTec, Uninformed Students: Student–Teacher Anomaly Detection with Discriminative Latent Embeddings. CVPR, 2020.
arXiv:1911.02357

Another implementation repo: https://github.com/denguir/student-teacher-anomaly-detection

Requirements - 依赖

python3
pytorch~=1.3
torchvision
numpy
opencv-python

Usage - 用法

Prepare datasets

imagenet (any image dataset)
MVTec_AD

Train a teacher network

choose a patch_size from (17, 33 or 65) and
python teacher_train.py

Train a student network

choose a patch_size(the teacher net should have been pretrained), and set st_id
python student_train.py

Evaluate

python evaluate.py
the res.jpg will be saved to the current directory.

TODO

metric learning and descriptor compactness in teacher_train.py
complete evaluate.py
...

Reference - 参考

https://github.com/erezposner/Fast_Dense_Feature_Extraction
https://github.com/denguir/student-teacher-anomaly-detection