search

cvlab-yonsei / MNAD

An official implementation of "Learning Memory-guided Normality for Anomaly Detection" (CVPR 2020) in PyTorch.
333 stars 82 forks source link

Visualization #14

Closed RohDonghyeon closed 3 years ago

RohDonghyeon commented 3 years ago

Thanks for the perfect code. The evaluation was able to proceed without any difficulty. Could you please explain how to visualize the result as a picture as suggested in the paper?

hyunjp commented 3 years ago

Thanks for your attentions to our work.

In Fig. 4, we visualized the magnitudes of the differences between input and output images to show errors of the reconstruction (or prediction) task. We depicted the regions of relatively large errors in red.