How to improve accuracy of the localisation of anomalies at gradient map?

marco-rudolph / differnet

This is the official repository to the WACV 2021 paper "Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows" by Marco Rudolph, Bastian Wandt and Bodo Rosenhahn.

215 stars 67 forks source link

How to improve accuracy of the localisation of anomalies at gradient map? #34

Closed volkov-maxim closed 3 years ago

volkov-maxim commented 3 years ago

Hi, Marco!

When I visualise the gradient map, I get incorrect localisation of anomalies (see examples below). For me, the ideal is when gradients repeat the contour of the anomaly. Is it possible?

? How can I improve the accuracy of the localisation of anomalies on gradient map?

0_orig 0_gradient_map 5_orig 5_gradient_map 19_orig 19_gradient_map 13_orig 13_gradient_map

marco-rudolph commented 3 years ago

Unfortunately, there are some failure cases for the localization. I would recommend techniques that are specialized on localization like this technique.