zqhang / AnomalyCLIP

Official implementation for AnomalyCLIP (ICLR 2024)
MIT License
275 stars 28 forks source link

Discerning anormal from normal images #42

Open unit-0002 opened 2 months ago

unit-0002 commented 2 months ago

Hey @tianyu0207 and @zqhang,

thank you again for your very helpful and insightful work.

I wanted to know, how it is possible to retrieve the anomaly score and based on that, decide if the image is anomalous or not. I tried searching for it myself, but could not find it.

A quick reference to the loc or variable would be enough.

Thank you very much in advance

unit-0002 commented 1 month ago

To clarify: In your module test_one_example.py, I only get the masking of the anomaly and not the score/ anomaly segmentation. How can I retrieve the anomaly score to segment good from bad images.