search

hcw-00 / PatchCore_anomaly_detection

Unofficial implementation of PatchCore anomaly detection
Apache License 2.0
320 stars 92 forks source link

Detection time is related to the training set size? #8

Closed xfby2016 closed 3 years ago

xfby2016 commented 3 years ago

Hi,I find that the detection time grows longer as the training set grows,Is that reasonable? The problem seems to related to this line of code, https://github.com/hcw-00/PatchCore_anomaly_detection/blob/5680042fd114abbfbef820edef1f61d566a128fd/train.py#L280 Or is that a problem?Thanks

hcw-00 commented 3 years ago

I think it is related to dataset size..

xfby2016 commented 3 years ago

I think it is related to dataset size.. Do you mean it is related to the number of pictures in the training set? So how much of dataset size is appropriate?

hcw-00 commented 3 years ago

Yes, It is related to the number of pictures. I don't know about that. But I think you should do some experiment and find your own conditions.

xfby2016 commented 3 years ago

Okay, thanks, I'll try. By the way,Is it reasonable to set the parameter N of the following code to a fixed value? Like the 50 https://github.com/hcw-00/PatchCore_anomaly_detection/blob/5680042fd114abbfbef820edef1f61d566a128fd/train.py#L280

hcw-00 commented 3 years ago

I don't have deep insight about that. However, in my opinion, it depends on your dataset.