search

SimonThomine / RememberingNormality

Unofficial implementation of "Remembering Normality: Memory-guided Knowledge Distillation for Unsupervised Anomaly Detection"
7 stars 0 forks source link

change resnet 18 to wide_resnet50_2 #4

Closed wjx213 closed 3 weeks ago

wjx213 commented 3 weeks ago

Hello,Thank you very much for your work. However, when I change 'resnet18' to 'wide_resnet50_2' in the 'config.yaml' file and run 'trainRM_RD.py', I encounter an error. May I ask how I can adjust other parameters if I want to change the backbone to 'wide_resnet50_2' so that the model can run? Thank you very much for your response.

SimonThomine commented 3 weeks ago

Hello, you're right, I forgot to adapt the memory modules for the wide resnet architecture (bottleneck instead of basicblock). The problem is now fixed. Please let me know if you encounter any other problem.

wjx213 commented 3 weeks ago

thank you for your reply, the problem has been fixed，but the results differ slightly from the paper, but I'm still very grateful for your work!

SimonThomine commented 3 weeks ago

No problem :) Let me know if you find any problem in the code that could explain the slight difference in results.