search

hq-deng / RD4AD

Anomaly Detection via Reverse Distillation from One-Class Embedding
MIT License
174 stars 31 forks source link

How to do your ablation study #1

Closed 22strongestme closed 2 years ago

22strongestme commented 2 years ago

Dear Author I want to do the ablation study of your network ,but i have some question about it 。 ablation_table if I want to use the single-layer features M1, how should I do the ablation study,like the first pic below or the latter one? ablation1 ablation2

hq-deng commented 2 years ago

Tis ablation study is from the first one. Please set "inputs[0], outputs[0]" in test.py line#83: "anomalymap, = cal_anomaly_map(inputs, outputs, img.shape[-1], amap_mode='a')". Similar operation for other layers.

hq-deng commented 2 years ago

The second graph is not reasonable. But you can remove the multi-scale fusion. By replace "decoder(bn(inputs))" by "decoder(inputs)" for both main.py and test.py.

22strongestme commented 2 years ago

thanks a lot ，I'll try it !!!