Why the PRO should be normalized from 0 ~ 1 to 0 ~ 0.3? - Githubissues

hq-deng / RD4AD

Anomaly Detection via Reverse Distillation from One-Class Embedding

MIT License

174 stars 31 forks source link

Why the PRO should be normalized from 0 ~ 1 to 0 ~ 0.3? #29

Open William9Baker opened 1 year ago