Giters

CAMMA-public / rendezvous

A transformer-inspired neural network for surgical action triplet recognition from laparoscopic videos.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Reproduce problem

Aawaqqqqq opened this issue · comments

commented

When I reproduced the code for the RDV article(RDV split), my result remains about 0.6784 | 0.4404 | 0.2609 | 0.1923 | 0.1316 | 0.0994 ,which are largely different from the official result 92.0 60.7 38.3 39.4 36.9 29.9, I want to know did you make any changes to the released code, or is there any difference between the actual Settings and the default ones?

commented

Dear user,

Visit the published paper for the details on the hyperparameters used for each type of results.

Thanks