Open woailuoshenqi opened 5 years ago
For those who are curious. I trained the model using this code and tested the performance using codes provided by foolwood. The following are the AUC I got: OTB2013: 0.5147 OTB2015: 0.4845
I don't think different training dataset will give this much difference. So, I am suspecting either the codes are incompatible (between this repo and foolwood's) or hyper-parameter setting problem.
I have not finished the test section yet. When I finished, I will let you know (: