MuseCP
MuseCP
I meet a same problem, have you solve it please?
the caffe I use is yours (FPN/caffe-fpn)
yes you need write a dataloeader yourself. return values are input，groundtruth，and a 1xn tensor contains class imformation。n is class num，1 is the label has the class and 0 is not...
yes I write the code myself too | | 翟墨 邮箱：[email protected] | 签名由 网易邮箱大师 定制 On 04/02/2019 10:14, 1535315854 wrote: @MuseCP Thanks, I have written the dataloader yesterday. I also...
i did not write the evaluate part，only output the final result | | 翟墨 邮箱：[email protected] | 签名由 网易邮箱大师 定制 On 04/02/2019 10:25, 1535315854 wrote: @MuseCP So what's your model performance?...