Singapore-mor

Results 10 comments of Singapore-mor

> > I run "conda activate envname" in the gitbash, and run "sh make.sh", everything is just ok. > > maybe you should check these: > > > > *...

> ## When I compile make.sh, the ERROR encounters, How to solve this? > Traceback (most recent call last): > File "build_modulated.py", line 40, in > extra_compile_args=extra_compile_args > File "D:\ProgramData\Anaconda3\lib\site-packages\torch\utils\cpp_extension.py",...

> Set the “results” folder in the folder, the “OTB100” folder in the “results” folder, and the result folders that are run out of various trackers in the “OTB100” folder,...

> when I build this project in detectron as a layer, the gpu memory is normal. > when I build this project as a extrrnal library. eg from dcn_v2 import...

> > if using vc++2015 or 2013, pytorch1.0, error occurs in file "pipmpl.h". I don't know why, so I comment the codes from line 121 to line 131, and compiled...

> 二分类的场景也可以使用,但是没有pair的效果好。另外二分类使用需要谨慎(泛化性能)。 谢谢!我理论上感觉不太适合用于二分类,但是还是想试一下,我看其他版本的circleloss的example都是给的pairwise,不太清楚如果我要用于二分类的话,feature和label应该是什么形式的,不知道方不方便加一下qq:2761520483咨询一下?打扰了

> 而且,search和template输出的box也是一样的,有点不理解。 你好,我也有相同的问题,觉得好像没有用到z的数据

> 您好,对于label生成部分的代码,存在一些疑问,询问一下。 > 对于`siamfc.py`中的函数`_create_labels(self, size)`部分,存在如下代码: > > ```python > r_pos = self.cfg.r_pos / self.cfg.total_stride > r_neg = self.cfg.r_neg / self.cfg.total_stride > labels = logistic_labels(x, y, r_pos, r_neg) > ``` >...

> Hi: > I found a problem when I run train_SIamfc.py. Can you please tell me how to solve it? Thanks! Do you solve the problem?

> hi, > Response maps是由相应的update机制(no-update, linear update, and updatenet), 通过输入之前帧的templates,实际预测出的。 > best! 嗯嗯,我知道response maps得到的过程,怎么把它给画成您在论文中给出的那个样子呢?想问下把它可视化出来的方法