Xin Wang

I need this feature, too. In my project, I use some pretrained model to generate labels. But these boxes' locations are not correct. I need to move them together sometimes.

界面完全可以用PyQt来写，比TK好用。

我是想知道马赛克模型是用什么数据训练出来的

赞，先下载看看，明天试着跑一下看啥效果。

Actually, the best way to download such datasets is to use cloud server. I used to use AWS to do this. However, there is still a problem waiting for us....

I wrote a script yesterday to download the dataset on AWS. After 12 hours, 600k images have been downloaded. (About 20% of the image links no long exist.) Even I...

Finally, I finished. It's about 50G, with about 17% links expired.

有个center class，里面存储的就是feature center，最初是随机初始化得到的。

> center最初是怎么计算出来的，在哪里实现的？ 我用另一种方式实现了center loss，欢迎使用讨论。

in line 77 and 79 in main.py, there is a little problem. > fpath=os.path.join(args.out_dir, '%s-train-epoch-%s.png' % (args.prefix, e))) Isn't it better to use "epoch-%d.png" rather than "epoch-%s.png"?