Shiwei Hou

[change_coco_17kps_12kps.zip](https://github.com/murdockhou/Single-Stage-Multi-person-Pose-Machines/files/4290400/change_coco_17kps_12kps.zip) @Dantju 参考下这个文件里的转换代码和转换后的json文件

Hi, 因为之前写的spm label有些问题, 所以修改了源码, 现在正在重新训练

Hi, thanks for your concern. Actually based on paper, the author used L2 loss for center point and SmoothL1 Loss for SPR regression. I tried to use these in code...

> I am facing the same issue, but I am using CPU to train. Me too. I use one GPU and using generator to create dataset, `model.fit` to train model....