Rafikg

@bddppq You mean that when we get this warning for example `Engine CUDNN is not available for operator MaxPool`, it means that `CUDNN` does not have `MaxPool` operator ?

@musab1234, Hi, did you get the model files? Could you share it?

@zyoohv could you elaborate the use of pixel_std and this instruction in `_xywh2cs` function ``` if center[0] != -1: scale = scale * 1.25 ```

@PaTricksStar , I have the same question. Also, what about the scale = scale*1.25 in this function ``` def _xywh2cs(self, x, y, w, h): center = np.zeros((2), dtype=np.float32) center[0] =...

Hi @fyu, I am interested in the segmentation network and I want to know the number of features maps (ouput_dim) that you use during the up-sampling branch

@rludlow Thanks for the explanation. I have compared between the cmu model implemented here and mobilenet_thin implemented in this github[openpose](https://github.com/ildoonet/tf-pose-estimation). For the prediction time with same size of the image...

In the original paper, they did not say anything about the training scheme (optimizer, learning rate, epochs,...). I reimplement your model with tf.keras and everything was good except the MultiSGD...

Thanks, @michalfaber, It is what I expected.

@DeeperDeeper, Thanks, I definitely agree!

@PrithivirajDamodaran @fernandocamargoai same problem!