Hao
https://github.com/tensorlayer/hyperpose/blob/master/requirements.txt @Gyx-One like that?
it is for balancing, you can find the answer on the paper.
@yuyuziliu the VGG is pre-trained using image with scale of 0 ~ 1
1. the inferencing speed is related to the size of you input image, it is normal. 2. for 1st time, tf need time to compile the graph, after that it...
why the training is so slow.. did you use GPU?
hi, @duxufeiqq , simply change the number of channel from 3 to your number then.
hello, you can now update the code base. this repo is updated to tf2.0.0a0 and tl2.
i think 384x384 should be fine for 3840x2160
hi, do you have human images in your training set? because DVI2K doesn't have any human images.
I see, I think your training images are too small, I suggest you to use image more than 1000+ pixels.