search

lmb-freiburg / hand3d

Network estimating 3D Handpose from single color images
GNU General Public License v2.0
802 stars 252 forks source link

is there a custom layer? #14

Closed dedoogong closed 6 years ago

dedoogong commented 6 years ago

I would like to start to port it to caffe(My goal is to implement just forward passes, not for training).

I have read your paper roughly and I wonder there are some custom layers which are not included in the existing TF or caffe layers.

Plus, I wonder, HandSegNet can be replaced with SSD hand detector as I thought HandSegNet is just for detecting hands and after that the hand-cropped patchs from the original image(not from the feature map of the last layer of the HandSegNet) is transferred to PoseNet.

dedoogong commented 6 years ago

Sorry but I have one more question. the pretrained weights are all saved in pickel format. But I need ckpt to convert to caffemodel format or pb format. Can I use tf.train.Saver().save to re-save the model after loading the pickle weights?

zimmerm commented 6 years ago

I didn't implement any custom operations for Tensorflow. So you can use a off the stock Tensorflow version.

Indeed you can replace HandSegNet with any other detection algorithm.

Yes, you can save the weights into a checkpoint after you initialized them from the pickle.