tf.contrib.slim evaluation

hyperconnect / MMNet

Code for Towards Real-Time Automatic Portrait Matting on Mobile Devices

https://arxiv.org/abs/1904.03816v1

Apache License 2.0

177 stars 36 forks source link

tf.contrib.slim evaluation #4

Open prnvjb opened 5 years ago

prnvjb commented 5 years ago

authors, thanks for opensourcing your code. I would like to know why did you choose tf.contrib.slim instead of tf.layers or just tf.keras? I'm trying to reimplement your using tf.keras which could be more simpler for more expremintation with tf.eager_execution to debug easily? what are implecations I should take to use tf.keras? Thanks!!

shurain commented 5 years ago

Our internal framework is still dependent on older TF versions. We're currently migrating to newer versions and plan to adopt appropriate coding style in the future, but for now, we decided that releasing our code was more important than waiting for the migration to complete.

tony2278 commented 5 years ago

HI，I made a few modification，and it can run on windows now. But how to use the models I trained?

tony2278 commented 5 years ago

do u have a demo about how to use this models？thanks. my email: loveu1061@126.com

ilous12 commented 5 years ago

Hi @tony2278 @shurain I tried to train for MModel with 4872 images, but never finished for 4 days Please let me know valid training.