search

sagieppel / Reconstruct-image-from-sparsely-sampled-pixels-using-fully-convolutional-neural-network-FCN-

Take image in which only small fraction of the pixels are known and reconstruct (upsample) the full image using fully convolutional neural net (tensorflow implementation).
MIT License
1 stars 2 forks source link

regarding running training script #1

Open huaiyanggongzi opened 7 years ago

huaiyanggongzi commented 7 years ago

Hi, thanks for sharing the code.

I am running the train script, by using python Train.py. Looks like it is running, are there any ways to monitor this running process, like printing out how many training iterations having been passed? I am not sure how long it will take for finishing the training.

This the current output,

`python Train.py Finished Running setting up vgg initialized conv layers ... Reading images list Number of Train images=20210 2017-07-04 15:16:14.178105: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations. 2017-07-04 15:16:14.178130: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations. 2017-07-04 15:16:14.178138: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.

`

sagieppel commented 7 years ago

The 3 final messages are related to Tensorflow installation and shouldn't effect you if you use GPU (if you dont use GPU training will take you forever). The training loss should be printed on screen once every ten training operation, which should accure once every few seconds, if it doesn't appear something strange happening.