thomas.hui.yang

found the reason PIL sees two frames in the image. [image read by skimage.io.imread have the wrong shape](https://github.com/scikit-image/scikit-image/issues/2406#issuecomment-271472754)

@cweill i changed it to 100 (the default number for tf.estimator.EvalSpec). it stayed almost the same, except print out evaluation progressbar. before: (steps=None): > INFO:tensorflow:Starting evaluation at 2018-11-07-09:00:11 > INFO:tensorflow:Graph...

@cweill Here is the full log. [full_log.txt](https://github.com/tensorflow/adanet/files/2560570/full_log.txt)

Tensorflow v1.10.0 improved local run behavior in tf.estimator.train_and_evaluate which does not reload checkpoints for evaluation, see [release.md](https://github.com/tensorflow/tensorflow/blob/master/RELEASE.md). It maybe the reason that adanet doesn't work well on tf 1.8.0.

@cweill Sorry I can't, for it is for production sometimes also. But I tried **on my own laptop** with **no gpu support** and TRAIN_STEPS=1000: - when using TF 1.8, the...

with same code on windows 10 and ubuntu 16.04, tensorflow v1.4.0, 20000 iterations: memory error occurs in ubuntu 16.04 while no error occurs in windows 10

add python path to your env variables

1 means it is a positive sample

try move all files in a folder, and move the folder into /home//. be sure you have the read and write permissions then cd to the folder and run `python...

edit win_build_opencv.cmd try: line 41(building opencv): add -DWITH_FFMPEG=ON