EOFError

[donglin8506]

Hello,@JingChaoLiu
I occured the problem about 'EOFError' when I train train.py. I can train without any error for much time (such as 24 hours), but after that , occurs problem as follow:

Surprisingly，After the problem occurs, I interrupt code, then I still python train_net.py for a period of time， but then error. Repeated appearance. Cycle.

[JingChaoLiu]

To reproduce the F-score of this paper, it's recommended to git clone or fork from the maskrcnn benchmark repository, instead of this repository.

In order to transfer the trained model to the maskrcnn benchmark framework and achieve the same F-score mentioned in the paper, there are some unnecessary modifications to keep consistency with the company-private framework, like the pixel value is in the range of [0, 1] rather than [0, 255], PIL.Image.Open() to cv2.imread(), StrideIn1x1=False, see this commit for more details.

1. Haven't been tested in the training stage, I'm not sure whether these meaningless modifications will cause some unexpected influences.
2. The maskrcnn benchmark repository which adds new features and fixes old bugs continuously, is more trustable.