search

longcw / yolo2-pytorch

YOLOv2 in PyTorch
1.54k stars 421 forks source link

next_batch: still epoch xxx start #17

Closed AceCoooool closed 7 years ago

AceCoooool commented 7 years ago

When I run python demo.py, I always get the epoch xxx start..., and the memroy up!~ I see the imdb.py ， it's confuse to me. Could everyone give me some advice? (I use python3.6)

longcw commented 7 years ago

demo.py don't use imdb.py. What's the error information? I suggest you write your own dataloader using torch.utils.data.Dataset. It is much easier to use than the imdb.py.

AceCoooool commented 7 years ago

thank you for your reply. I can run the demo, and I want to train by myself, so i run the train.py(and it using the ImageDataset in imdb--- the superclass of VOCDataset). No error information, but it always run in the part: except(StopIteration, AttributeError) in the next_batch，so i can not get the batch, and the while never stop(pool.imap to the memory, which cause the memory up~).

Thank you for the advise, I will try to write my own dataloder. (and after that, i will close the issue)

hyer commented 6 years ago

I run the train.py on GPU mode and I found the memory always increase, which make the train process killed. Can any one help. thx!