lmdb.Error when create dataset

[deep-practice]

Traceback (most recent call last): File "~/crnn-pytorch-master/tool/create_dataset.py", line 135, in createDataset(args.out, image_path_list, label_list) File "~/crnn-pytorch-master/tool/create_dataset.py", line 55, in createDataset env = lmdb.open(outputPath, map_size=1099511627776) lmdb.Error: ~/fake/lmdb: \ufffd\ufffd\ufffd\u033f\u057c\u4cbb\ufffd\u3863

[Holmeyoung]

Hi, what's the mode you use to create lmdb, file or folder? Did you do as readme ~~~

[deep-practice]

I use folder mode,and the dataset is loaded correctly. The code fails when run to the line env = lmdb.open(outputPath, map_size=1099511627776) After that,I check the lmdb output folder,there are 2 files:lock.mdb and data.mdb with length 8kb

[Holmeyoung]

Hi, did you use python3 to create? Please try to update the lmdb and set themap_size = 9000000000 to see if this will help

[deep-practice]

changing map_size to 9000000000 solves the problem,but why?

[Holmeyoung]

It defines the max storage space. Maybe your disk didn't have 1TB space

[deep-practice]

Thanks so much,I can run the trainning process now.Another question,if the sample in my dataset has variable space between characters,for example,"Never too old too learn ",is it ok to train a good model?

[Holmeyoung]

Yeah, it's OK.

[deep-practice]

Thanks so much.