Open v-shaoningli opened 1 year ago
I second this, I am also having trouble getting the data in the desired .lmdb format.
The lmdb data should be build automatically if you call the train.py function and you have the ImageNet dataset downloaded https://github.com/NVlabs/I2SB/blob/1ffdfaaf05495ef883ece2c1fe991b3049f814cc/train.py#L130-L131
Hi all! Thanks for your work I2SB!
I'm new in this field and not that familiar with your data process. Could you give a script on downloading and preprocessing the mentioned, e.g.,
train_faster_imagefolder.lmdb
data format?Really thanks for your help!