Kaiwen Yuan
Kaiwen Yuan
seems something wrong with your `lmdb ` loading? Wrong path? Use absolute path could be safer?
@gopi77 I still think it's the wrong path error. You should check your config file. I don't think 8GB memory is enough for training. Most straight forward way I think...
Similar case [here](https://github.com/torrvision/crfasrnn/issues/98#issuecomment-302171131)
Using either anaconda or docker can make your life way easier. In case you are interested in doing this in docker, I just wrapped up a dockerized version on top...
@ccomkhj did you solve this ? I believe this [line is the source of the issue](https://github.com/open-mmlab/mmdetection/commit/37022f5ec1acac3b21249c81e9d7cbabda1b18c9#diff-136b3f31689b4e74b64162ee5eb155777202e00231010a752df486a097a8633dR105), which assumes the `ann_file` and `data_prefix['img']` are two str, instead of list.
> > ``` > > train_dataloader = dict( > > batch_size=2, > > num_workers=2, > > persistent_workers=True, > > sampler=dict(type='DefaultSampler', shuffle=True), > > batch_sampler=dict(type='AspectRatioBatchSampler'), > > dataset=dict( > > type=dataset_type,...
@Dawson-huang Hey I actually plan to add a LSTM layers after the CNN feature extraction. After that, the results should be much better.
Sure! It will be great to have someone work with together! Feel free to submit any pull request.
@BluehackRano Sorry for delay of reply. Was in vacation and had no access to my Mac. So the problem is as it says, you need to either map your model...
@BluehackRano Any follow up man?