About GPU memory requirement.
Closed this issue · 2 comments
How much GPU memory is recommanded when training the model using ScutHead dataset? I've found a memory insufficience error when trying to train the model on a idel GPU with 4041MB memory and batch size 1.
Besides, there is some minor bugs in the "head_detection/data/create_scuthead.py" file, as args.dset_path is defined but args.dset_dir is used (and the same bugs to the arg.save_path and args.out_path).
Hello,
I'm not sure about the exact memory requirement. You can disable the transpose convolution to save a significant amount of memory. Note that this might lead to slightly inferior performances. Other than that, you can also reduce the input size (resolution). Again this can hurt the accuracy. I ran my code on 2080Ti, with ~11Gb of memory.
Thanks for pointing out the bug. Honestly, the python file is a copy paste from my notebook, so I could've made some mistakes there. Will be much happy to take a PR :-)
Thank you for your reply.