KERN
KERN copied to clipboard
yuweihao
pretrain_detector
您好，我在VG预训练detector时遇到这样的错误： from dataloaders.mscoco import CocoDetection,CocoDataLoader
ModuleNotFoundError: No module named 'dataloaders
Hi @wuxueliang123 ，this repo does not depend on mscoco. This part is from neural-motifs and you can refer to that repo.
Hello, I don't know why the pre-training model has been downloaded and shown to be damaged. I have downloaded it many times. Why is this? Could you please send it to me? Thank you
Hi @L6-hong , no need to decompress the model, just load it by torch.load('vg-faster-rcnn.tar').
Ok, thank you very much for your reply. I have another question I want to ask you, and I hope you can give me some advice. I found that there was a memory overflow when I was doing the experiment, so I reduced the data set, but there will still be a memory problem. Now I have reduced the data to 5 pictures and changed the batchsize to 1, but there are still problems: RuntimeError:cuda runtime error(2)：out of memory at /opt/conda/conda-bld/pytorch_15123874934/work/torch/lib/THC/generic/THCStorage.cu:58 Do you have a solution? Thank you.
------------------ 原始邮件 ------------------ 发件人: "yuweihao/KERN" <[email protected]>; 发送时间: 2020年12月16日(星期三) 晚上8:13 收件人: "yuweihao/KERN"<[email protected]>; 抄送: "李建红"<[email protected]>;"Mention"<[email protected]>; 主题: Re: [yuweihao/KERN] pretrain_detector (#19)
Hi @L6-hong , no need to decompress the model, just load it by torch.load('vg-faster-rcnn.tar').
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.