there was a probblem using my own datasets:RuntimeError: Sizes of tensors must match except in dimension 3. Got 425 and 423 (The offending index is 0)

Scalsol / mega.pytorch

Memory Enhanced Global-Local Aggregation for Video Object Detection, CVPR2020

Other

566 stars 115 forks source link

there was a probblem using my own datasets:RuntimeError: Sizes of tensors must match except in dimension 3. Got 425 and 423 (The offending index is 0) #108

Open LuoXiang-LX opened 2 years ago

LuoXiang-LX commented 2 years ago

Does anyone konw how to solve thie problem?

lemyanh2612 commented 2 years ago

Hi, did you solve this problem? I also had the same problem. This error happens only when training with RDN, FGFA, MEGA to over 20000 iterations. You should check tensor size of img['cur'] and img['ref']. I ignored all cases which get the different tensor sizes. Solved this problem.

meikorol commented 1 year ago

HI，how do you make own dataset and place it?I download vid dataset and it has videos、 jpgs of every frame and xmls,shall we make our dataset and place it like this?I think its so comlicated,and how can we make the txt files in the imageset?