"RuntimeError: The expanded size of the tensor (1200) must match the existing size (1199) at non-singleton dimension 1. Target sizes: [600, 1200, 3]. Tensor sizes: [600, 1199, 3] "

Hi All, got a Runtime error while training on the Open Images dataset after around 9000 iterations in the first epoch.

issue in: https://github.com/jwyang/faster-rcnn.pytorch/blob/358cecacf876717ff13988dc6396de10e265279c/lib/roi_data_layer/roibatchLoader.py#L177

File "/home/munzueta/Kaggle/ObjectDetection2019/faster-rcnn.pytorch/lib/roi_data_layer/roibatchLoader.py", line 177, in __getitem__ padding_data[:, :data_width, :] = data[0]

RuntimeError: The expanded size of the tensor (1200) must match the existing size (1199) at non-singleton dimension 1. Target sizes: [600, 1200, 3]. Tensor sizes: [600, 1199, 3]

I am not sure what might cause it, as I have not change the implementation of roibatchLoader.py yet and I been able to run the training for around 3h… Wondering with the np.ceil function might be causing the issue…

I could ignore it as the training works, but crashes when encountering a corner case

thanks!

[eleflea]

It might because a bug in croping image, it change data’s shape. I add a reassignment line to data_height and data_width. It seems solve this problem. https://github.com/jwyang/faster-rcnn.pytorch/blob/624608fd2f1fb332ef585062cfe51fabf718430d/lib/roi_data_layer/roibatchLoader.py#L160 data_height, data_width = data[0].size(0), data[0].size(1) add this line after line 160.

[peach1995]

i delete all files of .pkl in caches in the folder,and made it. i dont konw why ,maybe the older .pkl made it worse,you may try

I have fixed the problem with it, thank you!