How to get the multi output from the dataloader

1

My network accepts multi input data. The data has the same image content, but the resolution is different, like the first input contains the 320*320 image, the second one contains the 160*160 image, the third one contains the 80*80 image. How could I get the batch data from the dataloader?

1 Like
2

You can actually define your own dataset by inheriting the torch.utils.data.Dataset class. So maybe in your __getitem__ function, you can do return image320, image160, image80.
For more details, you can look at this tutorial: http://pytorch.org/tutorials/beginner/data_loading_tutorial.html

1 Like