search

yaoyao-liu / meta-transfer-learning

TensorFlow and PyTorch implementation of "Meta-Transfer Learning for Few-Shot Learning" (CVPR2019)
https://lyy.mpi-inf.mpg.de/mtl/
MIT License
731 stars 147 forks source link

Data Loader for CIFAR-FS #50

Closed YoungBigBird1985 closed 3 years ago

YoungBigBird1985 commented 3 years ago

Hi Liu, I want to run your code on CIFAR-FS, which has a 32 32 3 shape of Image, and cannot fit to your code. I try to change the parameter img_size from 84 to 32, but it has error with "Negative dimension size caused by subtracting 5 from 2 for 'pretrain-model/AvgPool' (op: 'AvgPool') with input shapes: [64,2,2,512]". Please help address this issue.

The data is downloaded from http://www.cs.toronto.edu/~kriz/

yaoyao-liu commented 3 years ago

Thanks for your interest in our work.

If you hope to run the code on CIFAR-FS, I suggest you resize the images to 84x84. This pre-processing is applied in many related papers, e.g., https://git.io/JOZI8.

Best, Yaoyao

YoungBigBird1985 commented 3 years ago

thanks.

YoungBigBird1985 commented 3 years ago

thanks.