Fixing the train-test resolution discrepancy

06/14/2019
by   Hugo Touvron, et al.
0

Data-augmentation is key to the training of neural networks for image classification. This paper first shows that existing augmentations induce a significant discrepancy between the typical size of the objects seen by the classifier at train and test time. We experimentally validate that, for a target test resolution, using a lower train resolution offers better classification at test time. We then propose a simple yet effective and efficient strategy to optimize the classifier performance when the train and test resolutions differ. It involves only a computationally cheap fine-tuning of the network at the test resolution. This enables training strong classifiers using small training images. For instance, we obtain 77.1 on 128x128 images, and 78 Conversely, when training a PNASNet at resolution 331x331 and further optimizing for test resolution 480x480, we obtain a test top-1 accuracy of 83.7

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset