The Augmented Image Prior: Distilling 1000 Classes by Extrapolating from a Single Image

TL;DR

This paper demonstrates that neural networks can learn meaningful visual representations from a single image using augmentation and knowledge distillation, achieving surprisingly high accuracy across multiple datasets and modalities.

Contribution

It introduces a framework for training neural networks from a single image with augmentations and knowledge distillation, revealing the power of the augmented image prior.

Findings

Achieves 94% accuracy on CIFAR-10 with a single image.

Attains 69% accuracy on ImageNet from one image.

Extends the approach successfully to video and audio data.

Abstract

What can neural networks learn about the visual world when provided with only a single image as input? While any image obviously cannot contain the multitudes of all existing objects, scenes and lighting conditions - within the space of all 256^(3x224x224) possible 224-sized square images, it might still provide a strong prior for natural images. To analyze this `augmented image prior' hypothesis, we develop a simple framework for training neural networks from scratch using a single image and augmentations using knowledge distillation from a supervised pretrained teacher. With this, we find the answer to the above question to be: `surprisingly, a lot'. In quantitative terms, we find accuracies of 94%/74% on CIFAR-10/100, 69% on ImageNet, and by extending this method to video and audio, 51% on Kinetics-400 and 84% on SpeechCommands. In extensive analyses spanning 13 datasets, we disentangle the effect of augmentations, choice of data and network architectures and also provide qualitative evaluations that include lucid `panda neurons' in networks that have never even seen one.