What makes ImageNet good for transfer learning?

TL;DR

This paper empirically investigates what properties of ImageNet contribute to its effectiveness for transfer learning, analyzing factors like data size, class granularity, and class count to understand their impact on feature quality.

Contribution

The study systematically evaluates how different aspects of ImageNet data influence transfer learning performance, providing insights into optimal data configurations.

Findings

Most data choices previously thought critical have minimal impact on transfer performance.

Coarse classes can be as effective as fine-grained classes for feature learning.

More classes or more examples per class both improve transfer results, but with diminishing returns.

Abstract

The tremendous success of ImageNet-trained deep features on a wide range of transfer tasks begs the question: what are the properties of the ImageNet dataset that are critical for learning good, general-purpose features? This work provides an empirical investigation of various facets of this question: Is more pre-training data always better? How does feature quality depend on the number of training examples per class? Does adding more object classes improve performance? For the same data budget, how should the data be split into classes? Is fine-grained recognition necessary for learning good features? Given the same number of training classes, is it better to have coarse classes or fine-grained classes? Which is better: more classes or more examples per class? To answer these and related questions, we pre-trained CNN features on various subsets of the ImageNet dataset and evaluated transfer performance on PASCAL detection, PASCAL action classification, and SUN scene classification tasks. Our overall findings suggest that most changes in the choice of pre-training data long thought to be critical do not significantly affect transfer performance.? Given the same number of training classes, is it better to have coarse classes or fine-grained classes? Which is better: more classes or more examples per class?