What makes ImageNet good for transfer learning?

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?

References in corpus (5)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Object Detectors Emerge in Deep Scene CNNs
• Analyzing the Performance of Multilayer Neural Networks for Object Recognition
• Understanding deep features with computer-generated imagery
• Unsupervised Feature Learning from Temporal Data

Cited by in corpus (15)

• On the Compactness, Efficiency, and Representation of 3D Convolutional Networks: Brain Parcellation as a Pretext Task
• Revisiting Unreasonable Effectiveness of Data in Deep Learning Era
• Parameter-Efficient Transfer Learning for NLP
• The Devil is in the Tails: Fine-grained Classification in the Wild
• Analysis and Optimization of Convolutional Neural Network Architectures
• An Analysis of Pre-Training on Object Detection
• Few-Shot Image Recognition by Predicting Parameters from Activations
• Human perception in computer vision
• Material Classification using Neural Networks
• An Investigation of Transfer Learning-Based Sentiment Analysis in Japanese
• Sequential modeling of Sessions using Recurrent Neural Networks for Skip Prediction
• Ship classification from overhead imagery using synthetic data and domain adaptation
• Growing a Brain: Fine-Tuning by Increasing Model Capacity
• Pay Attention to Convolution Filters: Towards Fast and Accurate Fine-Grained Transfer Learning
• Neural Program Meta-Induction