Good Practice in CNN Feature Transfer

TL;DR

This paper investigates effective CNN feature transfer techniques for image search and classification, highlighting the importance of input image size, layer pooling strategies, and combining features across layers to improve performance.

Contribution

It systematically studies CNN transfer practices, demonstrating the benefits of larger input images, layer pooling, and multi-layer feature combination for enhanced accuracy.

Findings

Using larger input images improves CNN transfer performance.

Pooling across CNN layers yields competitive accuracy, especially Conv5.

Combining features from multiple layers enhances evidence collection and accuracy.

Abstract

The objective of this paper is the effective transfer of the Convolutional Neural Network (CNN) feature in image search and classification. Systematically, we study three facts in CNN transfer. 1) We demonstrate the advantage of using images with a properly large size as input to CNN instead of the conventionally resized one. 2) We benchmark the performance of different CNN layers improved by average/max pooling on the feature maps. Our observation suggests that the Conv5 feature yields very competitive accuracy under such pooling step. 3) We find that the simple combination of pooled features extracted across various CNN layers is effective in collecting evidences from both low and high level descriptors. Following these good practices, we are capable of improving the state of the art on a number of benchmarks to a large margin.