Adapting Convolutional Neural Networks for Geographical Domain Shift

TL;DR

This paper presents a simple ensemble approach of CNNs with last-layer fine-tuning to address geographical domain shift in image classification, demonstrating effectiveness in a NeurIPS competition setting.

Contribution

It introduces a practical method for adapting CNNs to geographical domain shifts using limited labeled data and ensemble techniques.

Findings

Ensemble of CNNs with last-layer fine-tuning improves domain adaptation.

Method performs well in a competitive NeurIPS challenge setting.

Approach is simple, effective, and practical for real-world applications.

Abstract

We present the winning solution for the Inclusive Images Competition organized as part of the Conference on Neural Information Processing Systems (NeurIPS 2018) Competition Track. The competition was organized to study ways to cope with domain shift in image processing, specifically geographical shift: the training and two test sets in the competition had different geographical distributions. Our solution has proven to be relatively straightforward and simple: it is an ensemble of several CNNs where only the last layer is fine-tuned with the help of a small labeled set of tuning labels made available by the organizers. We believe that while domain shift remains a formidable problem, our approach opens up new possibilities for alleviating this problem in practice, where small labeled datasets from the target domain are usually either available or can be obtained and labeled cheaply.