CVApr 2, 2020

ProxyNCA++: Revisiting and Revitalizing Proxy Neighborhood Component Analysis

arXiv:2004.01113v2191 citations
AI Analysis

This work improves image retrieval performance for applications like visual search, though it is incremental as it builds on an existing method.

The paper tackles the problem of distance metric learning for image similarity by enhancing ProxyNCA with low temperature scaling, Global Max Pooling, and fast moving proxies, achieving a 22.9 percentage point average improvement in Recall@1 across four zero-shot retrieval datasets and state-of-the-art results on multiple benchmarks.

We consider the problem of distance metric learning (DML), where the task is to learn an effective similarity measure between images. We revisit ProxyNCA and incorporate several enhancements. We find that low temperature scaling is a performance-critical component and explain why it works. Besides, we also discover that Global Max Pooling works better in general when compared to Global Average Pooling. Additionally, our proposed fast moving proxies also addresses small gradient issue of proxies, and this component synergizes well with low temperature scaling and Global Max Pooling. Our enhanced model, called ProxyNCA++, achieves a 22.9 percentage point average improvement of Recall@1 across four different zero-shot retrieval datasets compared to the original ProxyNCA algorithm. Furthermore, we achieve state-of-the-art results on the CUB200, Cars196, Sop, and InShop datasets, achieving Recall@1 scores of 72.2, 90.1, 81.4, and 90.9, respectively.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes