MixBCT: Towards Self-Adapting Backward-Compatible Training

TL;DR

MixBCT introduces a unified backward-compatible training framework that adaptively aligns new features with old ones, improving retrieval system updates without costly database re-annotations.

Contribution

It proposes a novel loss function that adaptively constrains new features based on old feature distributions, enhancing backward compatibility across models of varying quality.

Findings

Outperforms previous backward-compatible training methods.

Effective on large-scale face recognition datasets.

Demonstrates superior retrieval accuracy and compatibility.

Abstract

Backward-compatible training circumvents the need for expensive updates to the old gallery database when deploying an advanced new model in the retrieval system. Previous methods achieved backward compatibility by aligning prototypes of the new model with the old one, yet they often overlooked the distribution of old features, limiting their effectiveness when the low quality of the old model results in a weakly feature discriminability. Instance-based methods like L2 regression take into account the distribution of old features but impose strong constraints on the performance of the new model itself. In this paper, we propose MixBCT, a simple yet highly effective backward-compatible training method that serves as a unified framework for old models of varying qualities. We construct a single loss function applied to mixed old and new features to facilitate backward-compatible training, which adaptively adjusts the constraint domain for new features based on the distribution of old features. We conducted extensive experiments on the large-scale face recognition datasets MS1Mv3 and IJB-C to verify the effectiveness of our method. The experimental results clearly demonstrate its superiority over previous methods. Code is available at https://github.com/yuleung/MixBCT .