Revisiting Training Strategies and Generalization Performance in Deep Metric Learning
This work addresses inconsistent comparisons in DML, offering a simple regularization to boost performance, but it is incremental as it builds on existing methods.
The study revisited training strategies in Deep Metric Learning, finding that objectives saturate more than previously reported and linking embedding space density to generalization, then proposed a regularization method that improved performance on standard benchmarks.
Deep Metric Learning (DML) is arguably one of the most influential lines of research for learning visual similarities with many proposed approaches every year. Although the field benefits from the rapid progress, the divergence in training protocols, architectures, and parameter choices make an unbiased comparison difficult. To provide a consistent reference point, we revisit the most widely used DML objective functions and conduct a study of the crucial parameter choices as well as the commonly neglected mini-batch sampling process. Under consistent comparison, DML objectives show much higher saturation than indicated by literature. Further based on our analysis, we uncover a correlation between the embedding space density and compression to the generalization performance of DML models. Exploiting these insights, we propose a simple, yet effective, training regularization to reliably boost the performance of ranking-based DML models on various standard benchmark datasets. Code and a publicly accessible WandB-repo are available at https://github.com/Confusezius/Revisiting_Deep_Metric_Learning_PyTorch.