Diverse Deep Feature Ensemble Learning for Omni-Domain Generalized Person Re-identification
This addresses the domain generalization challenge in person ReID, which is crucial for real-world applications where data comes from varied sources, though it appears incremental as it builds on existing ensemble and normalization techniques.
The paper tackles the problem of person re-identification (ReID) performance dropping when models are trained and tested across different datasets, proposing an omni-domain generalization approach that aims to be effective regardless of the number of domains involved. The result is that their method, D2FEL, significantly improves and matches state-of-the-art performance for both domain generalization and single-domain supervised benchmarks.
Person Re-identification (Person ReID) has progressed to a level where single-domain supervised Person ReID performance has saturated. However, such methods experience a significant drop in performance when trained and tested across different datasets, motivating the development of domain generalization techniques. However, our research reveals that domain generalization methods significantly underperform single-domain supervised methods on single dataset benchmarks. An ideal Person ReID method should be effective regardless of the number of domains involved, and when test domain data is available for training it should perform as well as state-of-the-art (SOTA) fully supervised methods. This is a paradigm that we call Omni-Domain Generalization Person ReID (ODG-ReID). We propose a way to achieve ODG-ReID by creating deep feature diversity with self-ensembles. Our method, Diverse Deep Feature Ensemble Learning (D2FEL), deploys unique instance normalization patterns that generate multiple diverse views and recombines these views into a compact encoding. To the best of our knowledge, our work is one of few to consider omni-domain generalization in Person ReID, and we advance the study of applying feature ensembles in Person ReID. D2FEL significantly improves and matches the SOTA performance for major domain generalization and single-domain supervised benchmarks.