Our Ph.D. student Yoonki Cho presented his work titled “Generalizable Person Re-identification via Balancing Alignment and Uniformity” at NeurIPS 2024 held from December 10th to December 15th, 2024. This study introduces the Balancing Alignment and Uniformity (BAU) framework to enhance domain generalization in person re-identification by maintaining a balance between alignment and uniformity, thereby improving robustness against distributional shifts. The proposed approach effectively leverages data augmentation to achieve state-of-the-art performance without requiring complex training procedures.
