Publications by authors named "Haoxuanye Ji"
Meta Pairwise Relationship Distillation for Unsupervised Person Re-Identification.
IEEE Trans Neural Netw Learn Syst
September 2024
Article Synopsis
- - Unsupervised person re-identification (Re-ID) faces challenges due to the absence of labeled data, often relying on inaccurate cluster estimates for pseudo labels, which can hinder performance.
- - The proposed method, called meta pairwise relationship distillation (MPRD), leverages graph convolutional networks (GCN) to create reliable pairwise relationships that help improve feature learning without needing to define cluster numbers.
- - Additionally, the method introduces two components: a hard sample deduction (HSD) module to identify problematic pseudo labels and a positive pair alignment (PPA) module to reduce redundancy in feature information, resulting in better performance on various datasets compared to existing unsupervised approaches.
Disentangled Sample Guidance Learning for Unsupervised Person Re-Identification.
IEEE Trans Image Process
September 2024
Unsupervised person re-identification (Re-ID) is challenging due to the lack of ground truth labels. Most existing methods employ iterative clustering to generate pseudo labels for unlabeled training data to guide the learning process. However, how to select samples that are both associated with high-confidence pseudo labels and hard (discriminative) enough remains a critical problem.
View Article and Find Full Text PDF