Multi-Task Learning With Coarse Priors for Robust Part-Aware Person Re-Identification.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Machine Learning

Background:

• Part-level representations are crucial for robust person re-identification (ReID).
• Existing methods struggle with feature quality due to body part misalignment in pedestrian images.

Purpose of the Study:

• To present a novel, robust, compact, and user-friendly method called the Multi-task Part-aware Network (MPN).
• To extract semantically aligned part-level features for improved person re-identification.

Main Methods:

• MPN employs multi-task learning (MTL) with one main task (MT) and one auxiliary task (AT) per body part.
• Auxiliary tasks use coarse location priors to guide main tasks in identifying part-relevant features via parameter and feature space alignment.
• The network learns high-quality parameters for independent feature extraction during inference without explicit part detection.

Main Results:

• MPN effectively solves the body part misalignment problem in person re-identification.
• The method achieves state-of-the-art performance on four large-scale ReID datasets.
• MPN demonstrates significant performance improvements over existing approaches.

Conclusions:

• MPN offers a compact and efficient solution for person re-identification by learning semantically aligned part-level features.
• The proposed method eliminates the need for body part detection during inference.
• MPN requires only coarse body part location priors for training, simplifying the process.