PLOT: Text-based Person Search with Part Slot Attention for Corresponding Part Discovery

TL;DR

This paper introduces PLOT, a novel framework for text-based person search that uses slot attention for autonomous part discovery and alignment, significantly improving retrieval accuracy without explicit part supervision.

Contribution

The paper presents a new part discovery module based on slot attention and a dynamic part attention mechanism, enhancing interpretability and performance in text-based person search.

Findings

Outperforms existing methods on three public benchmarks.

Effectively discovers and aligns human parts without explicit supervision.

Improves retrieval accuracy through dynamic part attention.

Abstract

Text-based person search, employing free-form text queries to identify individuals within a vast image collection, presents a unique challenge in aligning visual and textual representations, particularly at the human part level. Existing methods often struggle with part feature extraction and alignment due to the lack of direct part-level supervision and reliance on heuristic features. We propose a novel framework that leverages a part discovery module based on slot attention to autonomously identify and align distinctive parts across modalities, enhancing interpretability and retrieval accuracy without explicit part-level correspondence supervision. Additionally, text-based dynamic part attention adjusts the importance of each part, further improving retrieval outcomes. Our method is evaluated on three public benchmarks, significantly outperforming existing methods.