See Finer, See More: Implicit Modality Alignment for Text-based Person Retrieval
This work addresses the challenge of aligning visual and textual modalities for person retrieval, which is incremental as it builds on existing methods by avoiding explicit labeling and exploring finer-grained alignments.
The paper tackles the problem of text-based person retrieval by proposing an Implicit Visual-Textual (IVT) framework that learns a common latent space without explicit cross-modal alignments, achieving state-of-the-art performance on datasets like CUHK-PEDES, RSTPReID, and ICFG-PEDES.
Text-based person retrieval aims to find the query person based on a textual description. The key is to learn a common latent space mapping between visual-textual modalities. To achieve this goal, existing works employ segmentation to obtain explicitly cross-modal alignments or utilize attention to explore salient alignments. These methods have two shortcomings: 1) Labeling cross-modal alignments are time-consuming. 2) Attention methods can explore salient cross-modal alignments but may ignore some subtle and valuable pairs. To relieve these issues, we introduce an Implicit Visual-Textual (IVT) framework for text-based person retrieval. Different from previous models, IVT utilizes a single network to learn representation for both modalities, which contributes to the visual-textual interaction. To explore the fine-grained alignment, we further propose two implicit semantic alignment paradigms: multi-level alignment (MLA) and bidirectional mask modeling (BMM). The MLA module explores finer matching at sentence, phrase, and word levels, while the BMM module aims to mine \textbf{more} semantic alignments between visual and textual modalities. Extensive experiments are carried out to evaluate the proposed IVT on public datasets, i.e., CUHK-PEDES, RSTPReID, and ICFG-PEDES. Even without explicit body part alignment, our approach still achieves state-of-the-art performance. Code is available at: https://github.com/TencentYoutuResearch/PersonRetrieval-IVT.