Towards the Influence of Text Quantity on Writer Retrieval

2025-06-09

AI Summary

This study systematically investigates the impact of textual quantity on writer retrieval performance, focusing on line-level and word-level fine-grained scenarios. We propose a multi-granularity retrieval framework integrating handcrafted features (VLAD) and deep features (NetVLAD), evaluated across page-, line-, and word-level settings on the CVL and IAM datasets. Our work quantitatively establishes, for the first time, a nonlinear relationship between text length and retrieval accuracy: accuracy drops by 20–30% when only one line is available, while four or more lines restore over 90% of full-page retrieval performance. Key findings include: (i) text-dependent methods retain robustness in low-resource settings; (ii) handcrafted features suffer significant degradation with limited text; and (iii) NetVLAD substantially outperforms VLAD. These results provide both theoretical foundations and practical benchmarks for few-shot handwriting identification.

Technology Category

• Computer Vision: Image and Video Retrieval
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Learning with Manifolds

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

This paper investigates the task of writer retrieval, which identifies documents authored by the same individual within a dataset based on handwriting similarities. While existing datasets and methodologies primarily focus on page level retrieval, we explore the impact of text quantity on writer retrieval performance by evaluating line- and word level retrieval. We examine three state-of-the-art writer retrieval systems, including both handcrafted and deep learning-based approaches, and analyze their performance using varying amounts of text. Our experiments on the CVL and IAM dataset demonstrate that while performance decreases by 20-30% when only one line of text is used as query and gallery, retrieval accuracy remains above 90% of full-page performance when at least four lines are included. We further show that text-dependent retrieval can maintain strong performance in low-text scenarios. Our findings also highlight the limitations of handcrafted features in low-text scenarios, with deep learning-based methods like NetVLAD outperforming traditional VLAD encoding.

Problem

• Investigates text quantity impact on writer retrieval performance
• Compares line- and word-level retrieval with page-level methods
• Evaluates deep learning vs handcrafted features in low-text scenarios

Innovation

• Evaluates line- and word-level writer retrieval
• Analyzes deep learning vs handcrafted methods
• Demonstrates text-dependent retrieval effectiveness