VERSE: Visual Embedding Reduction and Space Exploration. Clustering-Guided Insights for Training Data Enhancement in Visually-Rich Document Understanding
Research paper by Ignacio de Rodrigo and collaborators. Published 2026-01-08. Topics: Vision-Language Models, visual embedding space, latent representations, synthetic data generation, MERIT Dataset.
Resource information
- Authors
- Ignacio de Rodrigo, Alvaro J. Lopez-Lopez, Jaime Boal
- Research area
- Computer vision
- Application domain
- Document processing
- AI methods
- Large language models
- Research topics
- cs.CV, cs.AI
- arXiv record date
- 2026-01-08
- Publication date
- 2026-01-08
- Publication
- arXiv
- Upstream last modified
- Jan 9, 2026
- Last record update
- Nov 3, 2026
Abstract
This work introduces VERSE, a methodology for analyzing and improving Vision-Language Models applied to Visually-rich Document Understanding by exploring their visual embedding space. VERSE enables the visualization of latent representations, supporting the assessment of model feasibility. It also facilitates the identification of problematic regions and guides the generation of synthetic data to enhance performance in those clusters. We validate the methodology by training on the synthetic MERIT Dataset and evaluating on its real-world counterpart, MERIT Secret. Results show that VERSE helps uncover the visual features associated with error-prone clusters, and that retraining with samples containing these features substantially boosts F1 performance without degrading generalization. Furthermore, we demonstrate that on-premise models such as Donut and Idefics2, when optimized with VERSE, match or even surpass the performance of SaaS solutions like GPT-4 and Pixtral.