Lightweight and Production-Ready PDF Visual Element Parsing
It addresses the practical need for accurate and efficient PDF visual element extraction in production environments, particularly for multimodal retrieval-augmented generation.
The paper presents a lightweight PDF parsing framework that achieves ≥96% visual element detection accuracy and 93% caption association accuracy, outperforming state-of-the-art parsers in multimodal RAG while reducing latency by over 2×.
PDF documents contain critical visual elements such as figures, tables, and forms whose accurate extraction is essential for document understanding and multimodal retrieval-augmented generation (RAG). Existing PDF parsers often miss complex visuals, extract non-informative artifacts (e.g., watermarks, logos), produce fragmented elements, and fail to reliably associate captions with their corresponding elements, which degrades downstream retrieval and question answering. We present a lightweight and production level PDF parsing framework that can accurately detect visual elements and associates captions using a combination of spatial heuristics, layout analysis, and semantic similarity. On popular benchmark datasets and internal product data, the proposed solution achieves $\geq96\%$ visual element detection accuracy and $93\%$ caption association accuracy. When used as a preprocessing step for multimodal RAG, it significantly outperforms state-of-the-art parsers and large vision-language models on both internal data and the MMDocRAG benchmark, while reducing latency by over $2\times$. We have deployed the proposed system in challenging production environment.