Boosting Document Parsing Efficiency and Performance with Coarse-to-Fine Visual Processing
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908934192758784 |
|---|---|
| author | Cui, Cheng Sun, Ting Liang, Suyin Gao, Tingquan Zhang, Zelun Liu, Jiaxuan Wang, Xueqing Zhou, Changda Liu, Hongen Lin, Manhui Zhang, Yue Zhang, Yubo Zhang, Jing Zhang, Jun Wei, Xing Liu, Yi Yu, Dianhai Ma, Yanjun |
| author_facet | Cui, Cheng Sun, Ting Liang, Suyin Gao, Tingquan Zhang, Zelun Liu, Jiaxuan Wang, Xueqing Zhou, Changda Liu, Hongen Lin, Manhui Zhang, Yue Zhang, Yubo Zhang, Jing Zhang, Jun Wei, Xing Liu, Yi Yu, Dianhai Ma, Yanjun |
| contents | Document parsing is a fine-grained task where image resolution significantly impacts performance. While advanced research leveraging vision-language models benefits from high-resolution input to boost model performance, this often leads to a quadratic increase in the number of vision tokens and significantly raises computational costs. We attribute this inefficiency to substantial visual regions redundancy in document images, like background. To tackle this, we propose PaddleOCR-VL, a novel coarse-to-fine architecture that focuses on semantically relevant regions while suppressing redundant ones, thereby improving both efficiency and performance. Specifically, we introduce a lightweight Valid Region Focus Module (VRFM) which leverages localization and contextual relationship prediction capabilities to identify valid vision tokens. Subsequently, we design and train a compact yet powerful 0.9B vision-language model (PaddleOCR-VL-0.9B) to perform detailed recognition, guided by VRFM outputs to avoid direct processing of the entire large image. Extensive experiments demonstrate that PaddleOCR-VL achieves state-of-the-art performance in both page-level parsing and element-level recognition. It significantly outperforms existing solutions, exhibits strong competitiveness against top-tier VLMs, and delivers fast inference while utilizing substantially fewer vision tokens and parameters, highlighting the effectiveness of targeted coarse-to-fine parsing for accurate and efficient document understanding. The source code and models are publicly available at https://github.com/PaddlePaddle/PaddleOCR. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_24326 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Boosting Document Parsing Efficiency and Performance with Coarse-to-Fine Visual Processing Cui, Cheng Sun, Ting Liang, Suyin Gao, Tingquan Zhang, Zelun Liu, Jiaxuan Wang, Xueqing Zhou, Changda Liu, Hongen Lin, Manhui Zhang, Yue Zhang, Yubo Zhang, Jing Zhang, Jun Wei, Xing Liu, Yi Yu, Dianhai Ma, Yanjun Computer Vision and Pattern Recognition Artificial Intelligence Information Retrieval Document parsing is a fine-grained task where image resolution significantly impacts performance. While advanced research leveraging vision-language models benefits from high-resolution input to boost model performance, this often leads to a quadratic increase in the number of vision tokens and significantly raises computational costs. We attribute this inefficiency to substantial visual regions redundancy in document images, like background. To tackle this, we propose PaddleOCR-VL, a novel coarse-to-fine architecture that focuses on semantically relevant regions while suppressing redundant ones, thereby improving both efficiency and performance. Specifically, we introduce a lightweight Valid Region Focus Module (VRFM) which leverages localization and contextual relationship prediction capabilities to identify valid vision tokens. Subsequently, we design and train a compact yet powerful 0.9B vision-language model (PaddleOCR-VL-0.9B) to perform detailed recognition, guided by VRFM outputs to avoid direct processing of the entire large image. Extensive experiments demonstrate that PaddleOCR-VL achieves state-of-the-art performance in both page-level parsing and element-level recognition. It significantly outperforms existing solutions, exhibits strong competitiveness against top-tier VLMs, and delivers fast inference while utilizing substantially fewer vision tokens and parameters, highlighting the effectiveness of targeted coarse-to-fine parsing for accurate and efficient document understanding. The source code and models are publicly available at https://github.com/PaddlePaddle/PaddleOCR. |
| title | Boosting Document Parsing Efficiency and Performance with Coarse-to-Fine Visual Processing |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Information Retrieval |
| url | https://arxiv.org/abs/2603.24326 |