EPISODE · Mar 26, 2026 · 24 MIN
MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding
from Daily Paper Cast · host Jingwen Liang, Gengyu Wang
🤗 Upvotes: 112 | cs.CV Authors: Hejun Dong, Junbo Niu, Bin Wang, Weijun Zeng, Wentao Zhang, Conghui He Title: MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding Arxiv: http://arxiv.org/abs/2603.22458v1 Abstract: Optical character recognition (OCR) has evolved from line-level transcription to structured document parsing, requiring models to recover long-form sequences containing layout, tables, and formulas. Despite recent advances in vision-language models, most existing systems rely on autoregressive decoding, which introduces sequential latency and amplifies error propagation in long documents. In this work, we revisit document OCR from an inverse rendering perspective, arguing that left-to-right causal generation is an artifact of serialization rather than an intrinsic property of the task. Motivated by this insight, we propose MinerU-Diffusion, a unified diffusion-based framework that replaces autoregressive sequential decoding with parallel diffusion denoising under visual conditioning. MinerU-Diffusion employs a block-wise diffusion decoder and an uncertainty-driven curriculum learning strategy to enable stable training and efficient long-sequence inference. Extensive experiments demonstrate that MinerU-Diffusion consistently improves robustness while achieving up to 3.2x faster decoding compared to autoregressive baselines. Evaluations on the proposed Semantic Shuffle benchmark further confirm its reduced dependence on linguistic priors and stronger visual OCR capability.
What this episode covers
🤗 Upvotes: 112 | cs.CV Authors: Hejun Dong, Junbo Niu, Bin Wang, Weijun Zeng, Wentao Zhang, Conghui He Title: MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding Arxiv: http://arxiv.org/abs/2603.22458v1 Abstract: Optical character recognition (OCR) has evolved from line-level transcription to structured document parsing, requiring models to recover long-form sequences containing layout, tables, and formulas. Despite recent advances in vision-language models, most existing systems rely on autoregressive decoding, which introduces sequential latency and amplifies error propagation in long documents. In this work, we revisit document OCR from an inverse rendering perspective, arguing that left-to-right causal generation is an artifact of serialization rather than an intrinsic property of the task. Motivated by this insight, we propose MinerU-Diffusion, a unified diffusion-based framework that replaces autoregressive sequential decoding with parallel diffusion denoising under visual conditioning. MinerU-Diffusion employs a block-wise diffusion decoder and an uncertainty-driven curriculum learning strategy to enable stable training and efficient long-sequence inference. Extensive experiments demonstrate that MinerU-Diffusion consistently improves robustness while achieving up to 3.2x faster decoding compared to autoregressive baselines. Evaluations on the proposed Semantic Shuffle benchmark further confirm its reduced dependence on linguistic priors and stronger visual OCR capability.
NOW PLAYING
MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding
No transcript for this episode yet
Similar Episodes
No similar episodes found.