Document understanding and GUI interaction are among the highest-value applications of Vision-Language Models (VLMs), yet they impose exceptionally heavy computational burden: fine-grained text and small UI elements demand high-resolution inputs that produce tens of thousands of visual tokens. We observe that this cost is largely wasteful -- across document and GUI benchmarks, only 22--71\% of image patches are pixel-unique, the rest being exact duplicates of another patch in the same image. We propose \textbf{PixelPrune}, which exploits this pixel-level redundancy through predictive-coding-based compression, pruning redundant patches \emph{before} the Vision Transformer (ViT) encoder. Because it operates in pixel space prior to any neural computation, PixelPrune accelerates both the ViT encoder and the downstream LLM, covering the full inference pipeline. The method is training-free, requires no learnable parameters, and supports pixel-lossless compression ($τ{=}0$) as well as controlled lossy compression ($τ{>}0$). Experiments across three model scales and document and GUI benchmarks show that PixelPrune maintains competitive task accuracy while delivering up to 4.2$\times$ inference speedup and 1.9$\times$ training acceleration. Code is available at https://github.com/OPPO-Mente-Lab/PixelPrune.

翻译：文档理解与图形用户界面交互是视觉语言模型（Vision-Language Models, VLMs）最高价值的应用场景之一，但这些任务带来了异常沉重的计算负担：精细文本与小尺寸界面元素需要高分辨率输入，进而产生数万个视觉令牌。我们观察到，这种开销在很大程度上是浪费的——在文档与GUI基准测试中，仅22\%至71\%的图像块具有像素唯一性，其余均为同一图像中其他图像块的精确副本。为此，我们提出\textbf{PixelPrune}方法，通过基于预测编码的压缩机制利用这种像素级冗余，在视觉Transformer（Vision Transformer, ViT）编码器之前修剪冗余图像块。由于该方法在神经网络计算前的像素空间执行操作，PixelPrune能同时加速ViT编码器与下游大语言模型（LLM），覆盖完整推理流程。该方法无需训练，不依赖任何可学习参数，同时支持像素无损压缩（$τ{=}0$）与可控有损压缩（$τ{>}0$）。在三种不同规模模型及文档与GUI基准测试上的实验表明，PixelPrune在保持竞争力任务准确率的同时，可实现高达4.2倍的推理加速与1.9倍的训练加速。代码开源于https://github.com/OPPO-Mente-Lab/PixelPrune。