홈 › Papers

Visual Merit or Linguistic Crutch? A Close Look at DeepSeek-OCR

2026-01-07 · Yunhao Liang, Ruixuan Ying, Bo Li, Hong Li, Kai Yan, Qingwen Li, Min Yang, Okamoto Satoshi, Zhe Cui, Shiwen Ni arxiv

DeepSeek-OCR utilizes an optical 2D mapping approach to achieve high-ratio vision-text compression, claiming to decode text tokens exceeding ten times the input visual tokens. While this suggests a promising solution for the LLM long-context bottleneck, we investigate a critical question: "Visual merit or linguistic crutch - which drives DeepSeek-OCR's performance?" By employing sentence-level and word-level semantic corruption, we isolate the model's intrinsic OCR capabilities from its language priors. Results demonstrate that without linguistic support, DeepSeek-OCR's performance plummets from approximately 90% to 20%. Comparative benchmarking against 13 baseline models reveals that traditional pipeline OCR methods exhibit significantly higher robustness to such semantic perturbations than end-to-end methods. Furthermore, we find that lower visual token counts correlate with increased reliance on priors, exacerbating hallucination risks. Context stress testing also reveals a total model collapse around 10,000 text tokens, suggesting that current optical compression techniques may paradoxically aggravate the long-context bottleneck. This study empirically defines DeepSeek-OCR's capability boundaries and offers essential insights for future optimizations of the vision-text compression paradigm. We release all data, results and scripts used in this study at https://github.com/dududuck00/DeepSeekOCR.

📄 PDF Abstract BibTeX arXiv:2601.03714

Code (0)

등록된 구현이 없습니다.

Tensegrity crutches with compliance from a pre-stressed self-tensile module improve ground reaction force profiles, speed, effort, comfort, and perceived stability

2026-05-04 · Jingxian Gu, Joanna Spyra, Andrew Walski, Lyla Elsaesser 외 arxiv

Purpose: Six million people use crutches as mobile aids in the US. Rigid designs with no axial mobility limit sensory feedback and lead to secondary injury on the upper joints. Spring-loaded designs offer compliance but …

VL-LTR: Learning Class-wise Visual-Linguistic Representation for Long-Tailed Visual Recognition

2021-11-26 · Changyao Tian, Wenhai Wang, Xizhou Zhu, Jifeng Dai 외

Deep learning-based models encounter challenges when processing long-tailed data in the real world. Existing solutions usually employ some balancing strategies or transfer learning to deal with the class imbalance proble…

Image Classification Long-tail Learning Transfer Learning

A Reinforcement Learning Based Controller to Minimize Forces on the Crutches of a Lower-Limb Exoskeleton

2024-01-31 · Aydin Emre Utku, Suzan Ece Ada, Muhammet Hatipoglu, Mustafa Derman 외

Metabolic energy consumption of a powered lower-limb exoskeleton user mainly comes from the upper body effort since the lower body is considered to be passive. However, the upper body effort of the users is largely ignor…

Deep Reinforcement Learning MuJoCo reinforcement-learning Reinforcement Learning (RL)

Densely Decoded Networks with Adaptive Deep Supervision for Medical Image Segmentation

2024-02-05 · Suraj Mishra, Danny Z. Chen

Medical image segmentation using deep neural networks has been highly successful. However, the effectiveness of these networks is often limited by inadequate dense prediction and inability to extract robust features. To …

Decoder Image Segmentation Medical Image Segmentation Semantic Segmentation

IMU-based Real-Time Crutch Gait Phase and Step Detections in Lower-Limb Exoskeletons

2026-01-15 · Anis R. Shakkour, David Hexner, Yehuda Bitton, Avishai Sintov arxiv

Lower limb exoskeletons and prostheses require precise, real time gait phase and step detections to ensure synchronized motion and user safety. Conventional methods often rely on complex force sensing hardware that intro…