One OCR model family handles recognition, parsing, and reasoning
PolyOCR uses a shared generation interface and competence-aware post-training to cover diverse text-centric vision tasks.
Chinese Tech
GuangJian Team · Kaili Huang · Yongshuo Zhang · Bingtao Fu · Changjiang Jiang · Chenfan Qu · +19 more
Ant Group
Research Digest··2 min read
The GuangJian Team presents PolyOCR, a family of multimodal models designed to perform OCR tasks that are usually handled by separate systems, including transcription, localization, document parsing, information extraction, and visual reasoning.
Why this paper
From Ant Group
In one line
PolyOCR achieves state-of-the-art performance across diverse OCR tasks by unifying instruction-following, a data engine, and competence-guided policy optimization.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (2 noted)
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§