👋 Open to Work

cuicheng

ChengCui

·

AI & ML interests

VLM; OCR; AI Data Infra

Recent Activity

new activity 2 days ago

PaddlePaddle/PaddleOCR-VL-1.5:Add MDPBench private evaluation result

new activity 2 days ago

tencent/HunyuanOCR:Some Questions About Benchmark

commentedon a paper 2 days ago

HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Better

View all activity

Organizations

New activity in PaddlePaddle/PaddleOCR-VL-1.5 2 days ago

Add MDPBench private evaluation result

#32 opened 9 days ago by

New activity in tencent/HunyuanOCR 2 days ago

Some Questions About Benchmark

#33 opened 2 days ago by

commented a paper 2 days ago

HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Better

Paper • 2607.04884 • Published 6 days ago • 7 •

authored 2 papers 3 days ago

PP-OCRv5: A Specialized 5M-Parameter Model Rivaling Billion-Parameter Vision-Language Models on OCR Tasks

Paper • 2603.24373 • Published Mar 25

RT-DocLayout: Real-Time End-to-End Document Layout Analysis with Reading Order in the Wild

Paper • 2606.23344 • Published 20 days ago • 1

liked a dataset 3 days ago

nanxidajun/NuosuBburma-OCR-Evaluation-Set

Viewer • Updated 12 days ago • 758 • 1.12k • 1

liked a model 3 days ago

tencent/Hy3

Text Generation • 299B • Updated 6 days ago • 8.21k • 702

New activity in baidu/Unlimited-OCR 3 days ago

Arabic language

#8 opened 14 days ago by

updated a collection 4 days ago

PaddleOCR-VL-1.6

Expanding the Frontier of Document Parsing with Under-Optimized Region Refinement and Progressive Post-Training • 7 items • Updated 4 days ago • 14

upvoted a paper 4 days ago

RT-DocLayout: Real-Time End-to-End Document Layout Analysis with Reading Order in the Wild

Paper • 2606.23344 • Published 20 days ago • 1

updated 2 collections 4 days ago

PaddleOCR-VL-1.6

Expanding the Frontier of Document Parsing with Under-Optimized Region Refinement and Progressive Post-Training • 7 items • Updated 4 days ago • 14

PaddleOCR-VL-1.5

Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing • 8 items • Updated 4 days ago • 19

updated 2 models 4 days ago

PaddlePaddle/PP-DocLayoutV3

Image Segmentation • Updated 4 days ago • 49.2k • 95

PaddlePaddle/PP-DocLayoutV3_safetensors

Object Detection • 33.3M • Updated 4 days ago • 545k • 31

upvoted a paper 4 days ago

PP-DocLayout: A Unified Document Layout Detection Model to Accelerate Large-Scale Data Construction

Paper • 2503.17213 • Published Mar 21, 2025 • 2

New activity in PaddlePaddle/PP-OCRv6_tiny_rec_onnx 4 days ago

Some people steal from you

#1 opened 23 days ago by

New activity in PaddlePaddle/PP-OCRv6_medium_rec 4 days ago

Demo for this model on Spaces

#1 opened 8 days ago by

New activity in PaddlePaddle/PP-OCRv6_medium_det 4 days ago

Inference speed

#1 opened 24 days ago by

New activity in PaddlePaddle/PP-OCRv6_medium_det_safetensors 4 days ago

Better polygon accuracy on handwritten/angled text with engine='transformers' — can postprocessing be decoupled from the transformers library?

#1 opened 14 days ago by

commented on PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters 9 days ago

The tiny version already runs at 97ms in-browser. For even lower latency, try some engineering trade-offs on top of it—like optimizing the pre and post-processing steps.