Qiyu Hou (侯启予)

Researcher, focused on ML, CV, Language Models, and Document Intelligence. Cuber.

Hi there, I’m Qiyu Hou (侯启予).

  • Welcome to my homepage. Here’s my blog page and cubing page.
  • I focus on ML, CV, Language Models (VLMs, LLMs, and MLLMs), and Document Intelligence (Document Layout Analysis, OCR, Table Structure Recognition, and Document Understanding).
  • I’m working on document processing pipelines, and agentic systems (RAG, Deep Search, and Deep Research).
  • A long time ago, I was really into cubing. In 2016, I built a robot that could solve a Rubik’s Cube in one second. That same year, I ranked among the world’s top 10 in the Rubik’s Clock event, with a highest ranking of NR-2 and AsR-3.
  • I’m currently based in Nanjing, Jiangsu, China.
Publications

TABLET: Table Structure Recognition using Encoder-only Transformers

Qiyu Hou, Jun Wang()

ICDAR-2025, arXiv:2506.07015

Synthesizing Realistic Data for Table Recognition

Qiyu Hou, Jun Wang(), Meixuan Qiao, Lujun Tian

ICDAR-2024, arXiv:2404.11100, Springer:10.1007/978-3-031-70533-5_22

Structure Diagram Recognition in Financial Announcements

Meixuan Qiao, Jun Wang(), Junfu Xiang, Qiyu Hou, Ruixuan Li

ICDAR-2023, arXiv:2304.13240, Springer:10.1007/978-3-031-41676-7_2

Automatically Constructing Knowledge Graphs from Ownership Structure Diagrams in Financial Announcements

Meixuan Qiao, Jun Wang(), Junfu Xiang, Qiyu Hou, Ruixuan Li

CCKS-2022

Reports

文档智能在金融领域的应用

DataFunCon-2024 @Shanghai, DataFunTalk:金融领域文档智能应用实践

Experiences

iWudao Tech: Pre-Research Center

Research Manager, 2020 to present

Fujitsu Nanjing: Advanced Technology Center

Researcher, 2017 to 2020