PeopleofSuperintelligence

Researcher

Hong-Yuan Mark Liao

Also known as 廖弘源, Mark Liao, Hong-Yuan Liao, Hong-Yuan M. Liao

Hong-Yuan Mark Liao is a computer scientist at Academia Sinica whose coauthored work on real-time object detection includes CSPNet and the YOLOv4, YOLOR, YOLOv7 and YOLOv9 detectors. After studying physics at National Tsing Hua University, he earned a doctorate in electrical engineering from Northwestern University in 1990 and joined Academia Sinica's Institute of Information Science in 1991. He became an IEEE Fellow in 2013 and, working with Chien-Yao Wang, coauthored CSPNet in 2019, YOLOv4 in 2020, YOLOR in 2021, YOLOv7 in 2022 and YOLOv9 in 2024. The pair received the 2026 Frontiers of Science Award. He is Distinguished Research Fellow and Director of the Institute of Information Science, a post he has held since 2018.

Badges

Timeline

  1. 1990

    Physics and a Northwestern PhD

    He studies physics at National Tsing Hua University, then completes a master's and doctoral degree in electrical engineering at Northwestern University in 1990.

  2. 1991–2004

    Joining Academia Sinica

    He joins the Institute of Information Science at Academia Sinica in July 1991. He serves as its deputy director from 1997 to 2000 and as acting director of the Institute of Applied Science and Engineering Research preparation office between 2001 and 2004.

  3. 1998–2016

    Research awards and IEEE Fellowship

    He receives the Academia Sinica Young Investigators' Award in 1998 and the National Science Council's Distinguished Research Award in 2003, 2010 and 2013. He is named an IEEE Fellow in 2013 for contributions to image and video forensics and security. The TECO Foundation gives him the TECO Award in 2016.

  4. 2019–2020

    CSPNet and YOLOv4

    In 2019, he coauthors CSPNet with Chien-Yao Wang and four others; the Cross Stage Partial Network cuts computation by 20% with equivalent or better accuracy on ImageNet. In 2020, he coauthors YOLOv4 with Alexey Bochkovskiy and Wang, which reaches 43.5% AP on MS COCO at about 65 FPS on a Tesla V100. His team develops both networks for a smart-city traffic-flow project that calculates traffic parameters in real time at crossroads to control traffic signs dynamically.

  5. 2021–2022

    YOLOR and YOLOv7

    In 2021, he coauthors YOLOR (You Only Learn One Representation), a unified network that encodes implicit and explicit knowledge to serve several tasks at once. In 2022, he coauthors YOLOv7 with Wang and Bochkovskiy, which reports 56.8% AP among real-time detectors running at 30 FPS or higher on a V100 GPU. From 2022, he and Wang lead a second four-year AI project funded by the National Science Council.

  6. 2024

    YOLOv9 with programmable gradient information

    In 2024, he coauthors YOLOv9 with Wang and I-Hau Yeh. The paper proposes programmable gradient information (PGI) to cope with information lost as data passes through deep networks, together with the lightweight GELAN architecture.

  7. 2026

    Frontiers of Science Award

    He and Wang receive the 2026 Frontiers of Science Award, announced by the Institute of Information Science.

  8. As of October 2026

    Institute director at Academia Sinica

    He serves as director of the Institute of Information Science from 2018, and Academia Sinica re-appoints him for a term from September 2024 to June 2027. As of October 2026, the institute lists him as Distinguished Research Fellow/Professor and Director.

Career

  1. Since 1991Academia SinicaDeputy Director, Acting Director

Education

  1. 1990Northwestern UniversityMaster's
  2. National Tsing Hua University

From the biography.

  1. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Academia Sinica. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  2. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Northwestern University. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Northwestern University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  4. Google has the highest combined h-index among AI organizationsIncludes Northwestern University. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Hong-Yuan Mark Liao yet. Send us a free-licensed photo.

  1. Mark Liao - Wikipedia en.wikipedia.org. Checked 21 October 2026.
  2. 廖弘源 - YOLOv4 誕生的奇妙過程 (14:40-15:00) - 2020 AI Academy 台灣人工智慧年會 conf2020.aiacademy.tw. Checked 21 October 2026.
  3. Seminar: Intelligence Traffic System in Smart City and YOLOv4 - Institute of Information Science, Academia Sinica homepage.iis.sinica.edu.tw. Checked 21 October 2026.
  4. CSPNet: A New Backbone that can Enhance Learning Capability of CNN arxiv.org. Checked 21 October 2026.
  5. YOLOv4: Optimal Speed and Accuracy of Object Detection arxiv.org. Checked 21 October 2026.
  6. 3M Maker: Multi-modal, Multi-task, Multiple Instance Learning for Trustable Multimedia Processing - Institute of Information Science, Academia Sinica homepage.iis.sinica.edu.tw. Checked 21 October 2026.
  7. You Only Learn One Representation: Unified Network for Multiple Tasks arxiv.org. Checked 21 October 2026.
  8. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors arxiv.org. Checked 21 October 2026.
  9. YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information arxiv.org. Checked 21 October 2026.
  10. 2026 announcements - Institute of Information Science, Academia Sinica homepage.iis.sinica.edu.tw. Checked 21 October 2026.
  11. Personnel - Academia Sinica sinica.edu.tw. Checked 21 October 2026.
  12. Research Fellows - Institute of Information Science, Academia Sinica iis.sinica.edu.tw. Checked 21 October 2026.