Minguk Kang

Ph.D. student in the Computer Vision Lab at POSTECH.

prof_minguk.jpg

I am a 5th year Ph.D. student in Graduate School of AI at POSTECH, where I am a member of the Computer Vision Lab under the supervision of Prof. Suha Kwak. From 2022 to 2024, I interned at Adobe Research, collaborating with wonderful colleagues: Taesung Park, Jun-Yan Zhu, Richard Zhang, Eli Shechtman, Sylvain Paris, and Connelly Barnes, where my research contributed to the development of Firefly. Previously, I received my B.S. in Engineering from Pusan National University.

My research focuses on computer vision, particularly visual generative modeling and its applications. I am fascinated by the diverse range of vision-related tasks that Generative Adversarial Networks (GANs) can assist with. I am also interested in training large-scale diffusion models and exploring their applications. If you would like to know more about my research projects, please feel free to reach out to me.

Email: mingukkang1994@gmail.com, mgkang@postech.ac.kr

Education

Feb, 2020 - Present Pohang University of Science and Technology (POSTECH), Pohang, South Korea
Integrated M.S./Ph.D. student in Graduate School of AI
Mar, 2013 - Aug, 2019 Pusan National University, Pusan, South Korea
B.S. in Engineering (Major: Mechanical Engineering and Minor: Statistics)

Experience

Jun, 2024 - Sep, 2024 Pika, Palo Alto, United States of America
Jul, 2022 - May, 2024 Adobe Research Creative Intelligence Lab, Remote work at Korea & Onsite work at San Francisco
Feb, 2020 - Present Computer Vision Lab, POSTECH, Pohang, South Korea
  • Integrated M.S./Ph.D. student
  • Supervisor: Prof. Suha Kwak
Aug, 2017 - Jan, 2020 Vision and Intelligent System Lab, Pusan National University, Pusan, South Korea

Softwares

  • Firefly is a text-to-image generative model developed by Adobe and integrated into Photoshop. My research contributed to the development of Firefly.
  • StudioGAN is a Pytorch library providing implementations of representative Generative Adversarial Networks for image generation. I am the primary inventor of the StudioGAN library.

Publications

  1. Distilling Diffusion Models into Conditional GANs
    Minguk Kang, Richard Zhang, Connelly Barnes, Sylvain Paris, Suha Kwak, Jaesik Park, Eli Shechtman, Jun-Yan Zhu,  and Taesung Park
    European Conference on Computer Vision (ECCV), 2024
    Paper Code Website
  2. Extending CLIP’s Image-Text Alignment to Referring Image Segmentation
    Seoyeon Kim, Minguk Kang, Dongwon Kim, Jaesik Park,  and Suha Kwak
    Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL), 2024
    Paper Code
  3. Fill-Up: Balancing Long-Tailed Data with Generative Models
    Joonghyuk Shin, Minguk Kang,  and Jaesik Park
    arXiv preprint arXiv:2306.07200 2023
    Paper Website
  4. StudioGAN: A Taxonomy and Benchmark of GANs for Image Synthesis
    Minguk Kang, Joonghyuk Shin,  and Jaesik Park
    IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) 2023
    Paper Code
  5. Holistic Evaluation of Text-to-Image Models
    Tony Lee, Michihiro Yasunaga, Chenlin Meng, Yifan Mai, Joon Sung Park, Agrim Gupta, Yunzhi Zhang, Deepak Narayanan, Hannah Benita Teufel, Marco Bellagente, Minguk Kang, Taesung Park, Jure Leskovec, Jun-Yan Zhu, Li Fei-Fei, Jiajun Wu, Stefano Ermon,  and Percy Liang
    Advances in Neural Information Processing Systems (NeurIPS), Datasets and Benchmarks Track, Spotlight 2023
    Paper Website
  6. Scaling up GANs for Text-to-Image Synthesis
    Minguk Kang, Jun-Yan Zhu, Richard Zhang, Jaesik Park, Eli Shechtman, Sylvain Paris,  and Taesung Park
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Highlight, 2023
    Paper Website
  7. Context-Aware Image Completion
    Jinoh Cho, Minguk Kang, Vibhav Vineet,  and Jaesik Park
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshop, 2023
    Paper
  8. Rebooting ACGAN: Auxiliary Classifier GANs with Stable Training
    Minguk Kang, Woohyeon Shim, Minsu Cho,  and Jaesik Park
    Advances in Neural Information Processing Systems (NeurIPS), 2021
    Paper Code
  9. ContraGAN: Contrastive Learning for Conditional Image Generation
    Minguk Kang,  and Jaesik Park
    Advances in Neural Information Processing Systems (NeurIPS), 2020
    Paper Code

Honors and Awards

Graduate School Presidential Science Scholarship (2024)
  • Received a scholarship totaling $26,000, which amounted to $1,450 per month over 18 months.
BK21 outstanding paper awards (2024)
  • 2st prize ($375) - Scaling up GANs for Text-to-Image Synthesis (CVPR2023)
BK21 outstanding paper awards (2022)
  • 1st prize ($750) - Rebooting ACGAN: Auxiliary Classifier GANs with Stable Training (NeurIPS2021)
Qualcomm Innovation Fellowship Korea (2021)
  • Winner ($3,000) - ContraGAN: Contrastive Learning for Conditional Image Generation (NeurIPS2020)
16th Samsung Electro­Mechanics Paper Awards (2020)
  • Silver prize ($2,500)
National Science and Engineering Scholarship (2013-2019)
  • Received full scholarship for 8 semesters ($15,000)