Qixin Hu

Qixin Hu 胡琪鑫

PhD Student @USC

  University of Southern California, Los Angeles, CA, USA

  qixinhu [at] usc [dot] edu

🧑‍🎓 About Me

Hi, my name is Qixin Hu (胡琪鑫). If you can't read Pinyin, you can pronounce it as "Chee-sheen Who." I am a second-year Ph.D. student at USC, working with Prof. C.-C. Jay Kuo. My research lies broadly in representation learning, generative AI, and foundation models. Specifically, I want to design simple and efficient learning algorithms and apply them to real-world applications.

I am very fortunate to have had opportunities to collaborate with Prof. Qi Dou, Prof. Alan Yuille, and Prof. Xinggang Wang. I am a recipient of the Viterbi Fellowship.

  I am open to collaboration — feel free to reach out if you're working on related topics.

🔥 Recent News

03/2026 Code for MemRoPE is released. We also made a demo website for it.
02/2026 One paper on training-free infinite video generation (MemRoPE) is submitted to ECCV 2026.
01/2026 One paper on eosinophil counting with green learning is accepted by ISBI 2026.
08/2025 I will serve as the Program Committee of AAAI 2026.
05/2025 I will serve as a Reviewer of NeurIPS 2025.
03/2025 I will serve as a Reviewer of APSIPA Transactions on Signal and Information Processing.
12/2024 One paper accepted by Generative Machine Learning Models in Medical Image Computing.
08/2024 Awarded Viterbi Fellowship.
06/2024 One paper on multi-organ trauma detection via VLM accepted by MICCAI 2024.
07/2023 I will serve as a Reviewer of ICML Workshop 2023.
06/2023 One paper on 3D animal datasets accepted by ICCV 2023.
04/2023 Code for Synthetic Tumors is released.
03/2023 One paper on label-free liver tumor segmentation accepted by CVPR 2023.

🚀 Open Source Contributions

Developed the inference engine for infinite video generation. Upgraded the existing architecture by designing an online relative RoPE to support infinite-length outputs.

Co-first Author & Primary Maintainer. Architected and implemented a novel evolving memory pipeline, enabling training-free stable infinite video generation.

Lead Author & Primary Maintainer. Pioneered successful demonstration of competitive liver tumor segmentation relying exclusively on synthetic training data.

Co-Author & Core Contributor. Engineered and integrated the vision-language fusion module, advancing the model's multi-organ trauma detection capabilities.

📝 Publications

MemRoPE

MemRoPE: Training-Free Infinite Video Generation via Evolving Memory Tokens

Submitted to ECCV 2026

Youngrae Kim*, Qixin Hu*, C.-C. Jay Kuo, Peter A. Beerel
*Equal Contribution

Eosinophil Counting with Green Learning

Eosinophil Counting with Green Learning

ISBI 2026

Qixin Hu, Yixing Wu, Mate Levente Nagyet, Hwan Dong, C.-C. Jay Kuo

Analyzing Tumors by Synthesis

Analyzing Tumors by Synthesis

Generative ML Models in Medical Image Computing, 2024

Yuxiang Lai, Qi Chen, ..., Qixin Hu, et al.

Language-Enhanced Trauma Detection

Language-Enhanced Local-Global Aggregation Network for Multi-Organ Trauma Detection

MICCAI 2024

Jianxun Yu, Qixin Hu, et al.

Synthetic Data as Validation

Synthetic Data as Validation

arXiv 2023

Qixin Hu, Alan L. Yuille, Zongwei Zhou

Animal3D

Animal3D: A Comprehensive Dataset of 3D Animal Pose and Shape

ICCV 2023

Jiacong Xu, ..., Qixin Hu, et al.

Label-free Liver Tumor Segmentation

Label-Free Liver Tumor Segmentation

CVPR 2023

Qixin Hu, Yixiong Chen, Junfei Xiao, Shuwen Sun, Jie-Neng Chen, Alan L. Yuille, Zongwei Zhou

Synthetic Tumors Make AI Segment Tumors Better

Synthetic Tumors Make AI Segment Tumors Better

NeurIPS Workshop 2022

Qixin Hu, Junfei Xiao, Yixiong Chen, Shuwen Sun, Jie-Neng Chen, Alan L. Yuille, Zongwei Zhou

Object Recognition with Speckles

Object Recognition for Remarkably Small Field-of-View with Speckles

APL 2021

Qixin Hu, Siyan Xu, Xue-wen Chen, Xinggang Wang, Ken Xingze Wang

💻 Research Experience

Foundation Models on Medical Image, CUHK

  • Position: Research Assistant (Sep 2023 – Jun 2024)
  • Advisor: Prof. Qi Dou
  • Achievement: MICCAI 2024, Rank top 0.5% in Foundation Model Benchmark

Label-efficient Learning Methods, Johns Hopkins University

Object Recognition with Small FoV, HUST

  • Position: Research Intern (Sep 2020 – Jun 2022)
  • Advisor: Prof. Xinggang Wang
  • Achievement: APL 2021

🏅 Honors & Awards

📙 Academic Service

PC Member · AAAI 2026 Reviewer · NeurIPS 2025 Reviewer · APSIPA Trans. on Signal & Info. Processing Reviewer · NeurIPS GenAI for Health Workshop 2024 Reviewer · ICML IMLH Workshop 2023