Hongjun Wang

Research Roadmap

Publications

Generalized Category Discovery under Domain Shifts thumbnail

Generalized Category Discovery under Domain Shifts: From Vision to Vision-Language Models
Hongjun Wang, Po Hu, Kai Han
ArXiv 2026
TL;NR: GCD under domain shifts, from vision foundation models to vision-language models

BibTeX Paper Project Page


@article{wang2026generalized,
  title   = {Generalized Category Discovery under Domain Shifts: From Vision to Vision-Language Models},
  author  = {Hongjun Wang and Po Hu and Kai Han},
  year    = {2026},
  journal = {arXiv preprint arXiv:2605.00906}
}

LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model
Inclusion AI, Tiwei Bie, Haoxing Chen, Tieyuan Chen, Zhenglin Cheng, Long Cui, Kai Gan, Zhicheng Huang, Zhenzhong Lan, Haoquan Li, Jianguo Li, Tao Lin, Qi Qin, Hongjun Wang^*, Xiaomei Wang, Haoyuan Wu, Yi Xin, Junbo Zhao
* one of the core contributors; individual authors are sorted by last name.
ArXiv 2026
TL;NR: Unified multimodal understanding, generation, editing, and interleaved data generation with a diffusion LLM

BibTeX Paper Code Model


@article{inclusionai2026llada20uni,
  title   = {LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model},
  author  = {{Inclusion AI} and Tiwei Bie and Haoxing Chen and Tieyuan Chen and Zhenglin Cheng and Long Cui and Kai Gan and Zhicheng Huang and Zhenzhong Lan and Haoquan Li and Jianguo Li and Tao Lin and Qi Qin and Hongjun Wang and Xiaomei Wang and Haoyuan Wu and Yi Xin and Junbo Zhao},
  year    = {2026},
  journal = {arXiv preprint arXiv:2604.20796}
}

MHPO: Modulated Hazard-aware Policy Optimization for Stable Reinforcement Learning
Hongjun Wang, Wei Liu, Weibo Gu, Xing Sun, Kai Han
ArXiv 2026
TL;NR: Stable RL algorithm via importance ratio control

BibTeX Paper


@article{wang2026mhpo0,
  title   = {MHPO: Modulated Hazard-aware Policy Optimization for Stable Reinforcement Learning},
  author  = {Hongjun Wang and Wei Liu and Weibo Gu and Xing Sun and Kai Han},
  year    = {2026},
  journal = {arXiv preprint arXiv: 2603.16929}
}

Scaling Spatial Propagation to Vision Foundation Models
Yitong Jiang, Collin McCarthy, Hongjun Wang, Hanrong Ye, Qi Dou, Tianfan Xue, Jinwei Gu, Jan Kautz, Hongxu Yin, Pavlo Molchanov, Sifei Liu
CVPR 2026
TL;NR: Scaling up GSPN for VLM

BibTeX Paper


@inproceedings{jiangscaling,
  title={Scaling Spatial Propagation to Vision Foundation Models},
  author={Jiang, Yitong and McCarthy, Collin and Wang, Hongjun and Ye, Hanrong and Dou, Qi and Xue, Tianfan and Gu, Jinwei and Yin, Hongxu and Molchanov, Pavlo and Kautz, Jan and Liu, Sifei},
  booktitle={CVPR},
  year={2026}
}

GSPN-2: Efficient Parallel Sequence Modeling
Hongjun Wang, Yitong Jiang, Collin McCarthy, David Wehr, Hanrong Ye, Xinhao Li, Ka Chun Cheung, Wonmin Byeon, Jinwei Gu, Ke Chen, Kai Han, Hongxu Yin, Pavlo Molchanov, Jan Kautz, Sifei Liu
NeurIPS 2025
TL;NR: Upgraded version of GSPN with a joint algorithm–system redesign

BibTeX Paper Code Project Page


@inproceedings{wanggspn2,
  title={GSPN-2: Efficient Parallel Sequence Modeling},
  author={Wang, Hongjun and Jiang, Yitong and McCarthy, Collin and Wehr, David and Ye, Hanrong and Li, Xinhao and Cheung, Ka Chun and Byeon, Wonmin and Gu, Jinwei and Chen, Ke and Han, Kai and Yin, Hongxu and Molchanov, Pavlo and Kautz, Jan and Liu, Sifei},
  booktitle={NeurIPS},
  year={2025}
}

Panoptic Captioning: Seeking An Equivalency Bridge for Image and Text
Kun-Yu Lin, Hongjun Wang, Weining Ren, Kai Han
NeurIPS 2025
TL;NR: Seek an injective mapping from image to text

BibTeX Paper Code Project Page


@inproceedings{lin2025panoptic,
  title={Panoptic Captioning: Seeking An Equivalency Bridge for Image and Text},
  author={Lin, Kun-Yu and Wang, Hongjun and Ren, Weining and Han, Kai},
  booktitle={NeurIPS},
  year={2025}
}

Fin3R: Fine-Tuning Feed-Foward 3D Reconstruction Models via Monocular Knowledge Distillation
Weining Ren, Hongjun Wang, Xiao Tan, Kai Han
NeurIPS 2025
TL;NR: feed-forward 3D reconstruction networks with distilled geometry knowledge

BibTeX Paper Code Project Page


@inproceedings{ren2025Fin3R,
  title={Fin3R: Fine-Tuning Feed-Foward 3D Reconstruction Models via Monocular Knowledge Distillation},
  author={Ren, Weining and Wang, Hongjun and Tan Xiao and Han, Kai},
  booktitle={NeurIPS},
  year={2025}
}

MultiEdit: Advancing Instruction-based Image Editing on Diverse and Challenging Tasks
Mingsong Li, Lin Liu, Hongjun Wang, Haoxing Chen, Xijun Gu, Shizhan Liu, Dong Gong, Junbo Zhao, Zhenzhong Lan, Jianguo Li
ArXiv
TL;NR: 107K high-quality image editing samples + 6 challenging editing tasks

BibTeX Paper Dataset


@article{li2025multiedit0,
  title   = {MultiEdit: Advancing Instruction-based Image Editing on Diverse and Challenging Tasks},
  author  = {Mingsong Li and Lin Liu and Hongjun Wang and Haoxing Chen and Xijun Gu and Shizhan Liu and Dong Gong and Junbo Zhao and Zhenzhong Lan and Jianguo Li},
  year    = {2025},
  journal = {arXiv preprint arXiv: 2509.14638}
}

Parallel Sequence Modeling via Generalized Spatial Propagation Network
Hongjun Wang, Wonmin Byeon, Jiarui Xu, Jinwei Gu, Ka Chun Cheung, Xiaolong Wang, Kai Han, Jan Kautz, Sifei Liu
CVPR 2025
TL;NR: Super fast attention - $O (\sqrt{N})$ ) with dense and long-context connection

BibTeX Paper Code Project Page Poster Wechat (in Chinese)


@inproceedings{wang2025parallel,
    author    = {Wang, Hongjun and Byeon, Wonmin and Xu, Jiarui and Gu, Jinwei and Cheung, Ka Chun and Wang, Xiaolong and Han, Kai and Kautz, Jan and Liu, Sifei},
    title     = {Parallel Sequence Modeling via Generalized Spatial Propagation Network},
    journal   = {CVPR},
    year      = {2025}
}

HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts
Hongjun Wang, Sagar Vaze, Kai Han
ICLR 2025
TL;NR: Challenging task: semantic + covariate shifts (mixed + unlabeled domains)

BibTeX Paper Code Project Page Poster


@inproceedings{wang2025hilo,
  title={HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts},
  author={Wang, Hongjun and Vaze, Sagar and Han, Kai},
  booktitle={ICLR},
  year={2025}
}

Dissecting Out-of-Distribution Detection and Open-Set Recognition: A Critical Analysis of Methods and Benchmarks
Hongjun Wang, Sagar Vaze, Kai Han
IJCV 2024
TL;NR: Unify and Benchmark OOD and OSR / New metrics

BibTeX Paper Code Project Page Wechat (in Chinese)


@article{wang2024dissect,
  title={Dissecting Out-of-Distribution Detection and Open-Set Recognition: A Critical Analysis of Methods and Benchmarks},
  author={Wang, Hongjun and Vaze, Sagar and Han, Kai},
  journal={IJCV},
  year={2024}
}

SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning
Hongjun Wang, Sagar Vaze, Kai Han
ICLR 2024
TL;NR: Prompt and representation alternative optimization

BibTeX Paper Code Project Page Poster Wechat (in Chinese)


@inproceedings{wang2024sptnet,
  title={SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning},
  author={Wang, Hongjun and Vaze, Sagar and Han, Kai},
  booktitle={ICLR},
  year={2024}
}

Generalist: Decoupling Natural and Robust Generalization
Hongjun Wang, Yisen Wang
CVPR 2023 (Highlight, Top 2.5%)
TL;NR: Bi-experts for clean and adversarial instances

BibTeX Paper Code


@inproceedings{wang2023generalist,
  title={Generalist: Decoupling Natural and Robust Generalization},
  author={Wang, Hongjun and Wang, Yisen},
  booktitle={CVPR},
  year={2023}
}

The Best Protection Is Attack: Fooling Scene Text Recognition with Minimal Pixels
Yikun Xu, Pengwen Dai, Zekun Li, Hongjun Wang, Xiaochun Cao
TIFS 2023
TL;NR: Security verification on OCR

BibTeX Paper


@article{xu2023best,
  title={The Best Protection Is Attack: Fooling Scene Text Recognition with Minimal Pixels},
  author={Xu, Yikun and Dai, Pengwen and Li, Zekun and Wang, Hongjun and Cao, Xiaochun},
  journal={TIFS},
  year={2023}
}

Self-ensemble Adversarial Training for Improved Robustness
Hongjun Wang, Yisen Wang
ICLR 2022
TL;NR: Smoothness helps adversarial robustness

BibTeX Paper Code


@inproceedings{wang2022selfensemble,
  title={Self-ensemble Adversarial Training for Improved Robustness},
  author={Wang, Hongjun and Wang, Yisen},
  booktitle={ICLR},
  year={2022}
}

A Hamiltonian Monte Carlo Method for Probabilistic Adversarial Attack and Learning
Hongjun Wang, Guanbin Li, Xiaobai Liu, Liang Lin
TPAMI 2020 (IF: ~24.31)
TL;NR: Hamiltonian Monte Carlo for generation diversity

BibTeX Paper Code


@article{wang2020hamiltonian,
  title={A Hamiltonian Monte Carlo method for probabilistic adversarial attack and learning},
  author={Wang, Hongjun and Li, Guanbin and Liu, Xiaobai and Lin, Liang},
  journal={T-PAMI},
  year={2020}
}

Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking
Hongjun Wang*, Guangrun Wang*, Ya Li, Dongyu Zhang, Liang Lin
CVPR 2020 (Oral, Top 3%)
TL;NR: Comprehensive security investigation on SoTA ReIDs

BibTeX Paper Code Wechat (in Chinese)


@inproceedings{wang2020transferable,
  title={Transferable, controllable, and inconspicuous adversarial attacks on person re-identification with deep mis-ranking},
  author={Wang, Hongjun and Wang, Guangrun and Li, Ya and Zhang, Dongyu and Lin, Liang},
  booktitle={CVPR},
  year={2020}
}

CamDrop: A New Explanation of Dropout and A Guided Regularization Method for Deep Neural Networks
Hongjun Wang, Guangrun Wang, Guanbin Li, Liang Lin
CIKM 2019 (Oral)
TL;NR: Semantics-guided dropout

BibTeX Paper Code


@inproceedings{wang2019camdrop,
  title={Camdrop: A new explanation of dropout and a guided regularization method for deep neural networks},
  author={Wang, Hongjun and Wang, Guangrun and Li, Guanbin and Lin, Liang},
  booktitle={CIKM},
  year={2019}
}

Crowd Counting using Deep Recurrent Spatial-Aware Network
Lingbo Liu, Hongjun Wang, Guanbin Li, Wanli Ouyang, Liang Lin
IJCAI 2018 (Oral)
TL;NR: Recurrent Spatial-Aware Refinement for crowd counting

BibTeX Paper


@inproceedings{liu2018crowd,
  title={Crowd Counting using Deep Recurrent Spatial-Aware Network},
  author={Liu, Lingbo and Wang, Hongjun and Li, Guanbin and Ouyang, Wanli and Lin, Liang},
  booktitle={IJCAI},
  year={2018}
}

About

Research Roadmap

Publications

Services

Conference & Journal Reviews

Teaching

Awards