A bit about my research

My current research interest mainly lies in multimodal understanding and generation. In the past, I developed algorithms for learning representations, understanding people, and segmenting images. Below is a selection of my publications, the details of which can be found on Google Scholar. ​​​

  • F.Y. Wang, H. Zhou, L. Yuan, S. Woo, B. Gong, B. Han, M.H. Yang, H. Zhang, Y. Zhu, T. Liu, L. Zhao. Image diffusion preview with consistency solver. Tech Report, 2026. [paper][code]

  • Gemini Team, Google. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. Tech Report, 2025. [paper][blog]
 
  • L. Zhao, S. Woo, Z. Wan, Y. Li, H. Zhang, B. Gong, H. Adam, X. Jia, T. Liu. Epsilon-VAE: Denoising as visual decoding​. ICML 2025. [paper]
 
  • Z. Li, H. Shi, Y. Gao, D. Liu, Z. Wang, Y. Chen, T. Liu, L. Zhao, H. Wang, D.N. Metaxas. The hidden life of tokens: Reducing hallucination of large vision-language models via visual information steering. ICML 2025. [paper]
 
  • Y. Sun, H. Zhou, L. Yuan, J.J. Sun, Y. Li, X. Jia, H. Adam, B. Hariharan, L. Zhao, T. Liu. Video creation by demonstration. Tech Report, 2024. [paper][website]
 
  • J.J. Sun, H. Zhou, L. Zhao, L. Yuan, B. Seybold, D. Hendon, F. Schroff, D.A. Ross, H. Adam, B. Hu*, T. Liu*. Video foundation models for animal behavior analysis. Tech Report, 2024. [paper] (* Equal contribution)
 
  • L. Yuan*, N.B. Gundavarapu*, L. Zhao*, H. Zhou*, Y. Cui, L. Jiang, X. Yang, M. Jia, T. Weyand, L. Friedman, M. Sirotenko, H. Wang, F. Schroff, M.-H. Yang, T. Liu, B. Gong. VideoGLUE: Video general understanding evaluation of foundation models. Transactions on Machine Learning Research (TMLR), 2024. ​[paper][code] (* Equal contribution)
 
  • X. Zhu, H. Zhou, P. Xing, L. Zhao, H. Xu, J. Liang, A. Hauptmann, T. Liu, A.C. Gallagher​. ​Open-vocabulary 3D semantic segmentation with text-to-image diffusion models. ECCV 2024. [paper][website]

  • L. Zhao*, N.B. Gundavarapu*, L. Yuan*, H. Zhou*, S. Yan**, J.J. Sun**, L. Friedman**, R. Qian**, T. Weyand, Y. Zhao, R. Hornung, F. Schroff, M.-H. Yang, D.A. Ross, H. Wang, H. Adam, M. Sirotenko***, T. Liu***, B. Gong***. VideoPrism: A foundational visual encoder for video understanding. ICML 2024. [paper][blog][website][hf] (* Equal primary contribution; ** Equal core technical contribution; *** Equal senior contribution, project leads)
 
  • Y. Zhao, L. Zhao, X. Zhou, J. Wu, C.-T. Chu, H. Miao, F. Schroff, H. Adam, T. Liu, B. Gong, P. Krähenbühl, L. Yuan. Distilling vision-language models on millions of videos. CVPR 2024. ​[paper][website]
 
  • Y. Xiong, L. Zhao, B. Gong, M.-H. Yang, F. Schroff, T. Liu, C.-J. Hsieh, L. Yuan. Structured video-language modeling with temporal grouping and spatial grounding. ICLR 2024. ​[paper]
 
  • Z. Li, L. Zhao, Z. Zhang, H. Zhang, D. Liu, T. Liu, D.N. Metaxas. Steering prototype with prompt-tuning for rehearsal-free continual learning​. WACV 2024. ​[paper]
 
  • L. Zhao, L. Yuan, B. Gong, Y. Cui, F. Schroff, M.-H. Yang, H. Adam, T. Liu. Unified visual relationship detection with vision and language models. ICCV 2023. ​[paper]
 
  • Q. Wang, L. Zhao, L. Yuan, T. Liu, X. Peng. Learning from semantic alignment between unpaired multiviews for egocentric video recognition. ICCV 2023. ​[paper]
 
  • Z. Xu*, M.D. Collins*, Y. Wang, L. Panait, S. Oh, S. Augenstein, T. Liu, F. Schroff, H.B. McMahan. Learning to generate image embeddings with user-level differential privacy. CVPR 2023. [paper] (* Equal contribution)
 
  • T.R. Scott, T. Liu, M.C. Mozer, A.C. Gallagher. An Empirical Study on Clustering Pretrained Embeddings: Is deep strictly better? Tech Report, 2022. [paper]
 
  • R. Qian, Y. Li, L. Yuan, B. Gong, T. Liu, M. Brown, S. Belongie, M.-H. Yang, H. Adam, Y. Cui. On temporal granularity in self-supervised video representation learning. BMVC 2022. [paper]
 
  • H. Zhou, A. Kadav, A. Shamsian, S. Geng, F. Lai, L. Zhao, T. Liu, M. Kapadia, H.P. Graf. COMPOSER: Compositional reasoning of group activity in videos with keypoint-only modality. ECCV 2022. [paper][code]

  • L. Yuan, R. Qian, Y. Cui, B. Gong, F. Schroff, M.-H. Yang, H. Adam, T. Liu. Contextualized spatio-temporal contrastive learning with self-supervision. CVPR 2022. [paper]
 
  • J. Zheng, X. Shi, A. Gorban, J. Mao, Y. Song, C.R. Qi, T. Liu, V.V. Chari, A. Cronman, Y. Zhou, C. Li, D. Anguelov. Multi-modal 3D human pose estimation with 2D weak supervision in autonomous driving. CVPR 2022 Workshop on Autonomous Driving. [paper]
 
  • J. Zhuang, B. Gong, L. Yuan, Y. Cui, H. Adam, N.C. Dvornek, S. Tatikonda, J.S. Duncan, T. Liu. Surrogate gap minimization improves sharpness-aware training. ICLR 2022. [paper][website][checkpoints]
 
  • T. Liu*, J.J. Sun*, L. Zhao, J. Zhao, L. Yuan, Y. Wang, L.-C. Chen, F. Schroff, H. Adam. View-invariant, occlusion-robust probabilistic embedding for human pose. International Journal of Computer Vision (IJCV), 130, 111-135, 2022. [paper][blog][code][website] (* Equal contribution)
 
  • L. Zhao, Y. Wang, J. Zhao, L. Yuan, F. Schroff, H. Adam, X. Peng, D.N. Metaxas, T. Liu. Learning view-disentangled human pose representation by contrastive cross-view mutual information maximization. CVPR 2021 (Oral). [paper][code]
  
  • L. Zhao, T. Liu, X. Peng, D.N. Metaxas. Maximum-entropy adversarial data augmentation for improved generalization and robustness. NeurIPS 2020. [paper][code]
 
  • J.J. Sun, J. Zhao, L.-C. Chen, F. Schroff, H. Adam, T. Liu. View-Invariant probabilistic embedding for human pose. ECCV 2020 (Spotlight). [paper][blog][talk][short talk][code][website]
 
  • B. Cheng, M.D. Collins, Y. Zhu, T. Liu, T.S. Huang, H. Adam, L.-C. Chen. Panoptic-DeepLab: A simple, strong, and fast baseline for bottom-up panoptic segmentation. CVPR 2020. [paper][code]
 
  • J.J. Sun, T. Liu, A.S. Cowen, F. Schroff, H. Adam, G. Prasad. EEV Dataset: Predicting expressions evoked by diverse videos. Tech Report, 2020. [paper][workshop]
   
  • B. Cheng, M.D. Collins, Y. Zhu, T. Liu, T.S. Huang, H. Adam, L.-C. Chen. Panoptic-DeepLab. ICCV 2019 Joint COCO and Mapillary Recognition Workshop (Winner of the Mapillary Vista Panoptic Segmentation Task, Best Paper and Most Innovative Award). [paper][code]
 
  • T.-J. Yang, M.D. Collins, Y. Zhu, J.-J. Hwang, T. Liu, X. Zhang, V. Sze, G. Papandreou, L.-C. Chen. DeeperLab: Single-shot image parser. Tech Report, 2019. [paper]
 
  • J.J. Sun, T. Liu, G. Prasad. GLA in MediaEval 2018 Emotional Impact of Movies Task. MediaEval 2018 Multimedia Benchmark Workshop (Winner of the Emotional Impact of Movies Task). [paper][talk]
 
  • P. Nguyen*, T. Liu*, G. Prasad, B. Han. Weakly supervised action localization by sparse temporal pooling network. CVPR 2018. [paper][code] (* Equal contribution)
​
  • T. Liu, M. Zhang, M. Javanmardi, N. Ramesh, T. Tasdizen. SSHMT: Semi-supervised hierarchical merge tree for electron microscopy image segmentation. ECCV 2016 (Spotlight).​ [paper][talk][code]
 
  • T. Liu, M. Seyedhosseini, T. Tasdizen. Image segmentation using hierarchical merge tree. IEEE Transactions on Image Processing, 25, pp. 4596-4607, 2016. [paper][code]
 
  • M. Javanmardi, M. Sajjadi, T. Liu, T. Tasdizen. Unsupervised total variational loss for semi-supervised learning of semantic segmentation. Tech Report, 2016. [paper]
 
  • I. Arganda-Carreras, S.C. Turaga, D.R. Berger, D. Ciresan, A. Giusti, L.M. Gambardella, J. Schmidhuber, D. Laptev, S. Dwivedi, J.M. Buhmann, T. Liu, M. Seyedhosseini, T. Tasdizen, L. Kamentsky, R. Burget, V. Uher, X. Tan, C. Sun, T.D. Pham, E. Bas, M.G. Uzunbas, A. Cardona, J. Schindelin, H.S. Seung. Crowdsourcing the creation of image segmentation algorithms for connectomics. Frontiers in Neuroanatomy, 9:142, 2015. [paper]
 
  • C. Jones, T. Liu, N.W. Cohan, M.H. Ellisman, T. Tasdizen. Efficient semi-automatic 3D segmentation for neuron tracing in electron microscopy images. Journal of Neuroscience Methods, 246, pp. 13-21, 2015. [paper]
 
  • T. Liu, C. Jones, M. Seyedhosseini, T. Tasdizen. A modular approach to 3D electron microscopy image segmentation. Journal of Neuroscience Methods, 226, 88-102, 2014. [paper][code]
 
  • T. Tasdizen, M. Seyedhosseini, T. Liu, C. Jones, E. Jurrus. Image segmentation for connectomics using machine learning. Computational Intelligence in Biomedical Imaging, pp. 237-278, Springer New York, 2014. [paper]
 
  • T. Liu, M. Seyedhosseini, M.H. Ellisman, T. Tasdizen. Watershed merge forest classification for electron microscopy image stack segmentation. ICIP 2013. [paper]
 
  • C. Jones, T. Liu, M.H. Ellisman, T. Tasdizen. Semi-automatic neuron segmentation in electron microscopy images via sparse labeling. ISBI 2013. [paper]
 
  • T. Liu, E. Jurrus, M. Seyedhosseini, M.H. Ellisman, T. Tasdizen. Watershed merge tree classification for electron microscopy image segmentation. ICPR 2012. [paper][code]