Logo Action Lab

The ACTION Lab is dedicated to teaching machines to better perceive the visual world. We focus on discovering fundamental principles and algorithms for solving machine visual perception problems. Our goal is to improve machine intelligence and explore new ways to make machines learn to change the world for social good.



03/2022. We have three papers accepted by CVPR 2022.

09/2021. We have one paper accepted by BMVC 2021.

07/2021. We have three papers accepted by ICCV 2021. Congratulations to Wentao and Junwen! [DEAR 😮 oral], [DRIVE], [Entailment]

04/2021. We have received funding of $360K from the Army Research Office. Cheers!

04/2021. Our paper "Multiple Instance Relational Learning for Video Anomaly Detection" is accepted by IJCNN, 2021. This is Xiwen Dengxiong's first paper. Congratulations to Xiwen and Wentao!

07/2020. We have two papers accepted by ACM Multimedia 2020. Congratulations to Wentao and Junwen!

07/2020. Our paper "Group Activity Prediction with Sequential Relational Anticipation Model" is accepted by ECCV, 2020. Congratulations to Junwen and Wentao! 

06/2020. Our paper "Object-Aware Centroid Voting for Monocular 3D Object Detection" is accepted by IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2020. Congratulations to Wentao! 

06/2020. RIT-18 dataset for understanding compositional group activity is available!

04/2020. Our paper "Few-shot Human Motion Prediction via Learning Novel Motion Dynamics" is accepted by International Joint Conference on Artificial Intelligence (IJCAI), 2020. This is Chuanqi Zang's first paper. Congratulations! 

04/2020. Our paper "RIT-18: A Novel Dataset for Compositional Group Activity Understanding" is accepted by CVPR Workshop 2020. This is Junwen's first paper after joining the ACTION Lab. Congratulations to Junwen, Haiting, and Hanbin! 

View all >


Generalized Visual Forecasting

Enable machines' capability of peeking into the future.

Data collection
A Multimodal Dynamic Bayesian Learning Framework for Complex Decision-making

Multimodal dynamic Bayesian learning framework to facilitate complex decision-making in various highly complicated battlefield scenarios and military tasks.

Our People

Photo of Yukong

Yu Kong, Ph.D.

Assistant Professor 
Ph.D. Program in Computing and Information Sciences
Golisano College of Computing and Information Sciences


Please Contact Dr. Yu Kong if you are interested in our work.

Contact Us >