Academically, generally speaking, I am interested in building a faster and stronger perception system. Concretely, I am interested in building an efficient, multimodal, and temporal perception system. Concretely:
Efficiency: I am interested in improving the efficiency from both model (compression and pruning) and data (self-supervised learning, pseudo-labeling, transfer learning) perspectives.
Multimodal: I am interested in incorporating signals from multiple sources and modalities like LiDAR and radars into the perception system (sensor fusion, lidar point clouds).
Temporal: I am interested in better leveraging the 4D video data and giving the perception system a better temporal sense (video understanding, visual tracking).
I've been honored to work with Humphrey Shi from SHI/IFP lab at UIUC and GaTech, Yuxiong Wang at UIUC, Jean Ponce at INRIA/NYU, Leonidas Guibas at Stanford, and Jose M. Alvarez 's amazing team at NVIDIA.
Academic Services: Reviewer for CVPR, WACV, IROS
Hardware-Aware Latency Pruning for Real-Time 3D Object Detection
M. Shen, L. Mao, J. Chen, X. Sun, O. Knieps, C. Maxim, J. M. Alvarez
2023 IEEE Intelligent Vehicles Symposium (IV) (IEEE IV), 2023
Revisiting Deformable Convolution for Depth Completion
X. Sun, J. Ponce, Y. Wang
2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023
Towards Better Structured Pruning Saliency by Reorganizing Convolution
X. Sun, H.Shi
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2024
LookAhead Dynamic Sparse Training
X. Sun, M. Shen, H. Yin, L. Mao, P. Molchanov, J. M. Alvarez
Refining Pre-Trained Motion Models
X. Sun, A.W. Harley, L. Guibas
Tag: Temporal Tag: Efficiency