K
Kelvin Wong
Researcher at Uber
Publications - 32
Citations - 536
Kelvin Wong is an academic researcher from Uber . The author has contributed to research in topics: Point cloud & Octree. The author has an hindex of 8, co-authored 32 publications receiving 199 citations. Previous affiliations of Kelvin Wong include University of Tokyo & University College London.
Papers
More filters
Proceedings ArticleDOI
LiDARsim: Realistic LiDAR Simulation by Leveraging the Real World
Sivabalan Manivasagam,Shenlong Wang,Kelvin Wong,Wenyuan Zeng,Mikita Sazanovich,Shuhan Tan,Bin Yang,Wei-Chiu Ma,Raquel Urtasun +8 more
TL;DR: LiDARsim as mentioned in this paper uses a deep neural network to produce deviations from the physics-based simulation, producing realistic LiDAR point clouds, which can be used for perception algorithms-testing on long-tail events and end-to-end closed-loop evaluation on safetycritical scenarios.
Identifying Unknown Instances for Autonomous Driving
TL;DR: A novel open-set instance segmentation algorithm for point clouds which can segment objects from both known and unknown classes in a holistic way and validate the effectiveness of the proposed method on large-scale self-driving datasets.
Posted Content
OctSqueeze: Octree-Structured Entropy Model for LiDAR Compression
TL;DR: In this article, a tree-structured conditional entropy model was proposed to encode the LiDAR points into an octree, a data-efficient structure suitable for sparse point clouds.
Proceedings ArticleDOI
OctSqueeze: Octree-Structured Entropy Model for LiDAR Compression
TL;DR: A novel deep compression algorithm to reduce the memory footprint of LiDAR point clouds and designs a tree-structured conditional entropy model that can be directly applied to octree structures to predict the probability of a symbol’s occurrence.
Proceedings ArticleDOI
SceneGen: Learning to Generate Realistic Traffic Scenes
TL;DR: SceneGen as mentioned in this paper is a neural autoregressive model of traffic scenes that eschews the need for rules and heuristics to model the complexity and diversity of real traffic scenes, thus inducing a content gap between synthesized traffic scenes versus real ones.