Efficient 3D Perception for Autonomous Vehicles

Explore cutting-edge advancements in efficient 3D perception for autonomous vehicles in this invited talk from the CVPR 2023 ECV Workshop. Delve into BEVFusion, a groundbreaking approach that unifies camera, LiDAR, and radar features in a shared bird's-eye view space, accelerating multi-task multi-sensor fusion by 40 times. Discover how this leading solution achieves top performance across popular 3D perception benchmarks for tasks such as object detection, tracking, and map segmentation. Learn about FlatFormer, an efficient point cloud transformer that enables real-time performance on edge GPUs, and SparseViT, which leverages spatial sparsity in 2D image transformers for improved speed without sacrificing accuracy. Gain insights into the latest research addressing the efficiency challenges in autonomous vehicle perception, paving the way for real-world applications.

Efficient 3D Perception for Autonomous Vehicles (Zhijian Liu @ CVPR 2023 ECV)