Abstract: Highlights•Propose SuPrNet, which combines 4D scene priors, deep features, and dynamic priors.•SuPrNet achieves SOTA on the Nuscenes and KITTI-Odometry datasets.•SuPrNet can be used for LiDAR NVS and temporal data generation tasks.•Ablation experiments show that SuPrNet is a product of framework innovation.
External IDs:dblp:journals/kbs/LiangCFZ24
Loading