HOPE: Hierarchical Spatial-temporal Network for Occupancy Flow Prediction

• DOI: 10.48550/arxiv.2206.10118
• Publication Date: 2022-01-01
• Authors: Hu, Yihan; Shao, Wenxin; Jiang, Bo; Chen, Jiajie; Chai, Siqi; Yang, Zhening; Qian, Jingyu; Zhou, Helong; Liu, Qiang
• Keywords: FOS: Computer and information sciences; 03 medical and health sciences; 0302 clinical medicine; Computer Vision and Pattern Recognition (cs.CV); Computer Science - Computer Vision and Pattern Recognition

In this report, we introduce our solution to the Occupancy and Flow Prediction challenge in the Waymo Open Dataset Challenges at CVPR 2022, which ranks 1st on the leaderboard. We have developed a novel hierarchical spatial-temporal network featured with spatial-temporal encoders, a multi-scale aggregator enriched with latent variables, and a recursive hierarchical 3D decoder. We use multiple losses including focal loss and modified flow trace loss to efficiently guide the training process. Our method achieves a Flow-Grounded Occupancy AUC of 0.8389 and outperforms all the other teams on the leaderboard.<br/>1st Ranking Solution for the Occupancy and Flow Prediction of the Waymo Open Dataset Challenges 2022 (http://cvpr2022.wad.vision/)<br/>