no code implementations • 6 May 2022 • Shuo Gu, Suling Yao, Jian Yang, Hui Kong
Instead of segmenting the moving objects directly, the network conducts single-scan-based semantic segmentation and multiple-scan-based moving object segmentation in turn.