Self-Supervised Learning for 3D Scene Flow Detection
3D Object Detection with a Self-supervised Lidar Scene Flow Backbone
We propose using a self-supervised training strategy to learn a general point cloud backbone model for downstream 3d vision tasks focusing mainly on the relation between the scene flow and detection tasks.Moreover, the perception of objects in the traffic scenarios heavily relies on making sense of the sparse data in the spatio-temporal context.In this way, self-supervised scene flow training constructs point motion features in the backbone, which help distinguish objects based on their different motion patterns used with a 3d detection head.Experiments on kitti and nuscenes benchmarks show that the proposed self-supervised pre-training increases 3d detection performance significantly.
