3D-SiamRPN: An End-to-End Learning Method for Real-Time 3D Single Object Tracking Using Raw Point Cloud

TL;DR

This paper introduces 3D-SiamRPN, a real-time end-to-end deep learning method for tracking a single object in 3D point cloud data, with high accuracy and generalization ability demonstrated on KITTI and H3D datasets.

Contribution

The paper proposes a novel 3D tracking network combining PointNet++ feature extraction, cross correlation modules, and a region proposal network for accurate, real-time 3D object tracking.

Findings

Achieves competitive success and precision on KITTI dataset.

Runs in real-time at 20.8 FPS.

Demonstrates good generalization on H3D dataset.

Abstract

3D single object tracking is a key issue for autonomous following robot, where the robot should robustly track and accurately localize the target for efficient following. In this paper, we propose a 3D tracking method called 3D-SiamRPN Network to track a single target object by using raw 3D point cloud data. The proposed network consists of two subnetworks. The first subnetwork is feature embedding subnetwork which is used for point cloud feature extraction and fusion. In this subnetwork, we first use PointNet++ to extract features of point cloud from template and search branches. Then, to fuse the information of features in the two branches and obtain their similarity, we propose two cross correlation modules, named Pointcloud-wise and Point-wise respectively. The second subnetwork is region proposal network(RPN), which is used to get the final 3D bounding box of the target object based on the fusion feature from cross correlation modules. In this subnetwork, we utilize the regression and classification branches of a region proposal subnetwork to obtain proposals and scores, thus get the final 3D bounding box of the target object. Experimental results on KITTI dataset show that our method has a competitive performance in both Success and Precision compared to the state-of-the-art methods, and could run in real-time at 20.8 FPS. Additionally, experimental results on H3D dataset demonstrate that our method also has good generalization ability and could achieve good tracking performance in a new scene without re-training.