3D Fully Convolutional Network for Vehicle Detection in Point Cloud

TL;DR

This paper extends 2D fully convolutional networks to 3D for vehicle detection in lidar point clouds, demonstrating significant performance improvements on the KITTI dataset for autonomous driving applications.

Contribution

It introduces a novel 3D fully convolutional network approach for point cloud object detection, advancing beyond previous methods.

Findings

Significant performance improvement over prior point cloud detection methods

Effective application of 3D FCN to lidar data for vehicle detection

Validated on KITTI dataset with promising results

Abstract

2D fully convolutional network has been recently successfully applied to object detection from images. In this paper, we extend the fully convolutional network based detection techniques to 3D and apply it to point cloud data. The proposed approach is verified on the task of vehicle detection from lidar point cloud for autonomous driving. Experiments on the KITTI dataset shows a significant performance improvement over the previous point cloud based detection approaches.