3D fully convolutional network for vehicle detection in point cloud

2D fully convolutional network has been recently successfully applied to the object detection problem on 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 significant performance improvement over the previous point cloud based detection approaches.

Authors

• Bo Li

Keywords

• Three-dimensional displays
• Two dimensional displays
• Object detection
• Vehicle detection
• Machine learning
• Laser radar
• Convolutional Network
• Point Cloud
• Fully Convolutional Network
• Detection In Point Clouds
• Detection Task
• Point Cloud Data
• Deep Learning
• Convolutional Neural Network
• Positive Samples
• Aspect Ratio
• Image Plane
• Object Recognition
• Bounding Box
• Ground Plane
• Square Grid
• Object Detection Task
• Object Recognition Task
• 3D Object Detection
• 3D Detection
• Predicted Bounding Box
• World Space
• Map Objects
• Object Bounding Boxes
• RGB-D Data
• Voxel Data
• 3D Bounding Box
• Pedestrian