PointFusion: Deep Sensor Fusion for 3D Bounding Box Estimation
This addresses 3D object detection for autonomous driving and indoor robotics by providing a generic fusion approach that works across diverse sensor setups.
The authors tackled 3D object detection by fusing image and point cloud data with a simple, application-agnostic method, achieving performance on-par or better than state-of-the-art on KITTI and SUN-RGBD datasets without dataset-specific tuning.
We present PointFusion, a generic 3D object detection method that leverages both image and 3D point cloud information. Unlike existing methods that either use multi-stage pipelines or hold sensor and dataset-specific assumptions, PointFusion is conceptually simple and application-agnostic. The image data and the raw point cloud data are independently processed by a CNN and a PointNet architecture, respectively. The resulting outputs are then combined by a novel fusion network, which predicts multiple 3D box hypotheses and their confidences, using the input 3D points as spatial anchors. We evaluate PointFusion on two distinctive datasets: the KITTI dataset that features driving scenes captured with a lidar-camera setup, and the SUN-RGBD dataset that captures indoor environments with RGB-D cameras. Our model is the first one that is able to perform better or on-par with the state-of-the-art on these diverse datasets without any dataset-specific model tuning.