CenterFusion: Center-based Radar and Camera Fusion for 3D Object Detection

By Ramin Nabati, Hairong Qi. IEEE Workshop/Winter Conference on Applications of Computer Vision, 2020

397 citations recorded by the source corpus.

The perception system in autonomous vehicles is responsible for detecting and tracking the surrounding objects. This is usually done by taking advantage of several sensing modalities to increase robustness and accuracy, which makes sensor fusion a crucial part of the perception system. In this paper, we focus on the problem of radar and camera sensor fusion and propose a middle-fusion approach to exploit both radar and camera data for 3D object detection. Our approach, called CenterFusion, first uses a center point detection network to detect objects by identifying their center points on the image. It then solves the key data association problem using a novel frustum-based method to associate the radar detections to their corresponding object’s center point. The associated radar detections are used to generate radar-based feature maps to complement the image features, and regress to object properties such as depth, rotation and velocity. We evaluate CenterFusion on the challenging nuScenes dataset, where it improves the overall nuScenes Detection Score (NDS) of the state-of-the-art camera-based algorithm by more than 12%. We further show that CenterFusion significantly improves the velocity estimation accuracy without using any additional temporal information. The code is available at https://github.com/mrnabati/CenterFusion.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Mechanical Engineering

View Paper: https://www.semanticscholar.org/paper/a7c88d4d3b6ebd0c69625053921730001f2c2a15

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

1. Guan Huang
2. Junjie Huang
3. Yue Wang
4. Yilun Wang
5. Hang Zhao
6. Jiangmiao Pang
7. Song Han
8. Xinge Zhu

In Computer Vision and Pattern Recognition

1. Song Han
2. Xiaogang Wang
3. Luc Van Gool
4. Dahua Lin
5. Dacheng Tao
6. Alan L. Yuille
7. Guan Huang
8. Hongsheng Li

In Mechanical Engineering

1. Guan Huang
2. Daniela Rus
3. Dahua Lin
4. Tai Wang
5. Yue Wang
6. Xinge Zhu
7. Junjie Huang
8. Jiangmiao Pang

Citations in the corpus

• BEVDet: High-performance Multi-camera 3D Object Detection in Bird-Eye-View (2021, neutral): “453 CenterFusion [27] Camera & Radar 0.”
• FUTR3D: A Unified Sensor Fusion Framework for 3D Detection (2022, background): “Conducting camera-radar fusion usually involves more complicated feature alignment techniques due to the sparsity of radar signals [3, 10, 18, 19].”
• BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation (2022, background): “[45], [46], [47] all lift image proposals into a 3D frustum.”
• Sensor and Sensor Fusion Technology in Autonomous Vehicles: A Review (2021, background): “Readers interested in detailed discussions about sensor fusion techniques and algorithms for various application ranging from environmental perception, including 2D or 3D object detection and lane tracking, to localization and mapping are recommended to refer to [16,26,181,187-199].”
• BEVDet4D: Exploit Temporal Cues in Multi-camera 3D Object Detection (2022, methodology): “For the first time, the precision of the velocity prediction in the camera-based methods notably exceeds the CenterFusion [27] 0.540 mAVE, who relies on the multi-sensor fusion with camera and radar for high precision in this aspect.”
• FCOS3D: Fully Convolutional One-Stage Monocular 3D Object Detection (2021, methodology): “For the approaches which use the input of RGB image and Radar data, we select CenterFusion [22] as the benchmark.”
• Probabilistic and Geometric Depth: Detecting Objects in Perspective (2021, methodology): “We also list benchmarks based on other data modality, including lightweight, real-time PointPillars [2] with LiDAR, CenterFusion [7] with RGB image and Radar, and CenterPoint [52] ensemble results with all the sensors.”
• BEVFusion: A Simple and Robust LiDAR-Camera Fusion Framework (2022, background): “There exist other works to leverage other modalities, such as fusing camera-radar by feature map concatenation [3, 18, 31, 30].”