Since currently LiDARs, cameras and radars are optimal solutions for autonomous driving perception tasks, DNN-based sensor fusion is an important milestone to achieve universal yet robust performance of prediction for 3D Object Detection.

Car provided by Integrant company is equipped with LiDARs, global shutter cameras and GPS sensor.

FADet is a multi-sensor 3D detection network, which specifically studies the characteristics of different sensors based on our local featured attention modules. For camera images, we propose dual-attention-based sub-module. For LiDAR point clouds, triple-attention-based sub-module is utilized while mixed-attention-based sub-module is applied for features of radar points. With local featured attention sub-modules, FADet has effective detection results in long-tail and complex scenes from camera, LiDAR and radar input.

Project status: under development.