Multiple Objects Detection and Localization using Data Fusion

Real-time object detection is one of the most fundamental computer vision processes, it is of great importance for many applications, such as video surveillance and autonomous driving. Object detection with multiple modalities of data is of great importance as it provides more information about the surrounding objects. Most 3D object detection approaches are based on utilizing expensive sensors and/or high computational requirements. Therefore, in this paper, we propose a low-cost real-time multiple objects detection platform using a low-cost 2D-LiDAR and an RGB camera. YOLOv3 has been adopted as the object detection algorithm, moreover, a novel calibration technique has been proposed to achieve mapping between the camera and LiDAR. Experiments were made on real-world scenarios; the performance of the system was validated by manually measuring the distances of detected objects. The system outputs surrounding objects along with their classification and their 3D location in the scene.