Open Access

ARTICLE

Camera-LiDAR Fusion for Enhanced Object Detection

Jianping Wu¹, Nian Li^2,*, Libin Dong³, Ping Zhang⁴

1 School of Computer Science and Technology, Changsha University of Science and Technology, Changsha, China
2 School of Physics and Electronic Science, Changsha University of Science and Technology, Changsha, China
3 Zhejiang Datang Wushashan Power Generation Co., Ltd., Ningbo, China
4 Fuel Business Division, Hunan Datang Xianyi Technology Co., Ltd., Changsha, China

* Corresponding Author: Nian Li. Email:

Journal on Artificial Intelligence 2026, 8, 259-271. https://doi.org/10.32604/jai.2026.075753

Received 07 November 2025; Accepted 19 January 2026; Issue published 12 May 2026

Abstract

This paper presents a static fusion framework that enhances object detection by integrating camera and LiDAR-based detection results. The proposed method focuses on associating 2D candidate bounding boxes from a camera detector with 3D candidate boxes from a LiDAR detector using an Intersection over Union (IoU)-based matching approach. To enhance the quality of 2D detection, we refine the baseline Cascade R-CNN detector by incorporating a dual self-attention mechanism into both the backbone and the region proposal network (RPN), resulting in the DA-Cascade R-CNN. This enhancement strengthens the network’s ability to detect small or distant objects by improving feature sensitivity and localization accuracy. Once 2D and 3D candidate boxes are obtained, they are associated through IoU-aware matching and subsequently refined using non-maximum suppression (NMS) to remove redundant or conflicting hypotheses across modalities, effectively preserving positive detection results to improve accuracy. Experimental results on the KITTI dataset demonstrate that the proposed static fusion method yields improved detection average precision for three different levels of difficulty compared to single-sensor baselines.

---

Keywords

---