Issue |
ITM Web Conf.
Volume 70, 2025
2024 2nd International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2024)
|
|
---|---|---|
Article Number | 01009 | |
Number of page(s) | 9 | |
Section | Traffic Prediction and Analysis | |
DOI | https://doi.org/10.1051/itmconf/20257001009 | |
Published online | 23 January 2025 |
A LiDAR - camera fusion detection method based on weight allocation
The Graduate School, Northwestern University, Evanston, IL 60208, United States
* Corresponding author: qinlei@Idy.edu.rs
In automatic driving target detection problem, the neural network is applied to two methods, vision, and LiDAR. These two methods have some relatively mature models based on neural networks. Combining the two to complement each other has become a hot topic. At present, most autonomous driving sensor fusion methods focus on fusion strategy and feature alignment, and there are few studies on the weight ratio of the two sensors after fusion in different environments. In this paper, a fusion target detection model of camera and LiDAR is proposed based on the weighted weight allocation method. The weighted fusion method is adopted, image feature points are extracted by Fast RCNN, and then LiDAR point cloud data is fused into the model by the weighted method, environment variables are introduced, and different weight allocation methods are output under different environments through full connection layer preprocessing. The results on the Nuscenes dataset show that compared with the results without weight assignment, the model can effectively achieve targeted weight assignment in different situations, and the performance is due to the single-sensor method.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.