Fusion4CA: Boosting 3D Object Detection via Comprehensive Image Exploitation

2026-03-05

AI Summary

This work addresses the overreliance on LiDAR and underutilization of visual information in existing bird’s-eye-view (BEV) fusion methods for 3D object detection. Building upon the BEVFusion framework, the authors propose a lightweight and efficient multimodal fusion strategy that calibrates image features with 3D geometric structure via a contrastive alignment module, enhances RGB information utilization during training through a camera-assisted branch, and refines BEV-space fusion by integrating a cognitive adapter with coordinate attention. The proposed method introduces only a 3.48% increase in parameters and achieves a 69.7% mAP on the nuScenes benchmark after just six training epochs—outperforming the baseline trained for 20 epochs by 1.2%. Furthermore, its generalization capability is validated in a simulated lunar environment.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Nowadays, an increasing number of works fuse LiDAR and RGB data in the bird's-eye view (BEV) space for 3D object detection in autonomous driving systems. However, existing methods suffer from over-reliance on the LiDAR branch, with insufficient exploration of RGB information. To tackle this issue, we propose Fusion4CA, which is built upon the classic BEVFusion framework and dedicated to fully exploiting visual input with plug-and-play components. Specifically, a contrastive alignment module is designed to calibrate image features with 3D geometry, and a camera auxiliary branch is introduced to mine RGB information sufficiently during training. For further performance enhancement, we leverage an off-the-shelf cognitive adapter to make the most of pretrained image weights, and integrate a standard coordinate attention module into the fusion stage as a supplementary boost. Experiments on the nuScenes dataset demonstrate that our method achieves 69.7% mAP with only 6 training epochs and a mere 3.48% increase in inference parameters, yielding a 1.2% improvement over the baseline which is fully trained for 20 epochs. Extensive experiments in a simulated lunar environment further validate the effectiveness and generalization of our method. Our code will be released through Fusion4CA.

Problem

• 3D object detection
• LiDAR-RGB fusion
• bird's-eye view
• image exploitation
• autonomous driving

Innovation

• BEV fusion
• contrastive alignment
• camera auxiliary branch
• cognitive adapter
• coordinate attention