Fast vehicle detection based on colored point cloud with bird's eye view representation.

Area of Science:

• Computer Vision
• Robotics
• Sensor Fusion

Background:

• Autonomous vehicles rely on RGB cameras and LiDAR for accurate detection, but early-level fusion methods struggle with modality differences.
• Existing fusion techniques often fail to leverage the complementary strengths of camera and LiDAR data effectively.

Purpose of the Study:

• To propose a simple and effective vehicle detection method using an early-fusion strategy.
• To enhance detection accuracy by fusing RGB camera data with LiDAR point clouds.

Main Methods:

• Developed a method that unifies augmented 7D colored point clouds into 2D Bird's-Eye View (BEV) grids.
• Employed a Feature Fusion (2F) detection module for multi-scale feature extraction from BEV images.
• Utilized cor-calibration to eliminate null point clouds and augmented LiDAR data with color information.

Main Results:

• Fusing RGB images with point clouds significantly improved detection accuracy compared to using raw point clouds alone.
• The proposed method achieved a fast inference time of 0.05 seconds per frame.
• Validation on KITTI and NuScenes datasets demonstrated the method's effectiveness.

Conclusions:

• The proposed early-fusion strategy effectively addresses modality differences between RGB cameras and LiDAR.
• Unified 2D BEV grids and feature fusion offer a simple yet powerful approach for robust vehicle detection.
• The method provides a computationally efficient solution for real-time autonomous driving systems.