EPNet++: Cascade Bi-Directional Fusion for Multi-Modal 3D Object Detection.

Area of Science:

• Computer Vision
• Robotics
• Sensor Fusion

Background:

• 3D object detection is crucial for autonomous systems.
• LiDAR and camera data offer complementary information for improved detection.
• Current methods struggle with sparse data scenarios.

Purpose of the Study:

• To enhance multi-modal 3D object detection performance and robustness.
• To introduce a novel fusion module and loss function for better feature representation and confidence scoring.
• To evaluate the detector's effectiveness in sparse point cloud environments.

Main Methods:

• Proposed EPNet++ architecture incorporating a Cascade Bi-directional Fusion (CB-Fusion) module.
• Introduced a Multi-Modal Consistency (MC) loss to ensure score consistency between modalities.
• Conducted experiments on KITTI, JRDB, and SUN-RGBD datasets.

Main Results:

• EPNet++ significantly outperforms state-of-the-art methods on benchmark datasets.
• Demonstrated superior performance and robustness in highly sparse point cloud scenarios.
• Achieved remarkable margins in sparse cases, validating the approach's effectiveness.

Conclusions:

• The proposed CB-Fusion module and MC loss enhance feature representation and detection reliability.
• EPNet++ offers a robust solution for 3D object detection, particularly in challenging sparse environments.
• This work presents a promising direction for cost-effective LiDAR sensor utilization in autonomous driving.