EPNet with Self-Attention for Fast and Accurate 3D Object Detection

3D object detection is a critical task in autonomous driving that requires fast and accurate processing. In this paper, we propose a novel method that utilizes both LiDAR point clouds and RGB images to improve the processing speed and detection accuracy of 3D object detection. Specifically, we investigate existing method and incorporate an effective self-attention mechanism. Our experimental results demonstrate that the proposed method achieves a processing speed that is 5.03 times faster than existing method, while also improving the accuracy by 2.61%.