The paper presents AVOD (Aggregate View Object Detection), a neural network architecture designed for autonomous driving scenarios. AVOD combines LIDAR point clouds and RGB images to generate features that are shared by a region proposal network (RPN) and a second-stage detection network. The RPN uses a novel architecture capable of multimodal feature fusion to generate reliable 3D object proposals for multiple classes in road scenes. The second-stage detection network performs accurate oriented 3D bounding box regression and category classification. The proposed architecture achieves state-of-the-art results on the KITTI 3D object detection benchmark while running in real-time with a low memory footprint, making it suitable for deployment on autonomous vehicles. The paper also discusses related work, including hand-crafted features, single-shot detectors, monocular-based methods, and 3D region proposal networks. The experimental results show that AVOD outperforms existing methods in terms of 3D proposal recall, 3D object detection accuracy, and computational efficiency.The paper presents AVOD (Aggregate View Object Detection), a neural network architecture designed for autonomous driving scenarios. AVOD combines LIDAR point clouds and RGB images to generate features that are shared by a region proposal network (RPN) and a second-stage detection network. The RPN uses a novel architecture capable of multimodal feature fusion to generate reliable 3D object proposals for multiple classes in road scenes. The second-stage detection network performs accurate oriented 3D bounding box regression and category classification. The proposed architecture achieves state-of-the-art results on the KITTI 3D object detection benchmark while running in real-time with a low memory footprint, making it suitable for deployment on autonomous vehicles. The paper also discusses related work, including hand-crafted features, single-shot detectors, monocular-based methods, and 3D region proposal networks. The experimental results show that AVOD outperforms existing methods in terms of 3D proposal recall, 3D object detection accuracy, and computational efficiency.