FCAF3D: Fully Convolutional Anchor-Free 3D Object Detection

TL;DR

FCAF3D introduces a fully convolutional, anchor-free 3D object detection method using voxel-based sparse convolutions, achieving state-of-the-art results on multiple indoor scene datasets without relying on prior object shape assumptions.

Contribution

It presents a novel, fully convolutional, anchor-free approach with a new parametrization of oriented bounding boxes for improved generalization in 3D detection.

Findings

Achieves state-of-the-art mAP@0.5 on ScanNet V2, SUN RGB-D, and S3DIS datasets.

Handles large-scale scenes efficiently with a single pass.

Outperforms previous methods by significant margins in accuracy.

Abstract

Recently, promising applications in robotics and augmented reality have attracted considerable attention to 3D object detection from point clouds. In this paper, we present FCAF3D - a first-in-class fully convolutional anchor-free indoor 3D object detection method. It is a simple yet effective method that uses a voxel representation of a point cloud and processes voxels with sparse convolutions. FCAF3D can handle large-scale scenes with minimal runtime through a single fully convolutional feed-forward pass. Existing 3D object detection methods make prior assumptions on the geometry of objects, and we argue that it limits their generalization ability. To get rid of any prior assumptions, we propose a novel parametrization of oriented bounding boxes that allows obtaining better results in a purely data-driven way. The proposed method achieves state-of-the-art 3D object detection results in terms of mAP@0.5 on ScanNet V2 (+4.5), SUN RGB-D (+3.5), and S3DIS (+20.5) datasets. The code and models are available at https://github.com/samsunglabs/fcaf3d.