Manual-Label Free 3D Detection via An Open-Source Simulator
This addresses the high cost of labeled data for 3D detection in autonomous driving, offering a promising alternative, though it is incremental as it builds on existing simulators and domain adaptation techniques.
The paper tackles the problem of expensive manual labeling for LiDAR-based 3D object detection by proposing a manual-label free method using synthetic data from the CARLA simulator and a Domain Adaptive VoxelNet to bridge the distribution gap to real scenarios, achieving 76.66% mAP on BEV mode and 56.64% mAP on 3D mode on the KITTI evaluation set.
LiDAR based 3D object detectors typically need a large amount of detailed-labeled point cloud data for training, but these detailed labels are commonly expensive to acquire. In this paper, we propose a manual-label free 3D detection algorithm that leverages the CARLA simulator to generate a large amount of self-labeled training samples and introduces a novel Domain Adaptive VoxelNet (DA-VoxelNet) that can cross the distribution gap from the synthetic data to the real scenario. The self-labeled training samples are generated by a set of high quality 3D models embedded in a CARLA simulator and a proposed LiDAR-guided sampling algorithm. Then a DA-VoxelNet that integrates both a sample-level DA module and an anchor-level DA module is proposed to enable the detector trained by the synthetic data to adapt to real scenario. Experimental results show that the proposed unsupervised DA 3D detector on KITTI evaluation set can achieve 76.66% and 56.64% mAP on BEV mode and 3D mode respectively. The results reveal a promising perspective of training a LIDAR-based 3D detector without any hand-tagged label.