PillarNeSt: Embracing Backbone Scaling and Pretraining for Pillar-based 3D Object Detection
This work addresses the challenge of improving 3D object detection for autonomous driving by scaling and pretraining backbones, though it is incremental as it adapts existing 2D methods to point clouds.
The paper tackled the problem of pillar-based 3D object detectors not leveraging 2D backbone scaling and pretraining, showing that using dense ConvNets pretrained on large-scale image datasets as backbones significantly improves performance, with PillarNeSt outperforming existing detectors by a large margin on nuScenes and Argoversev2 datasets.
This paper shows the effectiveness of 2D backbone scaling and pretraining for pillar-based 3D object detectors. Pillar-based methods mainly employ randomly initialized 2D convolution neural network (ConvNet) for feature extraction and fail to enjoy the benefits from the backbone scaling and pretraining in the image domain. To show the scaling-up capacity in point clouds, we introduce the dense ConvNet pretrained on large-scale image datasets (e.g., ImageNet) as the 2D backbone of pillar-based detectors. The ConvNets are adaptively designed based on the model size according to the specific features of point clouds, such as sparsity and irregularity. Equipped with the pretrained ConvNets, our proposed pillar-based detector, termed PillarNeSt, outperforms the existing 3D object detectors by a large margin on the nuScenes and Argoversev2 datasets. Our code shall be released upon acceptance.