CPD(CommonsensePrototype-basedDetector) is a high-performance unsupervised 3D object detection framework. CPD first constructs Commonsense Prototype (CProto) characterized by high-quality bounding box and dense points, based on commonsense intuition. Subsequently, CPD refines the low-quality pseudo-labels...
CVPR 2024·Hai Wu,Shijia Zhao,Xun Huang,Chenglu Wen,Xin Li,Cheng Wang· The prevalent approaches of unsupervised 3D object detection follow cluster-based pseudo-label generation and iterative self-training processes. However, the challenge arises due to the sparsity of LiDAR scans, which leads to...
* 题目: UniBEV: Multi-modal 3D Object Detection with Uniform BEV Encoders for Robustness against Missing Sensor Modalities* PDF: arxiv.org/abs/2309.1451* 作者: Shiming Wang,Holger Caesar,Liangliang Nan,Julian F. P. Kooij* 其他: 6 pages, 5 figures 三维视觉-分割 2篇 * 题目: Addressing ...
The prevalent approaches of unsupervised 3D object detection follow cluster-based pseudo-label generation and iterative self-training processes. However, the challenge arises due to the sparsity of LiDAR scans, which leads to pseudo-labels with erroneous size and position, resulting in subpar detection...
1. Introduction 3D object detection is a fundamental task in various real-world scenarios, such as autonomous driving [24, 35] and robot navigation [29], aiming to detect and localize traffic-related objects such as cars, pedestrians, and cy- Corresponding author: W....
我们提出的融合方法实现了Waymo打开数据集,KITTI检测数据集和Kitti MOT数据集的每个对象深度估计的最新性能。我们进一步证明,通过简单地用融合增强的深度替换估计的深度,我们可以在单眼3D感知任务(包括检测和跟踪)方面取得重大改进。 * Delving into thePre-trainingParadigm of Monocular 3D Object Detection...
We present a new domain adaptive self-training pipeline, named ST3D, for unsupervised domain adaptation on 3D object detection from point clouds. First, we pre-train the 3D detector on the source domain with our proposed random object scaling strategy for mitigating the negative effects of source...
3D object detection from point clouds is crucial in safety-critical autonomous driving. Although many works have made great efforts and achieved significant progress on this task, most of them suffer from expensive annotation cost and poor transferability to unknown data due to the domain gap. Recen...
在VOC、COCO、Cityscapes和ImageNet上的大量实验表明,DetCo不仅在一系列2D和3D实例级检测任务上优于最近的方法,而且在图像分类上也具有竞争力。比如在ImageNet分类上,DetCo比InsLoc和DenseCL这两个当代专为物体检测而设计的作品,top-1准确率分别好了6.9%和5.0%。而且,在COCO检测上,DetCo比带SwAV和Mask R-CNN C4...
得益于全局图像和局部patch之间的多级监督和对比学习设计,DetCo在不牺牲图像分类的情况下,成功地提高了目标检测的传输能力,与当代自我监督的同行相比。 在PASCAL VOC[15]、COCO[28]和Cityscapes[6]上进行的大量实验表明,当转移到一系列2D和3D实例级检测任务时,DetCo的性能优于以前的最先进方法 ...