论文阅读 [TPAMI-2022] Disentangling Monocular 3D Object Detection: From Single to Multi-Class Recognitio

论文阅读 [TPAMI-2022] Disentangling Monocular 3D Object Detection: From Single to Multi-Class Recognition

论文搜索(studyai.com)

搜索论文: Disentangling Monocular 3D Object Detection: From Single to Multi-Class Recognition

搜索论文: http://www.studyai.com/search/whole-site/?q=Disentangling+Monocular+3D+Object+Detection:+From+Single+to+Multi-Class+Recognition

关键字(Keywords)

Three-dimensional displays; Two dimensional displays; Measurement; Shape; Object detection; Feature extraction; Estimation; Computer vision; object recognition; vision and scene understanding; 3D/Stereo scene analysis; object detection

机器学习; 机器视觉

自监督学习; 检测分割; 场景解析

摘要(Abstract)

In this paper we introduce a method for multi-class, monocular 3D object detection from a single RGB image, which exploits a novel disentangling transformation and a novel, self-supervised confidence estimation method for predicted 3D bounding boxes.

在本文中，我们介绍了一种从单个RGB图像中检测多类单目3D目标的方法，该方法利用了一种新的解纠缠变换和一种新的用于预测3D边界盒的自监督置信度估计方法。.

The proposed disentangling transformation isolates the contribution made by different groups of parameters to a given loss, without changing its nature.

提出的解纠缠变换分离了不同参数组对给定损耗的贡献，而不改变其性质。.

This brings two advantages: i) it simplifies the training dynamics in the presence of losses with complex interactions of parameters; and ii) it allows us to avoid the issue of balancing independent regression terms.

这带来了两个好处：i）它简化了存在损耗时的训练动态，并具有复杂的参数交互作用；ii）它允许我们避免平衡独立回归项的问题。.

We further apply this disentangling transformation to another novel, signed Intersection-over-Union criterion-driven loss for improving 2D detection results.

我们进一步将这种解纠缠变换应用于另一种新的、基于联合准则的符号交叉驱动的损失，以改善二维检测结果。.

We also critically review the AP metric used in KITTI3D and resolve a flaw which affected and biased all previously published results on monocular 3D detection.

我们还严格审查了KITTI3D中使用的AP指标，并解决了一个缺陷，该缺陷影响并偏向了之前发布的所有单目3D检测结果。.

Our improved metric is now used as official KITTI3D metric.

我们改进的度量现在被用作官方的KITTI3D度量。.

We provide extensive experimental evaluations and ablation studies on the KITTI3D and nuScenes datasets, setting new state-of-the-art results.

我们对KITTI3D和nuScenes数据集进行了广泛的实验评估和消融研究，得出了最新的结果。.

We provide additional results on all the classes of KITTI3D as well as nuScenes datasets to further validate the robustness of our method, demonstrating its ability to generalize for different types of objects…

我们在KITTI3D的所有类别以及nuScenes数据集上提供了额外的结果，以进一步验证我们的方法的稳健性，展示其对不同类型对象的概括能力。。.

作者(Authors)

[‘Andrea Simonelli’, ‘Samuel Rota Bulò’, ‘Lorenzo Porzi’, ‘Manuel López Antequera’, ‘Peter Kontschieder’]