论文阅读 [CVPR 2022]：一种基于统一查询范式的点云理解方法（A Unified Query-based Paradigm for Point Cloud Understanding）

论文阅读 【CVPR-2022】 A Unified Query-based Paradigm for Point Cloud Understanding

studyai.com

搜索论文: A Unified Query-based Paradigm for Point Cloud Understanding

摘要(Abstract)

3D point cloud understanding is an important component in autonomous driving and robotics.

三维点云理解是自动驾驶和 机器人技术 的重要组成部分。

In this paper  , we present a novel Embedding-Querying paradigm (EQ-Paradigm) for 3D understanding tasks including detection, segmentation and classification.

在本文中，我们提出了一种新的嵌入查询范式（EQ范式），用于3D理解任务，包括检测、分割和分类。

EQ-Paradigm is a unified paradigm that enables the combination of any existing 3D backbone   architectures with different task heads.

EQ范式是一种统一的范式，可以将任何现有的3D主干架构与不同的任务头结合起来。

Under the EQ-Paradigm, the input is firstly encoded in the embedding stage with an arbitrary feature extraction architecture, which is independent of tasks and heads.

在EQ范式下，输入首先在嵌入阶段进行编码，采用任意的特征提取架构，该架构独立于任务和头部。

Then, the querying stage enables the encoded features to be applicable for diverse task heads. This is achieved by introducing an intermediate representation, i.e., Q-representation, in the querying stage to serve as a bridge between the embedding stage and task heads. We design a novel Q-Net as the querying stage network.

然后，查询阶段使编码的特征适用于各种各样的任务头。这是通过在查询阶段引入一个中间表示（即Q表示）来实现的，该中间表示被作为嵌入阶段和任务头之间的桥梁。我们设计了一个新的Q-Net作为查询阶段网络。

Extensive experimental results on various 3D tasks show that EQ-Paradigm in tandem with Q-Net is a general and effective pipeline  , which enables a flexible collaboration of backbones and heads, and further boosts the performance of the state-of-the-art methods. All codes and models will be published soon.

在各种3D任务上的大量实验结果表明，EQ范式与Q-Net相结合是一种通用而有效的流程，可以实现主体和头部的灵活协作，并进一步提高最先进方法的性能。所有代码和模型将很快发布。

免责声明：本文系网络转载或改编，未找到原创作者，版权归原作者所有。如涉及版权，请联系删