OG RGB+D

OG RGB+D是一个全新的步态识别数据库，称为OG RGB+D数据库，它突破了其他步态数据库的限制，通过多重同步的方式包含了各种遮挡（自遮挡、主动遮挡、被动遮挡）的多模态步态数据。 Azure Kinect DK 传感器数据采集系统（多 Kinect SDAS），也可应用于安全情况。因为 Azure Kinect DK 可以同时采集多模态数据，支持不同类型的步态识别算法，尤其是能够有效获取以相机为中心的多人 3D 姿态，多视角比单视角更好地处理遮挡。特别是，OG RGB+D 数据库通过融合多 Kinect 收集的数据，提供准确的轮廓和优化的人体 3D 关节数据 (OJ)，在遮挡下人体姿势表示更加准确。描述来自：Multi-Kinect SDAS 收集的遮挡下步态识别的基准OG RGB+D is a novel gait recognition database, termed the OG RGB+D Database. It breaks through the limitations of existing gait databases by collecting multi-modal gait data covering various occlusion scenarios including self-occlusion, active occlusion and passive occlusion via multiple synchronous acquisition methods. The Azure Kinect DK-based sensor data acquisition system (Multi-Kinect SDAS) can also be applied to security scenarios. Since Azure Kinect DK can simultaneously collect multi-modal data and support different types of gait recognition algorithms, it is particularly effective in acquiring camera-centric multi-person 3D poses, and multi-view settings can handle occlusions better than single-view settings. Notably, the OG RGB+D database fuses data collected by multiple Kinect sensors to provide accurate silhouettes and optimized human 3D joint data (OJ), enabling more accurate human pose representation under occlusion conditions. The description is sourced from: Benchmark for Gait Recognition Under Occlusions Collected by Multi-Kinect SDAS

搜集汇总

背景与挑战

背景概述

OG RGB+D是一个步态识别数据库，通过多重同步采集多模态数据，包含各种遮挡情况，利用Azure Kinect DK传感器实现多视角处理，提供准确的轮廓和3D关节数据。该数据集由西北工业大学于2021年发布，旨在为遮挡下的步态识别提供基准。