AIST_plus_plus

displayName: AIST++
labelTypes: []
license:
- CC BY 4.0
mediaTypes: []
paperUrl: https://arxiv.org/pdf/2101.08779.pdf
publishDate: "2021"
publishUrl: https://google.github.io/aistplusplus_dataset/
publisher:
- University of California
- Google Research
- University of Southern California
tags:
- Dance movement
taskTypes:
- 3D Human Pose Estimation

---
# 数据集介绍
## 简介
AIST++ 是一个 3D 舞蹈数据集，其中包含从真实舞者与音乐配对重建的 3D 动作。 AIST++ 舞蹈动作数据集是从 AIST 舞蹈视频数据库构建的。对于多视图视频，设计了一个精细的管道来估计相机参数、3D 人体关键点和 3D 人体舞蹈动作序列：它为 1010 万张图像提供 3D 人体关键点注释和相机参数，涵盖 9 个视图中的 30 个不同主题。这些属性使其成为具有 3D 人体关键点注释的最大和最丰富的现有数据集。它还包含 1,408 个 3D 人类舞蹈动作序列，表示为关节旋转以及根轨迹。舞蹈动作平均分布在 10 种舞蹈流派中，有数百种编舞。运动持续时间从 7.4 秒不等。至 48.0 秒。所有的舞蹈动作都有相应的音乐。

## Download dataset
:modelscope-code[]{type="git"}

displayName: AIST++
labelTypes: []
license:
- CC BY 4.0
mediaTypes: []
paperUrl: https://arxiv.org/pdf/2101.08779.pdf
publishDate: "2021"
publishUrl: https://google.github.io/aistplusplus_dataset/
publisher:
- University of California
- Google Research
- University of Southern California
tags:
- Dance movement
taskTypes:
- 3D Human Pose Estimation

---

# Dataset Introduction
## Overview
AIST++ is a 3D dance dataset containing 3D motions reconstructed from real dancers paired with music. The AIST++ dance motion dataset is constructed from the AIST Dance Video Database. For multi-view videos, a sophisticated pipeline was designed to estimate camera parameters, 3D human keypoints and 3D human dance motion sequences: it provides 3D human keypoint annotations and camera parameters for 10.1 million images, covering 30 distinct subjects across 9 views. These attributes make it the largest and most abundant existing dataset with 3D human keypoint annotations. It also contains 1,408 3D human dance motion sequences, represented as joint rotations and root trajectories. Dance motions are evenly distributed across 10 dance genres, with hundreds of choreographies. The duration of motions ranges from 7.4 seconds to 48.0 seconds. All dance motions have corresponding music.

## Download Dataset
:modelscope-code[]{type="git"}

搜集汇总

背景与挑战

背景概述

AIST_plus_plus是一个包含3D舞蹈动作和音乐配对的数据集，提供10.1百万张图像的3D人体关键点注释和相机参数，涵盖30名不同舞者和9种视角。数据集包含1,408个3D舞蹈动作序列，均匀分布在10种舞蹈风格中，每个动作时长从7.4秒到48.0秒不等，并配有相应音乐。