计算机视觉--视觉3D数据库

Posted on | In , , , |

[TOC]

计算机视觉领域

https://www.paperswithcode.com/area/computer-vision

UMD Faces 面部数据集

UMD Faces Dataset

UMD Faces Dataset 是一个面部数据集，主要用于身份鉴定研究，它拥有 8501 个主题共计 367,920 个面孔。该数据集分为静止图像和视频帧两部分，其中静止图像包含 367,888 张图，共计 8277 个主题；视频帧则包含 22,000 个主题视频，共计 370 万个带注释的视频帧。

https://hyper.ai/datasets/5537

3D Face

300W

300 Faces In-the-Wild Challenge (300-W)

官网介绍：https://ibug.doc.ic.ac.uk/resources/300-W/

中文介绍：https://blog.csdn.net/lgh0824/article/details/88536215

3DDFA的生成数据

http://www.cbsr.ia.ac.cn/users/xiangyuzhu/projects/3DDFA/main.htm

300W-3D
300W-3D-Face

300W-LP 合成了300W的大姿态人脸图像。

1
2
3
[300W-3D]: https://drive.google.com/file/d/0B7OEHD3T4eCkRFRPSXdFWEhRdlE/view?usp=sharing 
[300W-3D-Face]: https://drive.google.com/file/d/0B7OEHD3T4eCkZmgzUWZfd2FVVWs/view?usp=sharing 
[300W-LP]: https://drive.google.com/file/d/0B7OEHD3T4eCkVGs0TkhUWFN6N1k/view?usp=sharing

AFLW2000-3D

  • 项目：3DDFA

    AFLW2000-3D由AFLW数据库的前2000张图片及其三维信息组成。三维信息由3DMM重建（Blanz et.al A morphable model for the synthesis of 3d faces, SIGGRAPH’99）得到，并且包含68个特征点的三维信息。该数据库的三维数据精准度存在争议。
    ————————————————

1
2
3
/media/simon/新加卷1/dataset/3d_face/AFLW2000/
image00040.jpg
image00040.mat
1
2
3
4
5
6
7
8
9
10
11
12
具体包含以下内容：

1）pt2d：21个二维点
2）Illum_Para：1×10 光照参数
3）Color_Para：1×7 颜色参数
4）Tex_Para： 199×1 纹理参数
5）Shape Para： 199×1 形状参数
6）Exp_Para： 29×1 表情参数
7）Pose： 1×7 姿态参数，分别为：pitch， yaw， roll， translation(dx，dy，dz)，scale
8）pt3d_68： 3×68 三维特征点
————————————————
原文链接：https://blog.csdn.net/AuntieLee/article/details/105940291

image00040image00040.mat

3D Body

Human 3.6M

  • 常用于3D人体姿态估计

  • 11个人

Diversity and Size

  • • 3.6 million 3D human poses and corresponding images
  • • 11 professional actors (6 male, 5 female)
  • • 17 scenarios (discussion, smoking, taking photo, talking on the phone…)

Accurate Capture and Synchronization

  • • High-resolution 50Hz video from 4 calibrated cameras
  • • Accurate 3D joint positions and joint angles from high-speed motion capture system
  • • Pixel-level 24 body part labels for each configuration
  • • Time-of-flight range data
  • • 3D laser scans of the actors
  • • Accurate background subtraction, person bounding boxes

Support for Development

  • • Precomputed image descriptors
  • • Software for visualization and discriminative human pose prediction
  • • Performance evaluation on withheld test set

FAUST 2014

  • 300个2D扫描人体数据
数据库 网址 header 2
FAUST http://faust.is.tue.mpg.de/ A data set containing 300 real, high-resolution human scans, with automatically computed ground-truth correspondences. 一个包含300个真实，高分辨率人体扫描的数据集，具有自动计算的地面实况对应关系（Max Planck Tubingen）
Dynamic FAUST 3D人体动态数据库, 提供40,000个原始网格和对齐网格的数据集

SAESAR

SMPL使用的数据集

CAESAR http://store.sae.org/caesar/ 美国和欧洲的表面人体测量资源项目
10,000美元，2,400名男性和女性

Dyma

http://dyna.is.tue.mpg.de/

使用十个对象的40,000多次扫描

TOSCA

dataset contains synthetic meshes of fixed topology with artist-defined deformations.

数据集包含具有艺术家定义的变形的固定拓扑的合成网格。

synthetic dataset that is widely used for evaluation of mesh registration methods. It provides 80 artificially created meshes of animals and people (with 3 subjects in a dozen different poses each).

SHREC

为TOSCA添加了各种⼈造噪声⽹格，

SCAPE

Buff

http://buff.is.tue.mpg.de/

Given static 3D scans or 3D scan sequences (in pink), we estimate the naked shape under clothing (beige)

给定3D扫描序列，我们的模型评估不穿衣服的人体。

includes high resolution 3D scan sequences of 3 males and 3 females in different clothing styles.

3男3女的高分辨率的穿不同衣服的数据集。

参考资料：

开源数据库统计：

https://blog.csdn.net/m0_37570854/article/details/88736189

3D人体重建技术：

https://graphics.soe.ucsc.edu/data/BodyModels/index.html

  • SCAPE - (paper website (data) - The correspondence between the meshes above is identical to that in SCAPE
  • FAUST - (website-data-code) - different people in different poses
  • BodyLabs - (web) - Company commercializing body model tools
  • MPI - (website-data-code) - different people in different poses from a different MPI lab
  • MPII Human Shape (website-data-code) - CAESAE dataset processed by MPI
  • CAESAR - (website-data) - Sells complete CAESAR dataset with more measurements and authorized to provide commercial license.
  • other - (web) - Other data I find around my lab that might be helpful to someone

Web Face Recognition Training Datasets (Updating)

Github-DatasetZoo

CASIA-Webface (10K ids/0.5M images) [1]

baidu

dropbox

CelebA (10K ids/0.2M images) [2]

UMDFace (8K ids/0.37M images) [3]

baidu

dropbox

VGG2 (9K ids/3.31M images) [4]

baidu

dropbox

VGG2-Face HD()

90G

MS1M-IBUG (85K ids/3.8M images) [5,6]

baidu

dropbox

MS1M-ArcFace (85K ids/5.8M images) [5,7] (Recommend)

baidu

dropbox

Asian-Celeb (94K ids/2.8M images)[8] (Recommend)

baidu faces_glintasia.zip

dropbox faces_glintasia.zip

DeepGlint (181K ids/6.75M images) [8] (Recommend)

baidu

dropbox

IMDB-Face (59K ids/1.7M images) [9]

Celeb500k (500K ids/50M images) [10]

MegaFace (672K ids/4.7M images) [11]

Face Recognition Validation Datasets

CFP-FP (500 ids/7K images/7K pairs)[12]

AgeDB-30 (570 ids/12,240 images/6K pairs)[13,6]

LFW (5749 ids/13233 images/6K pairs)[14]

CALFW (5749 ids/13233 images/6K pairs)[15]

CPLFW (5749 ids/13233 images/6K pairs)[16]

Face Recognition Image Test Datasets

MegaFace

IJB (IJB-B, IJB-C)

TrillionPairs

NIST

Face Recognition Video Test Datasets

YTF

IQIYI

Reference

[1] Dong Yi, Zhen Lei, Shengcai Liao, Stan Z. Li. Learning Face Representation from Scratch. arXiv:1411.7923, 2014.

[2] Ziwei Liu, Ping Luo, Xiaogang Wang, Xiaoou Tang. Deep Learning Face Attributes in the Wild, ICCV, 2015.

[3] Bansal Ankan, Nanduri Anirudh, Castillo Carlos D, Ranjan Rajeev, Chellappa, Rama. UMDFaces: An Annotated Face Dataset for Training Deep Networks, arXiv:1611.01484v2, 2016.

[4] Qiong Cao, Li Shen, Weidi Xie, Omkar M. Parkhi, Andrew Zisserman. VGGFace2: A dataset for recognising faces across pose and age. FG, 2018.

[5] Yandong Guo, Lei Zhang, Yuxiao Hu, Xiaodong He, Jianfeng Gao. Ms-celeb-1m: A dataset and benchmark for large-scale face recognition. ECCV, 2016.

[6] Jiankang Deng, Yuxiang Zhou, Stefanos Zafeiriou. Marginal loss for deep face recognition, CVPRW, 2017.

[7] Jiankang Deng, Jia Guo, Stefanos Zafeiriou. Arcface: Additive angular margin loss for deep face recognition, arXiv:1801.07698, 2018.

[8] http://trillionpairs.deepglint.com/

[9] Wang Fei, Chen Liren, Li Cheng, Huang Shiyao, Chen Yanjie, Qian Chen, Loy, Chen Change. The Devil of Face Recognition is in the Noise, ECCV, 2018.

[10] Cao Jiajiong, Li Yingming, Zhang Zhongfei, Celeb-500K: A Large Training Dataset for Face Recognition, ICIP, 2018.

[11] Nech Aaron, Kemelmacher-Shlizerman Ira, Level Playing Field For Million Scale Face Recognition, CVPR, 2017.

[12] Sengupta Soumyadip, Chen Jun-Cheng, Castillo Carlos, Patel Vishal M, Chellappa Rama, Jacobs David W, Frontal to profile face verification in the wild, WACV, 2016.

[13] Moschoglou, Stylianos and Papaioannou, Athanasios and Sagonas, Christos and Deng, Jiankang and Kotsia, Irene and Zafeiriou, Stefanos, Agedb: the first manually collected, in-the-wild age database, CVPRW, 2017.

[14] Gary B. Huang, Manu Ramesh, Tamara Berg, and Erik Learned-Miller. Labeled Faces in the Wild: A Database for Studying Face Recognition in Unconstrained Environments, 2007.

[15] Zheng Tianyue, Deng Weihong, Hu Jiani, Cross-age lfw: A database for studying cross-age face recognition in unconstrained environments, arXiv:1708.08197, 2017.

[16] Zheng, Tianyue, and Weihong Deng. Cross-Pose LFW: A Database for Studying Cross-Pose Face Recognition in Unconstrained Environments, 2018.