人脸识别数据集 Face Databases

本文介绍了多个用于面部识别研究的人脸数据库,包括ARFaceDatabase、Richard's MIT database等,涵盖了不同条件下的图像数量、颜色、尺寸及获取方式等关键信息。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

Face Databases

AR Face Database
Richard's MIT database
CVL Database
The Psychological Image Collection at Stirling
Labeled Faces in the Wild
The MUCT Face Database
The Yale Face Database B
The Yale Face Database 
PIE Database
The UMIST Face Database 
Olivetti - Att - ORL 
The Japanese Female Facial Expression (JAFFE) Database
The Human Scan Database
The University of Oulu Physics-Based Face Database 
XM2VTSDB 


Databases with over 100 unique individuals in them

Name:AR Face Database 
Color Images:Yes 
Image Size:576 x 768 
Number of unique people:126; 70 Male, 56 Female 
Number of pictures per person:26 
Different Conditions:All frontal views of: neutral expression, smile, anger, scream, left light on, right light on, all sides lights on, wearing sun glasses, wearing sun glassses and left light on, wearing sun glasses and right light on, wearing scarf, wearing scarf and left light on, wearing scarf and right light on; second sessions repeated same conditions. 
Citation reference:A.M. Martinez and R. Benavente. The AR Face Database. CVC Technical Report #24, June 1998 
Available:Yes 



Name:Richard's MIT database 
Color Images:Yes 
Image Size:480 x 640 
Number of unique people:154; 82 Male, 74 Female 
Number of pictures per person:6 
Different Conditions:2 frontal, profile, and 3/4 view - also some additional pictures of students wearing makeup
Available:Yes 



Name:CVL Database 
Color Images:Yes 
Image Size:640 x 480 
Number of unique people:114; 108 Male, 6 Female 
Number of pictures per person:7 
Different Conditions:profile left/right, 45 degrees left/right, frontal, frontal smile, frontal smile with teeth 
Citation reference:Franc Solina, Peter Peer, Borut Batagelj, Samo Juvan, Jure Kovac, "Color-based face detection in the "15 seconds of fame" art installation", In: Mirage 2003, Conference on Computer Vision / Computer Graphics Collaboration for Model-based Imaging, Rendering, image Analysis and Graphical special Effects, March 10-11 2003, INRIA Rocquencourt, France, Wilfried Philips, Rocquencourt, INRIA, 2003, pp. 38-47. 
Available :Yes 



Name:The Psychological Image Collection at Stirling (PICS) 
Note:The Psychological Image Collection at Stirling contains a number of databases, below are listed the four largest ones. 
Citation reference:Not sure - contact Peter Hancock (pjbh1@stir.ac.uk) to find out who to reference 
Available :Yes 



Name:notingham scans 
Color Images:No 
Image Size:438 x 538 
Number of unique people:100; 50 Male, 50 Female 
Number of pictures per person:1 
Different Conditions:All frontal views 



Name:aberdeen 
Color Images:Yes 
Image Size:432 x 528 
Number of unique people:116 
Number of pictures per person:Ranges from 1 to around 18, there is a total of 690 image in the database 
Different Conditions:All frontal view, two main differences in lighting and some expression variation 



Name:nott-faces-originals 
Color Images:No 
Image Size:288 x 384 
Number of unique people:around 100 - almost all male 
Number of pictures per person:7 
Different Conditions:Mostly frontal, a few profile 



Name:notingham scans 
Color Images:No 
Image Size:438 x 538 
Number of unique people:100; 50 Male, 50 Female 
Number of pictures per person:1 
Different Conditions:All frontal views 



Name:Stirling_faces 
Color Images:No 
Image Size:280 x 365 
Number of unique people:36; 18 Males, 18 Females 
Number of pictures per person:~9 
Different Conditions:Frontal and profile views 



Name:Labeled Faces in the Wild 
Color Images:Yes 
Image Size:150 x 150 
Number of unique people:13,233 
Number of pictures per person:Varies, but 1,680 individuals with more than 1 image 
Different Conditions:Un-posed photos, but mainly frontal views 
Citation reference: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. University of Massachusetts, Amherst, Technical Report 07-49, October, 2007. 
Available :Free download from website 



Name:The MUCT Face Database 
Color Images:Yes 
Image Size:480 x 640 
Number of unique people:276 
Number of pictures per person:10 to 15 
Different Conditions:frontal and three-quarter views, different lighting sets, manual landmarks 
Citation reference:S. Milborrow, J. Morkel, and F. Nicolls. The MUCT Landmarked Face Database. PRASA. 2010. 
Available :Yes 



Databases with less than 100 unique individuals in them


Name:The Yale Face Database B 
Color Images:No 
Image Size:640 x 480 
Number of unique people:10 
Number of pictures per person:576 
Different Conditions:9 poses x 64 illumination conditions 
Citation reference:Georghiades, A.S. and Belhumeur, P.N. and Kriegman, D.J. From Few to Many: Illumination Cone Models for Face Recognition under Variable Lighting and Pose. IEEE Trans. Pattern Anal. Mach. Intelligence 23(6):643-660 (2001). 
Available :Yes - facial marker points also available 


Name:The Yale Face Database 
Color Images:No 
Image Size:320 x 243 
Number of unique people:15 
Number of pictures per person:11 
Different Conditions:center-light, w/glasses, happy, left-light, w/no glasses, normal, right-light, sad, sleepy, surprised, wink 
Citation reference:P. N. Bellhumer, J. Hespanha, and D. Kriegman. Eigenfaces vs. fisherfaces: Recognition using class specific linear projection. IEEE Transactions on Pattern Analysis and Machine Intelligence, Special Issue on Face Recognition, 17(7):711--720, 1997. 
Available :Yes 


Name:PIE Database 
Color Images:Yes 
Image Size:640 x 486 
Number of unique people:68 
Number of pictures per person:A lot (40GB total) - we have a subset of one picture per individual 
Different Conditions:13 different poses, 43 different illumination conditions, and with 4 different expressions. 
Citation reference:The CMU Pose, Illumination, and Expression Database T. Sim, S. Baker, and M. Bsat IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, No. 12, December, 2003 
Available :Subset (one picture of each individual); to get the whole database (40GB) send a hard drive to simonb@cs.cmu.edu 


Name:The UMIST Face Database 
Color Images:No 
Image Size:220 x 220 
Number of unique people:20 
Number of pictures per person:From around 19 to 36 
Different Conditions:Various angles from left profile to right profile. 
Citation reference:TCharacterizing Virtual Eigensignatures for General Purpose Face Recognition. Daniel B Graham and Nigel M Allinson. Face Recognition: From Theory to Applications, NATO ASI Series F, Computer and Systems Sciences, Vol. 163. H. Wechsler, P. J. Phillips, V. Bruce, F. Fogelman-Soulie and T. S. Huang (eds), pp 446-456, 1998. 
Available :Yes 


Name:Olivetti - Att - ORL 
Color Images:No 
Image Size:92 x 112 
Number of unique people:40 
Number of pictures per person:10 
Different Conditions:All frontal and slight tilt of the head 
Citation reference:Ferdinando Samaria, Andy Harter. Parameterisation of a Stochastic Model for Human Face Identification. Proceedings of 2nd IEEE Workshop on Applications of Computer Vision, Sarasota FL, December 1994 
Available :Yes 


Name:The Japanese Female Facial Expression (JAFFE) Database 
Color Images:No 
Image Size:256 x 256 
Number of unique people:10 
Number of pictures per person:7 
Different Conditions:7 different emotional facial expressions 
Citation reference:Coding Facial Expressions with Gabor Wavelets Michael J. Lyons, Shigeru Akamatsu, Miyuki Kamachi & Jiro Gyoba Proceedings, Third IEEE International Conference on Automatic Face and Gesture Recognition, April 14-16 1998, Nara Japan, IEEE Computer Society, pp. 200-205. 
Available :Yes 


Name:The Human Scan Database 
Color Images:No 
Image Size:384 x 286 
Number of unique people:23 
Number of pictures per person:around 66 on average 
Different Conditions:mainly frontal views, a bunch random different photos of different people 
Citation reference:O. Jesorsky, K. Kirchberg, R. Frischholz.Robust Face Detection Using the Hausdorff Distance. In J. Bigun and F. Smeraldi, editors, Audio and Video based Person Authentication - AVBPA 2001, pages 90-95. Springer, 2001. 
Available :Yes - facial points also available 




Databases that cost money


Name:The University of Oulu Physics-Based Face Database 
Color Images:Yes 
Image Size:428 x 569 
Number of unique people:125 
Number of pictures per person:16 
Different Conditions:All frontal images: 16 different camera calibration and illuminations 
Citation reference:E. Marszalec, B. Martinkauppi, M. Soriano, M. Pietik鋓nen (2000), 'A physics-based face database for color research', Journal of Electronic Imaging Vol. 9 No. 1 pp. 32-38. 
Available :No - Cost $50, email Prof. Matti Pietikainen (mkp@ee.oulu.fi) 


Name:XM2VTSDB 
Color Images:Yes 
Image Size:576 x 720 
Number of unique people:295 
Number of pictures per person:Depends on database purchased - 4 in frontal database; 4 of both left/right profile in profile database 
Different Conditions:frontal, profile, VRML modles, speech, and more, see website 
Citation reference:K Messer, J Matas, J Kittler, J Luettin, and G maitre. Xm2vtsdb: The extended m2vts database. In Second International Conference of Audio and Video-based Biometric Person Authentication, March 1999 
Available :No: costs a fair amount to purchase - Frontal Image Set = $153, Side Profile Image Set = $229.5 



Other useful links

Computer Vision Test Images  - a page with links to other sites with more images 

Irfan View - a great image viewer that has useful conversion and renaming utilities 


from: http://web.mit.edu/emeyers/www/face_databases.html#ar

### 字符识别数据集下载 对于训练或测试字符识别模型的需求,可以考虑以下几个常见的OCR数据集: #### 常见的 OCR 数据集 1. **IAM Handwriting Database**: 这是一个广泛使用的手写文本数据集,包含了大量分割好的单词和句子图像。该数据集适用于离线 handwriting recognition 的研究工作[^1]。 2. **MNIST Dataset**: 虽然 MNIST 主要针对数字分类任务设计,但它也可以作为一个简单的字符识别入门级数据集。它由 70,000 张灰度的手写数字图片组成,每张大小为 28×28 像素[^3]。 3. **EMNIST Dataset**: EMNIST 是基于 MNIST 扩展而来的更大规模的数据集,不仅包含数字还增加了大写字母、小写字母等多种类别标签,非常适合多类别的字符识别实验。 4. **IIIT-HWS (Indian Institute of Technology Hyderabad Word Segmentation)**: 提供了丰富的印地语和其他印度本地语言文字样本集合,同时也支持英语印刷体和平面扫描文档上的应用案例分析. 5. **Multi-language dataset (IC19)**: 来自论文《Revisiting Scene Text Recognition: A Data Perspective》(ICCV, 2023),这个多语言场景文本识别数据集中涵盖了多种自然环境下的复杂背景条件下的真实世界照片素材,特别适合跨文化适应性的算法验证需求[^2]. 6. **SynthText**: SynthText 是一种合成方法生成的大规模英文学术文章标题样式字符串库,通过渲染技术模拟不同光照效果、字体风格变化等因素构建而成,极大促进了深度学习框架下端到端解决方案的发展进程. 这些资源都可以在网络上找到公开版本进行免费获取或者申请授权访问权限后合法利用于科研项目当中。具体链接地址可能随时间有所调整,请自行搜索最新发布页面确认详情信息。 ```python import requests url = 'http://www.fki.inf.unibe.ch/databases/iam-handwriting-database' response = requests.get(url) if response.status_code == 200: print('Successfully accessed IAM Handwriting Database.') else: print(f'Failed to access database with status code {response.status_code}.') ``` 上述代码片段展示了一个简单的方法去尝试连接至某个特定在线数据库网站实例——这里以 IAM Handwriting Database为例说明如何编写一段基础脚本来检测目标网址是否可达。
评论 2
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值