Abstract—Face recognition has witnessed a good deal of awareness because of its various applications in several fields like computer vision
Abstract—Face recognition has witnessed a good deal of awareness because of its various applications in several fields like computer vision, security, pattern recognition and computer graphics, however still is a challenging and active research area. This paper has presented an extensive review of face databases for constrained and unconstrained Environments. Face databases are used for the face detection and recognition algorithm testing and they have been designed to analyze the effectiveness of recognition of face algorithms. The paper has focused mostly on novel databases that are freely available for the research purposes. Most of the popular face databases are briefly introduced and compared.
Keywords- face recognition, face database, expression, occlusion.
INTRODUCTION
Over the last few years analysis in recognition of face has moved from 2D to 3D. The demand for 3D face data has caused the need of 3D databases. The paper initially gives an introduction of publicly present 2D and 3D face databases for constrained and unconstrained Environments. The presence of different databases requires a quantitative analysis of those databases so as to compare more precisely the executions of the different algorithms present in literature 8, 9, 10, 14, 15. The development of algorithms robust to illumination, pose, facial expression, age, occlusion changes demands databases of adequate size that involves cautiously controlled deviations of these features. Also, typical databases are required to relatively analyze techniques.
Presently several existing databases utilized for recognition of faces that contrast in lighting conditions, size, pose, expressions and occlusions. The previous facial databases mainly consists of frontal pictures, like the local set of data set collected from 115 people. Nowadays, the facial databases were observed to conquer the deviations in pose, illuminations, imaging directions, ethnicity, gender and facial expressions. Few most latest databases conquer the deviations in picture sizes, compression, occlusions and are collected from different origins like social media and web 10, 11.
This paper have presented an extensive survey of face databases for constrained and unconstrained environments. Section II describes the overview of various face databases. It focuses mostly on novel databases that are freely available for research purposes. Section III describes some of the recent face databases. Section IV compares the various popular face databases. Finally, Section V concludes the paper.
FACE DATABASES
Over the last few decades, several face databases have been constructed to analyze the effectiveness of recognition of faces approaches. The brief introduction of selected databases is as follows. In most cases the link to database download is provided.
The AR database
This 12 is a databases which has practical occlusions and are open to the public. It has higher than four thousand color pictures of one hundred twenty six people faces (seventy male and fifty six female). The pictures endure from several changes in facial expressions, illuminations and occlusions (i.e., scarves and sunglasses). They were collected underneath perfectly controlled situations. There were no restrictions obligatory on wear of individuals (clothes, glasses, etc.), makeup, hair style, etc. For every person, 26 images were captured in two sessions (two weeks apart) 1.
The limitations of the AR database are that it only contains two types of occlusions, i.e., scarf and sunglasses, and the location of the occlusion is either on the upper face or lower face. This database can be downloaded from the link http://rvl1.ecn.purdue.edu/~aleix/aleix face DB.html 12.
The Extended Yale B database
It has two thousand four hundred fourteen frontal face pictures of thirty eight persons in sixty four different lighting conditions. It has for every person in a typical pose, a picture
with surrounding (background) lighting was too gathered. The pictures are classified into 4 subsets in line with the lighting angle in concern to the axis of camera. The set one and set a pair of cowl the angular range 0? to 25?, the set three extends from 25? to 50?, the Subset four covers 50? to 77?, and Subset 5 covers angles which are larger than 78?. In order to simulate various levels of contiguous occlusions, the most used scheme is to revive an indiscriminately placed square patch from every test picture with a catarrhine picture that has analogous texture with the human face. The position of the occlusion is indiscriminately chosen. The sizes of the synthetic occlusions vary in the range of 10% to 50% of the original image 2, 13.
The FRGC database
The well-known 3D expression databases are the “Face Recognition Grand Challenge” (FRGC) databases. It had a high footprint on the advancement of face recognition techniques. Therefore it’s conjointly thought of as the reference databases for substantiating the 3D recognition of face techniques. The Face Recognition Grand Challenge (FRGC) database has eight thousand fourteen pictures from four hundred sixty six subjects in difference periods. For each subject in each session, there are 4 controlled stationary pictures, 2 uncontrolled stationary pictures, and a single 3D picture. The still pictures have changes like lighting, time-lapse and expression variations etc.
The unconstrained pictures were captured in changing illuminations. Every group of unconstrained pictures have 2 expressions, neutral and smiling. To simulate the randomly located occlusions, one can relocate a randomly placed square patch from every picture by a black block. The position of the occlusion is indiscriminately chosen. The size of the black block varies in the range of 10% to 50% of the original image 2. This database can be downloaded from the link https://www.idiap.ch/software/bob.
The LFW database
This is database that has face pictures constructed for analyzing the issue of uncontrolled recognition of faces which has 13,233 face pictures of 5,749 people assembled from the web. The pictures are collected in uncontrollable situations that have wide changes in expression, illumination, pose, time-gap and several kinds of occlusions. These faces is were detected solely by the Viola-Jones face detector. Every face is being labelled by the name of the person to be photographed. 1,680 of the persons photographed have 2 or more different pictures in the database. The aim of face authentication below the LFW database’s contract is to verify if a pair of face photos belongs to identical person or not. The pictures are present as 250 x 250 pixel JPEG pictures. In this database, most of pictures are in color, though some are grayscale solely 4.
CAS-PEAL Database
It contains of pictures from 66-1040 persons (595 male, 445 female) in 7 groups: expression, pose, accessory, lighting, time, background, and distance. For the pose subset, 9 cameras located in a hemicycle around the person were utilized. The pictures were captured continuously within a small time span (2 seconds) 2. This database can be downloaded from the link http://www.jdl.ac.cn/peal/index.html.
FERET
It was a collective work of Dr. Wechsler and Dr. Phillips. The pictures were gathered in an exceedingly semi-controlled condition. In order to keep a degree of uniformity within the database, the same physical arrangement was utilized in every span of photography. Because the instrumentation had to be assembled once more for each period, there have been few minor changes in photos collected on dissimilar dates. It was gathered in fifteen periods from August 1993 to July 1996.
It has 1564 groups of pictures in total a 14,126 that consists of 1199 persons and 365 duplicate groups of pictures. A duplicate set is another group of pictures of an individual present in the database that was generally captured on a dissimilar day. This database can be downloaded from the link http://www.nist.gov/humanid/feret/. The color FERET dataset can be downloaded from the link http://www.nist.gov/humanid/colorferet/.
