![ENHANCEMENT OF FACE RETRIVAL DESIGEND FOR
MANAGING HUMAN ASPECTS
Devarapalli Lakshmi Sowmya1, V.Sreenatha Sarma2
1M.Tech Student, Dept of CSE, Audisankara College of Engineering & Technology, Gudur, A.P, India
2Associate Professor, Dept of CSE, Audisankara College of Engineering & Technology, Gudur, A.P, India
ABSTRACT:
Traditional methods of content-based image retrieval make use of image content like color,
texture as well as gradient to symbolize images. By combine low-level characteristics with
high level human features, we are capable to discover enhanced feature representations and
attain improved retrieval results. The recent effort explains automatic attribute recognition
has sufficient quality on numerous different human attributes. Content-based face image
retrieval is strongly associated to the problems of face recognition however they focus on
finding appropriate feature representations in support of scalable indexing systems. To
leverage capable human attributes automatically identified by attribute detectors in support of
getting better content-based face image retrieval, we put forward two orthogonal systems
named attribute enhanced sparse coding as well as attribute-embedded inverted indexing.
Attribute-embedded inverted indexing believes human attributes of chosen query image in a
binary signature as well as make available resourceful recovery in online stage. Attribute-
enhanced sparse coding make use of global structure and employ quite a lot of human
attributes to build semantic-aware code words in offline stage. The projected indexing system
can be effort less integrated into inverted index, consequently maintaining a scalable
structure.
Keywords: Content-based image retrieval, Attribute recognition, Feature representations,
Binary signature, Semantic-aware code words.
1. INTRODUCTION:
In the recent times, human attributes of
automatically detected have been revealed
capable in various applications. To get
better the value of attributes, relative
attributes were applied. Multi-attribute
space was introduced to standardize
assurance scores from various attributes
[4]. By means of automatically detected
human attributes, excellent performance
was achieved on retrieval of keyword
based face image as well as face
verification detectors in support of search
of similar attribute. Due to increase of
photo sharing or social network services,
there increase tough needs for extensive
content-based retrieval of face image.
Content-based face image retrieval is
strongly associated to the problems of face
recognition however they focus on finding
appropriate feature representations in
support of scalable indexing systems [8].
As face recognition generally necessitate
substantial computation outlay in support
of dealing with high dimensional features
as well as generate explicit models of
classification, it is non-trivial to directly
pertain it towards tasks of face retrieval.
Even though images obviously have
extremely high dimensional representation,
those within similar class generally lie on a
low dimensional subspace [1]. Sparse
coding can make use of semantics of
information and attain capable results in
numerous different applications for
instance image classification as well as
face recognition. Even though these works
accomplish significant performance on
keyword based face image recovery as
well as face recognition, we put forward to
make use of effectual ways to merge low-
level features and automatically noticed
facial attributes in support of scalable face
image retrieval [11]. Human attributes are
high level semantic description concerning
an individual. The recent effort explains
289
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in](https://image.slidesharecdn.com/iaetsd-enhancementoffaceretrivaldesigendfor-150407215543-conversion-gate01/85/Iaetsd-enhancement-of-face-retrival-desigend-for-1-320.jpg)
![automatic attribute recognition has
sufficient quality on numerous different
human attributes. Using human attributes,
numerous researchers have attained
capable results in various applications for
instance face verification, identification of
face, keyword based face image recovery
as well as similar attribute search [3]. Even
though human attributes have been
revealed practical on applications
associated towards face images, it is not
trivial towards concerning in retrieval task
of content-based face image due to quite a
lot of reasons. Human attributes simply
enclose limited dimensions. When there
are more over numerous people in dataset,
it loses discrimin ability as assured people
may have comparable attributes [14].
Human attributes are represented as vector
concerning floating points. It does not
effort well with increasing extensive
indexing methods, and consequently it
suffers from slow reply and scalability
concern when the data size is enormous.
2. METHODOLOGY:
Traditional methods of content-based
image retrieval make use of image content
like color, texture as well as gradient to
symbolize images [13]. Traditional
methods in support of face image retrieval
typically employ low-level features to
correspond to faces but low-level
characteristics are be short of of semantic
meanings as well as face images typically
include high intra class variance thus the
recovery results are unacceptable [9].
When specified a face image query,
retrieval of content-based face image effort
to discover comparable face images from
huge image database. It is an enabling
knowledge for numerous applications
includes automatic face annotation,
investigation of crime. By combining low-
level characteristics with high-level human
features, we are proficient to discover
enhanced feature representations and attain
improved retrieval results [7]. To deal with
extensive information, mainly two types of
indexing systems are employed. Numerous
studies have leveraged inverted indexing
or else hashbased indexing pooled with
bag-of-word representation as well as
local features to attain well-organized
similarity search. Even though these
methods can attain high precision on rigid
object recovery, they go through from low
recall difficulty due to semantic gap [2].
The significance as well as sheer amount
of human face photos makes
manipulations of extensive human face
images actually significant research
difficulty and facilitate numerous real
world applications. In recent times, some
researchers have fixed on bridging
semantic gap by discovery of semantic
image representations to get better
performance of content-based image
retrieval [16]. To leverage capable human
attributes automatically identified by
attribute detectors in support of getting
better content-based face image retrieval,
we put forward two orthogonal systems
namedchosen query image in a binary
signature as well as make available
resourceful recovery in online stage [12].
Attribute-enhanced sparse coding make
use of global structure and employ quite a
lot of human attributes to build semantic-
aware code words in offline stage. By
incorporating these methods, we put up an
extensive content-based face image
retrieval scheme by taking benefits of low-
level features as well as high-level
semantic [5]. When a query image arrives,
it will experience same process to get hold
of sparse code words as well as human
attributes, and make use of these code
words by binary attribute signature to get
back images in index system. By means of
sparse coding, a mark is a linear grouping
of column vectors of dictionary [15]. As
learning dictionary with a huge vocabulary
is lengthy we can presently make use of
randomly sampled image patch as
dictionary and pass over prolonged
dictionary learning measure. To believe
human attributes in sparse illustration, we
initially put forward to make use of
dictionary selection to compelte images
with various attribute values to enclose
290
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in](https://image.slidesharecdn.com/iaetsd-enhancementoffaceretrivaldesigendfor-150407215543-conversion-gate01/85/Iaetsd-enhancement-of-face-retrival-desigend-for-2-320.jpg)
![various codewords [10]. For a single
human aspect, we separate dictionary
centroids into two various subsets, images
all the way through positive attribute
scores will make use of subset as well as
images by negative attribute scores will
employ the other [6]. For cases of
numerous attributes, we separate sparse
representation into numerous segments
based on number of features, and every
section of sparse representation is
produced depending on distinct aspect.
Fig1: An overview of structure of proposed system
3. RESULTS:
Attribute-enhanced sparse coding make
use of global structure and employ quite a
lot of human attributes to build semantic-
aware code words in offline stage.
Attribute embedded inverted indexing
additionally believe local attribute
signature concerning attribute-enhanced
sparse coding as well as attribute-
embedded inverted indexing as shown in
fig1. Attribute-embedded inverted
indexing believes human attributes of
query image and still make sure proficient
recovery in online stage. The experimental
result illustrate that by means of code
words generated by projected coding
system, we can decrease the quantization
error as well as attain salient gains in face
recovery on public datasets. The projected
indexing system can be effortlessly
integrated into inverted index,
consequently maintaining a scalable
structure. Certain informative attributes
were discovered in support of face
retrieval across various datasets and these
aspects are capable for other applications.
Attribute-enhanced sparse codewords
would additionally get better accurateness
of retrieval of content-based face image.
4. CONCLUSION:
Even though human attributes have been
revealed practical on applications
associated towards face images, it is not
trivial towards concerning it in retrieval
task of content based face image due to
quite a lot of reasons. Traditional methods
in support of face image retrieval typically
employ low level features to correspond to
faces but low-level characteristics are be
short of semantic meanings as well as face
images typically include high intra-class
variance thus the recovery results are
unacceptable. Using human attributes,
numerous researchers have attained
capable results in various applications for
instance face verification, identification of
face, keyword based face image recovery
as well as similar attribute search.
Numerous studies have leveraged inverted
indexing or else hash based indexing
pooled with bag-of word representation as
well as local features to attain well-
organized similarity search. To believe
human attributes in sparse illustration, we
initially put forward to make use of
dictionary selection to compel images with
various attribute values to enclose various
code words. The significance as well as
sheer amount of human face photos makes
manipulations of extensive human face
images actually significant research
difficulty and facilitate numerous real
world applications. In recent times, some
researchers have fixed on bridging
semantic gap by discovery of semantic
image representations to get better
performance of content-based image
retrieval. The experimental result illustrate
that by means of codewords generated by
291
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in](https://image.slidesharecdn.com/iaetsd-enhancementoffaceretrivaldesigendfor-150407215543-conversion-gate01/85/Iaetsd-enhancement-of-face-retrival-desigend-for-3-320.jpg)
![projected coding system, we can decrease
the quantization error as well as attain
salient gains in face recovery on public
datasets.
REFERENCES:
[1] J. Wright, A. Yang, A. Ganesh, S.
Sastry, and Y. Ma, “Robust face
recognition via sparse representation,”
IEEE Transactions on Pattern Analysis and
Machine Intelligence (PAMI), 2009
[2] Scalable Face Image Retrieval using
Attribute-Enhanced Sparse Code words
Bor-Chun Chen, Yan-Ying Chen, Yin-Hsi
Kuo, Winston H. Hsu,2013
[3] G. B. Huang, M. Ramesh, T. Berg, and
E. Learned-Miller, “Labeled faces in the
wild: A database for studying face
recognition in unconstraine denvironments
, ” University of Massachusetts, Amherst,
Tech.Rep. 07-49, October 2007.
[4] Y.-H. Kuo, H.-T. Lin, W.-H. Cheng,
Y.-H. Yang, and W. H. Hsu
,“Unsupervised auxiliary visual words
discovery for large scale image object
retrieval,” IEEE Conference on Computer
Vision and Pattern Recognition, 2011
[5] J. Yang, K. Yu, Y. Gong, and T.
Huang, “Linear spatial pyramid matching
using sparse coding for image
classification,” IEEE Conference on
Computer Vision and Pattern Recognition,
2009
[6] N. Kumar, A. C. Berg, P. N.
Belhumeur, and S. K. Nayar,
“Describablevisual attributes for face
verification and image search,” in IEEE
Transactionson Pattern Analysis and
Machine Intelligence (PAMI),
SpecialIssue on Real-World Face
Recognition, Oct 2011
[7] O. Chum, J. Philbin, J. Sivic, M. Isard
and A. Zisserman, “Total
Recall:Automatic Query Expansion with a
Generative Feature Model for
ObjectRetrieval,” IEEE International
Conference on Computer Vision, 2007
[8] A. Torralba, K. P. Murphy, W. T.
Freeman, and M. A. Rubin, “Context base
dvision system for place and object
recognition,” International Conference on
Computer Vision, 2003.
[9] J. Wang, J. Yang, K. Yu, F. Lv, T.
Huang, and Y. Gong, “Locality
constrained linear coding for image
classification,” IEEE Conference on
Computer Vision and Pattern Recognition,
2010.
[10] Z. Wu, Q. Ke, J. Sun, and H.-Y.
Shum, “Scalable face image retrieval with
identity-based quantization and multi-
reference reranking,” IEEE Conference on
Computer Vision and Pattern Recognition,
2010.
[11] W. Scheirer and N. Kumar and P.
Belhumeur and T. Boult, “Multi-Attribute
Spaces: Calibration for Attribute Fusion
and Similarity Search,” IEEE Conference
on Computer Vision and Pattern
Recognition, 2012.
[12] J. Wang, S. Kumar, and S.-F. Chang,
“Semi-supervised hashing for scalable
image retrieval,” IEEE Conference on
Computer Vision andPattern Recognition,
2010.
[13] M. Douze and A. Ramisa and C.
Schmid, “Combining Attributes and Fisher
Vectors for Efficient Image Retrieval,”
IEEE Conference on Computer Vision and
Pattern Recognition, 2011
[14] S. Lazebnik, C. Schmid, and J. Ponce,
“Beyond bags of features :Spatial pyramid
matching for recognizing natural scene
categories,”IEEE Conference on Computer
Vision and Pattern Recognition, 2006
[15] H. Jegou, M. Douze, and C. Schmid,
“Hamming embedding and weak
geometric consistency for large scale
image search,” European Conference on
Computer Vision, 2008
[16] W. Scheirer, N. Kumar, K. Ricanek,
T. E. Boult, and P. N. Belhumeur, “Fusing
with context: a bayesian approach to
combining descriptive attributes,”
International Joint Conference on
Biometrics, 2011.
292
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in](https://image.slidesharecdn.com/iaetsd-enhancementoffaceretrivaldesigendfor-150407215543-conversion-gate01/85/Iaetsd-enhancement-of-face-retrival-desigend-for-4-320.jpg)
Iaetsd enhancement of face retrival desigend for
- 1. ENHANCEMENT OF FACE RETRIVAL DESIGEND FOR MANAGING HUMAN ASPECTS Devarapalli Lakshmi Sowmya1, V.Sreenatha Sarma2 1M.Tech Student, Dept of CSE, Audisankara College of Engineering & Technology, Gudur, A.P, India 2Associate Professor, Dept of CSE, Audisankara College of Engineering & Technology, Gudur, A.P, India ABSTRACT: Traditional methods of content-based image retrieval make use of image content like color, texture as well as gradient to symbolize images. By combine low-level characteristics with high level human features, we are capable to discover enhanced feature representations and attain improved retrieval results. The recent effort explains automatic attribute recognition has sufficient quality on numerous different human attributes. Content-based face image retrieval is strongly associated to the problems of face recognition however they focus on finding appropriate feature representations in support of scalable indexing systems. To leverage capable human attributes automatically identified by attribute detectors in support of getting better content-based face image retrieval, we put forward two orthogonal systems named attribute enhanced sparse coding as well as attribute-embedded inverted indexing. Attribute-embedded inverted indexing believes human attributes of chosen query image in a binary signature as well as make available resourceful recovery in online stage. Attribute- enhanced sparse coding make use of global structure and employ quite a lot of human attributes to build semantic-aware code words in offline stage. The projected indexing system can be effort less integrated into inverted index, consequently maintaining a scalable structure. Keywords: Content-based image retrieval, Attribute recognition, Feature representations, Binary signature, Semantic-aware code words. 1. INTRODUCTION: In the recent times, human attributes of automatically detected have been revealed capable in various applications. To get better the value of attributes, relative attributes were applied. Multi-attribute space was introduced to standardize assurance scores from various attributes [4]. By means of automatically detected human attributes, excellent performance was achieved on retrieval of keyword based face image as well as face verification detectors in support of search of similar attribute. Due to increase of photo sharing or social network services, there increase tough needs for extensive content-based retrieval of face image. Content-based face image retrieval is strongly associated to the problems of face recognition however they focus on finding appropriate feature representations in support of scalable indexing systems [8]. As face recognition generally necessitate substantial computation outlay in support of dealing with high dimensional features as well as generate explicit models of classification, it is non-trivial to directly pertain it towards tasks of face retrieval. Even though images obviously have extremely high dimensional representation, those within similar class generally lie on a low dimensional subspace [1]. Sparse coding can make use of semantics of information and attain capable results in numerous different applications for instance image classification as well as face recognition. Even though these works accomplish significant performance on keyword based face image recovery as well as face recognition, we put forward to make use of effectual ways to merge low- level features and automatically noticed facial attributes in support of scalable face image retrieval [11]. Human attributes are high level semantic description concerning an individual. The recent effort explains 289 INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT ISBN: 378 - 26 - 138420 - 5 www.iaetsd.in
- 2. automatic attribute recognition has sufficient quality on numerous different human attributes. Using human attributes, numerous researchers have attained capable results in various applications for instance face verification, identification of face, keyword based face image recovery as well as similar attribute search [3]. Even though human attributes have been revealed practical on applications associated towards face images, it is not trivial towards concerning in retrieval task of content-based face image due to quite a lot of reasons. Human attributes simply enclose limited dimensions. When there are more over numerous people in dataset, it loses discrimin ability as assured people may have comparable attributes [14]. Human attributes are represented as vector concerning floating points. It does not effort well with increasing extensive indexing methods, and consequently it suffers from slow reply and scalability concern when the data size is enormous. 2. METHODOLOGY: Traditional methods of content-based image retrieval make use of image content like color, texture as well as gradient to symbolize images [13]. Traditional methods in support of face image retrieval typically employ low-level features to correspond to faces but low-level characteristics are be short of of semantic meanings as well as face images typically include high intra class variance thus the recovery results are unacceptable [9]. When specified a face image query, retrieval of content-based face image effort to discover comparable face images from huge image database. It is an enabling knowledge for numerous applications includes automatic face annotation, investigation of crime. By combining low- level characteristics with high-level human features, we are proficient to discover enhanced feature representations and attain improved retrieval results [7]. To deal with extensive information, mainly two types of indexing systems are employed. Numerous studies have leveraged inverted indexing or else hashbased indexing pooled with bag-of-word representation as well as local features to attain well-organized similarity search. Even though these methods can attain high precision on rigid object recovery, they go through from low recall difficulty due to semantic gap [2]. The significance as well as sheer amount of human face photos makes manipulations of extensive human face images actually significant research difficulty and facilitate numerous real world applications. In recent times, some researchers have fixed on bridging semantic gap by discovery of semantic image representations to get better performance of content-based image retrieval [16]. To leverage capable human attributes automatically identified by attribute detectors in support of getting better content-based face image retrieval, we put forward two orthogonal systems namedchosen query image in a binary signature as well as make available resourceful recovery in online stage [12]. Attribute-enhanced sparse coding make use of global structure and employ quite a lot of human attributes to build semantic- aware code words in offline stage. By incorporating these methods, we put up an extensive content-based face image retrieval scheme by taking benefits of low- level features as well as high-level semantic [5]. When a query image arrives, it will experience same process to get hold of sparse code words as well as human attributes, and make use of these code words by binary attribute signature to get back images in index system. By means of sparse coding, a mark is a linear grouping of column vectors of dictionary [15]. As learning dictionary with a huge vocabulary is lengthy we can presently make use of randomly sampled image patch as dictionary and pass over prolonged dictionary learning measure. To believe human attributes in sparse illustration, we initially put forward to make use of dictionary selection to compelte images with various attribute values to enclose 290 INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT ISBN: 378 - 26 - 138420 - 5 www.iaetsd.in
- 3. various codewords [10]. For a single human aspect, we separate dictionary centroids into two various subsets, images all the way through positive attribute scores will make use of subset as well as images by negative attribute scores will employ the other [6]. For cases of numerous attributes, we separate sparse representation into numerous segments based on number of features, and every section of sparse representation is produced depending on distinct aspect. Fig1: An overview of structure of proposed system 3. RESULTS: Attribute-enhanced sparse coding make use of global structure and employ quite a lot of human attributes to build semantic- aware code words in offline stage. Attribute embedded inverted indexing additionally believe local attribute signature concerning attribute-enhanced sparse coding as well as attribute- embedded inverted indexing as shown in fig1. Attribute-embedded inverted indexing believes human attributes of query image and still make sure proficient recovery in online stage. The experimental result illustrate that by means of code words generated by projected coding system, we can decrease the quantization error as well as attain salient gains in face recovery on public datasets. The projected indexing system can be effortlessly integrated into inverted index, consequently maintaining a scalable structure. Certain informative attributes were discovered in support of face retrieval across various datasets and these aspects are capable for other applications. Attribute-enhanced sparse codewords would additionally get better accurateness of retrieval of content-based face image. 4. CONCLUSION: Even though human attributes have been revealed practical on applications associated towards face images, it is not trivial towards concerning it in retrieval task of content based face image due to quite a lot of reasons. Traditional methods in support of face image retrieval typically employ low level features to correspond to faces but low-level characteristics are be short of semantic meanings as well as face images typically include high intra-class variance thus the recovery results are unacceptable. Using human attributes, numerous researchers have attained capable results in various applications for instance face verification, identification of face, keyword based face image recovery as well as similar attribute search. Numerous studies have leveraged inverted indexing or else hash based indexing pooled with bag-of word representation as well as local features to attain well- organized similarity search. To believe human attributes in sparse illustration, we initially put forward to make use of dictionary selection to compel images with various attribute values to enclose various code words. The significance as well as sheer amount of human face photos makes manipulations of extensive human face images actually significant research difficulty and facilitate numerous real world applications. In recent times, some researchers have fixed on bridging semantic gap by discovery of semantic image representations to get better performance of content-based image retrieval. The experimental result illustrate that by means of codewords generated by 291 INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT ISBN: 378 - 26 - 138420 - 5 www.iaetsd.in
- 4. projected coding system, we can decrease the quantization error as well as attain salient gains in face recovery on public datasets. REFERENCES: [1] J. Wright, A. Yang, A. Ganesh, S. Sastry, and Y. Ma, “Robust face recognition via sparse representation,” IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2009 [2] Scalable Face Image Retrieval using Attribute-Enhanced Sparse Code words Bor-Chun Chen, Yan-Ying Chen, Yin-Hsi Kuo, Winston H. Hsu,2013 [3] G. B. Huang, M. Ramesh, T. Berg, and E. Learned-Miller, “Labeled faces in the wild: A database for studying face recognition in unconstraine denvironments , ” University of Massachusetts, Amherst, Tech.Rep. 07-49, October 2007. [4] Y.-H. Kuo, H.-T. Lin, W.-H. Cheng, Y.-H. Yang, and W. H. Hsu ,“Unsupervised auxiliary visual words discovery for large scale image object retrieval,” IEEE Conference on Computer Vision and Pattern Recognition, 2011 [5] J. Yang, K. Yu, Y. Gong, and T. Huang, “Linear spatial pyramid matching using sparse coding for image classification,” IEEE Conference on Computer Vision and Pattern Recognition, 2009 [6] N. Kumar, A. C. Berg, P. N. Belhumeur, and S. K. Nayar, “Describablevisual attributes for face verification and image search,” in IEEE Transactionson Pattern Analysis and Machine Intelligence (PAMI), SpecialIssue on Real-World Face Recognition, Oct 2011 [7] O. Chum, J. Philbin, J. Sivic, M. Isard and A. Zisserman, “Total Recall:Automatic Query Expansion with a Generative Feature Model for ObjectRetrieval,” IEEE International Conference on Computer Vision, 2007 [8] A. Torralba, K. P. Murphy, W. T. Freeman, and M. A. Rubin, “Context base dvision system for place and object recognition,” International Conference on Computer Vision, 2003. [9] J. Wang, J. Yang, K. Yu, F. Lv, T. Huang, and Y. Gong, “Locality constrained linear coding for image classification,” IEEE Conference on Computer Vision and Pattern Recognition, 2010. [10] Z. Wu, Q. Ke, J. Sun, and H.-Y. Shum, “Scalable face image retrieval with identity-based quantization and multi- reference reranking,” IEEE Conference on Computer Vision and Pattern Recognition, 2010. [11] W. Scheirer and N. Kumar and P. Belhumeur and T. Boult, “Multi-Attribute Spaces: Calibration for Attribute Fusion and Similarity Search,” IEEE Conference on Computer Vision and Pattern Recognition, 2012. [12] J. Wang, S. Kumar, and S.-F. Chang, “Semi-supervised hashing for scalable image retrieval,” IEEE Conference on Computer Vision andPattern Recognition, 2010. [13] M. Douze and A. Ramisa and C. Schmid, “Combining Attributes and Fisher Vectors for Efficient Image Retrieval,” IEEE Conference on Computer Vision and Pattern Recognition, 2011 [14] S. Lazebnik, C. Schmid, and J. Ponce, “Beyond bags of features :Spatial pyramid matching for recognizing natural scene categories,”IEEE Conference on Computer Vision and Pattern Recognition, 2006 [15] H. Jegou, M. Douze, and C. Schmid, “Hamming embedding and weak geometric consistency for large scale image search,” European Conference on Computer Vision, 2008 [16] W. Scheirer, N. Kumar, K. Ricanek, T. E. Boult, and P. N. Belhumeur, “Fusing with context: a bayesian approach to combining descriptive attributes,” International Joint Conference on Biometrics, 2011. 292 INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING AND TECHNOLOGY INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR SKILL DEVELOPMENT ISBN: 378 - 26 - 138420 - 5 www.iaetsd.in