Department of Computer Science & Engineering

&ldquo;IIT Jodhpur is committed to impart quality education to facilitate the creation of qualified, competent and responsible human resources to meet the emerging technological challenges of the world. It is working towards creating an equitable learning platform that is positive, accessible and effective in the face of the modern societal and scientific advancements.

Flexibility in curriculum, in all its academic programs, encompassing elements of teaching, research, creativity, innovation and entrepreneurship is the most unique feature of IIT Jodhpur. The Department of Science and Technology (DST), under the National Mission on Cyber Physical Systems, has identified IIT Jodhpur as a Technology &amp; Innovation Hub in the area of Augmented Reality and Virtual Reality. The institute has also been nominated as the coordinator for the Jodhpur City and Knowledge Cluster by the Office of Principal Scientific Adviser, Government of India.

The sprawling 852-acre IIT Jodhpur campus is a majestic tribute to the deep-rooted diverse cultural heritage of the region. The campus is an architectural masterpiece, which is slated to become an international exemplar of sustainability with net-zero energy, and advanced water and waste management strategies.&rdquo;

IIT Jodhpur

Kinship verification has a number of applications such as organizing large collections of images and recognizing resemblances among humans. In this paper, first, a human study is conducted to understand the capabilities of human mind and to identify the discriminatory areas of a face that facilitate kinship-cues. The visual stimuli presented to the participants determine their ability to recognize kin relationship using the whole face as well as specific facial regions. The effect of participant gender and age and kin-relation pair of the stimulus is analyzed using quantitative measures such as accuracy, discriminability index d', and perceptual information entropy. Utilizing the information obtained from the human study, a hierarchical kinship verification via representation learning (KVRL) framework is utilized to learn the representation of different face regions in an unsupervised manner. We propose a novel approach for feature representation termed as filtered contractive deep belief networks (fcDBN). The proposed feature representation encodes relational information present in images using filters and contractive regularization penalty. A compact representation of facial images of kin is extracted as an output from the learned model and a multi-layer neural network is utilized to verify the kin accurately. A new WVU kinship database is created, which consists of multiple images per subject to facilitate kinship verification. The results show that the proposed deep learning framework (KVRL-fcDBN) yields the state-of-the-art kinship verification accuracy on the WVU kinship database and on four existing benchmark data sets. Furthermore, kinship information is used as a soft biometric modality to boost the performance of face verification via product of likelihood ratio and support vector machine based approaches. Using the proposed KVRL-fcDBN framework, an improvement of over 20% is observed in the performance of face verification. © 1992-2012 IEEE.

World University root node

Fulltext

IEEE Transactions on Image Processing

Hierarchical representation learning for kinship verification

Abundance and availability of video capture devices, such as mobile phones and surveillance cameras, have instigated research in video face recognition, which is highly pertinent in law enforcement applications. While the current approaches have reported high accuracies at equal error rates, performance at lower false accept rates requires significant improvement. In this paper, we propose a novel face verification algorithm, which starts with selecting feature-rich frames from a video sequence using discrete wavelet transform and entropy computation. Frame selection is followed by representation learning-based feature extraction, where three contributions are presented: 1) deep learning architecture, which is a combination of stacked denoising sparse autoencoder (SDAE) and deep Boltzmann machine (DBM); 2) formulation for joint representation in an autoencoder; and 3) updating the loss function of DBM by including sparse and low rank regularization. Finally, a multilayer neural network is used as the classifier to obtain the verification decision. The results are demonstrated on two publicly available databases, YouTube Faces and Point and Shoot Challenge. Experimental analysis suggests that: 1) the proposed feature-richness-based frame selection offers noticeable and consistent performance improvement compared with frontal only frames, random frames, or frame selection using perceptual no-reference image quality measures and 2) joint feature learning in SDAE and sparse and low rank regularization in DBM helps in improving face verification performance. On the benchmark Point and Shoot Challenge database, the algorithm yields the verification accuracy of over 97% at 1% false accept rate whereas, on the YouTube Faces database, over 95% verification accuracy is observed at equal error rate. © 2005-2012 IEEE.

IEEE Transactions on Information Forensics and Security

Face Verification via Learned Representation on Feature-Rich Video Frames

Face recognition systems are susceptible to presentation attacks such as printed photo attacks, replay attacks, and 3D mask attacks. These attacks, primarily studied in visible spectrum, aim to obfuscate or impersonate a person's identity. This paper presents a unique multispectral video face database for face presentation attack using latex and paper masks. The proposed Multispectral Latex Mask based Video Face Presentation Attack (MLFP) database contains 1350 videos in visible, near infrared, and thermal spectrums. Since the database consists of videos of subjects without any mask as well as wearing ten different masks, the effect of identity concealment is analyzed in each spectrum using face recognition algorithms. We also present the performance of existing presentation attack detection algorithms on the proposed MLFP database. It is observed that the thermal imaging spectrum is most effective in detecting face presentation attacks. © 2017 IEEE.

IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops

Face Presentation Attack with Latex Masks in Multispectral Videos

The main objective of the paper is to recommend an essential improvement to the existing Multi-Domain Convolutional Neural Network tracker (MDNet) which is used to track unknown object in a video-stream. MDNet is able to handle major basic tracking challenges like fast motion, background clutter, out of view, scale variations etc. through offline training and online tracking. We pre-train the Convolutional Neural Network (CNN) offline using many videos with ground truth to obtain a target representation in the network. In online tracking the MDNet uses large number of random sample of windows around the previous target for estimating the target in the current frame which make its tracking computationally complex while testing or obtaining the track. The major contribution of the paper is to give guided samples to the MDNet rather than random samples so that the computation and time required by the CNN while tracking could be greatly reduced. Evaluation of the proposed algorithm is done using the videos from the ALOV300++ dataset and the VOT dataset and the results are compared with the state of art trackers. © 2017 IEEE.

2017 23rd National Conference on Communications, NCC 2017

Computationally efficient deep tracker: Guided MDNet

With a rapid growth in smartphone technology, there is a need to provide secured access to critical data using personal authentication. Existing access mechanisms such as pin and password suffer due to lack of security from shoulder-surfing attacks. Use of biometric modalities such as fingerprint are currently explored in existing smartphones as a more secure authentication mechanism. Typically, fingerprint capturing requires an extra sensor, adding to the cost of the device as well as denying backend services to existing smartphone devices. Using a rear camera captured fingerphoto image provides a cheap alternate option, without the need for a dedicated sensor for capturing images. However, unlike fingerprints fingerphoto images are captured in a more uncontrolled environment including any outdoor conditions. Hence, fingerphoto matching is prone to many challenges such as varying environmental illumination and surrounding background. We propose a novel end-to-end fingerphoto matching pipeline by studying the effect of different environmental conditions in fingerphoto matching. The pipeline consists of the following major contributions: (i) a segmentation technique to extract the fingerphoto region of interest from varying background, (ii) an enhancement module to neutralize the illumination imbalance and increase the ridge-valley contrast, (iii) a scattering network based fingerphoto representation technique to deal with the pose variations, whose resultant features are invariant to geometric transformations, and (iv) a learning based matching technique to accommodate maximum variations occurring in fingerphoto images. To experimentally study the challenging conditions such as background and illumination, we create a publicly available fingerphoto dataset, IIITD SmartPhone Fingerphoto Database v1, along with the corresponding live-scan prints. The experiments performed on the dataset shows that the proposed matching pipeline provides an improved performance when compared with some of the existing approaches. © 2017 Elsevier Ltd. All rights reserved.

Human Recognition in Unconstrained Environments: Using Computer Vision, Pattern Recognition and Machine Learning Methods for Biometrics

Fingerphoto authentication using smartphone camera captured under varying environmental conditions

Automatically finding out existing building layouts from a repository is always helpful for an architect to ensure reuse of design and timely completion of projects. In this paper, we propose Deep Architecture for fiNdIng alikE Layouts (DANIEL). Using DANIEL, an architect can search from the existing projects repository of layouts (floor plan), and give accurate recommendation to the buyers. DANIEL is also capable of recommending the property buyers, having a floor plan image, the corresponding rank ordered list of alike layouts. DANIEL is based on the deep learning paradigm to extract both low and high level semantic features from a layout image. The key contributions in the proposed approach are: (i) novel deep learning framework to retrieve similar floor plan layouts from repository; (ii) analysing the effect of individual deep convolutional neural network layers for floor plan retrieval task; and (iii) creation of a new complex dataset ROBIN (Repository Of BuildIng plaNs), having three broad dataset categories with 510 real world floor plans.We have evaluated DANIEL by performing extensive experiments on ROBIN and compared our results with eight different state-of-The-art methods to demonstrate DANIEL's effectiveness on challenging scenarios. © 2017 IEEE.

Proceedings of the International Conference on Document Analysis and Recognition, ICDAR

DANIEL: A Deep Architecture for Automatic Analysis and Retrieval of Building Floor Plans

Wireless communication is one of the most dynamic and vibrant areas of technology development in the communication field today. It has been found that severe climatic conditions disturb the propagation of electromagnetic signals at higher frequencies (greater than 30 MHz). The disturbance is mainly due to molecular absorption by oxygen for frequencies ranging between 60 and 118 GHz and due to water vapour in 22, 183 and 325 GHz bands. Rain and fog has the most significant impact, since the size of the rain drops is of the order of the wavelength of the transmitted signal. This results in energy absorption by the rain drops themselves and as a secondary effect energy is scattered by the drops. The frequency selective absorption characteristics of the atmosphere can be approximated by a transfer function. In most of the practical channels when the signal propagates through the atmosphere the effect of many factors on the signal has to be considered along with the free space propagation

diplom

Development of Field Propagation Model for Urban Area.

Multi-hop Wireless Mesh Networks (WMNs) is a promising new technique for communication with routing protocol designs being critical to the effective and efficient of these WMNs. A common approach for routing traffic in these networks is to select a minimal distance from source to destination as in wire-line networks. Opportunistic Routing(OR) makes use of the broadcasting ability of wireless network and is especially very helpful for WMN because all nodes are static. Our proposed scheme of Multicast Opportunistic Routing(MOR) in WMNs is based on the broadcast transmissions and Learning Au-tomata (LA) to expand the potential candidate nodes that can aid in the process of retransmission of the data. The receivers are required to be in sync with one another in order to avoid duplicated broadcasting of data which is generally achieved by formulating the forwarding candidates according to some LA based metric. The most adorable aspect of this protocol is that it intelligently "learns" from the past experience and improves its performance. The results obtained via this approach of MOR, shows that the proposed scheme outperforms with some existing sachems and is an improved and more effective version of opportunistic routing in mesh network. © 2017 ACM.

ACM International Conference Proceeding Series

Algorithm for Multicast Opportunistic Routing in Wireless Mesh Networks

Centrality measures have perpetually been helpful to find the foremost central or most powerful node within the network. There are numerous strategies to compute centrality of a node however in social networks betweenness centrality is the most widely used approach to bifurcate communities within the network, to find out the susceptibility within the complex networks and to generate the scale free networks whose degree distribution follows the power law. In this paper, we've computed betweenness centrality by identifying communities lying within the network. Our algorithm efficiently updates the centrality of the nodes whenever any edge or vertex addition or deletion takes place within the dynamic network by modifying solely a subset of vertices. For the vertex addition, Incremental Algorithm has been used in which Streaming graphs has also been considered. Brandes approach is the most widely used approach for finding out the betweenness centrality however it's still expensive for growing networks since it takes O(mn+n2logn) amount of time and O(n+m) space however our approach efficiently updates the centrality of the nodes by taking O(|S|n+|S|nlogn) amount of time where |S| is the subset of the vertices, m is the number of edges, n is the number of vertices and |S| ≤ n holds true. © 2017 ACM.

Betweenness centrality updation and community detection in streaming graphs using incremental algorithm

Multi-hop Wireless Mesh Networks (WMNs) is a promising new technique for communication with routing protocol designs being critical to the effective and efficient of these WMNs. A common approach for routing traffic in these networks is to select a minimal distance from the source to destination as in wireless networks. Opportunistic Routing (OR) makes use of the broadcasting ability of wireless network and is especially very helpful for WMN because all nodes are static. Our proposed scheme of Multicast Opportunistic Routing (MOR) in WMNs is based on the broadcast transmissions and Learning Automata (LA) to expand the potential candidate nodes that can aid in the process of retransmission of the data. The receivers are required to be in sync with one another in order to avoid duplicated broadcasting of data which is generally achieved by formulating the forwarding candidates according to some LA based metric. The most helpful aspect of this protocol is that it intelligently learns from the past experience and improves its performance. The results obtained via this approach of MOR show that the proposed scheme outperforms some existing schemes and is an improved and more effective version of opportunistic routing in a mesh network. Copyright © 2017 Inderscience Enterprises Ltd.