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

Several mathematical models have been proposed for recognizing face images with age variations. However, effect of change in body-weight is also an interesting covariate that has not been much explored. This paper presents a novel approach to incorporate the weight variations during feature learning process. In a deep learning architecture, we propose incorporating the body-weight in terms of a regularization function which helps in learning the latent variables representative of different weight categories. The formulation has been proposed for both Autoencoder and Deep Boltzmann Machine. On extended WIT database of 200 subjects, the comparison with a commercial system and an existing algorithm show that the proposed algorithm outperforms them by more than 9% at rank-10 identification accuracy. © 2015 IEEE.

World University root node

2015 IEEE 7th International Conference on Biometrics Theory, Applications and Systems, BTAS 2015

Regularizing deep learning architecture for face recognition with weight variations

International Journal of Advanced Research in Electrical, Electronics and {\ldots}

A Review on Robot Arm Using Haptic

Broadcasting problem is an important issue in the wireless networks, especially in dynamic wireless networks. In dynamic wireless networks the node density and mobility is high, due to which several problems arise during broadcasting. Two major problems faced are namely, Broadcast Storm Problem and Disconnected network problem. In a highly dense network, if information is being flooded in a loop, it could lead to broadcast storm. The broadcast storm may eventually crash the entire network and lead to loss of information. Mobility of the nodes may lead to the problem of Disconnected Network. If the two nodes sending and receiving information are mobile with different speeds, it could lead to a disconnection between them as soon as the receiver moves out of the communication range. In this paper, we are trying to solve both the problems based on our proposed algorithms. © 2015 IEEE.

International Conference on Electrical, Electronics, Signals, Communication and Optimization, EESCO 2015

Improvised broadcast algorithm for wireless networks

The changes in environmental parameters may demand switching between underlying topologies for better performance of distributed message passing applications. Arbitrary topology switching using distributed tree construction may lead to loss or redundancy in delivery of application messages. In this work, a set of algorithms has been proposed for dynamic switching between two spanning trees to offer better adaptivity towards the environment for different applications. Here, two extreme cases of spanning trees, a Breadth First Search (BFS) tree and a Depth First Search (DFS) tree, rooted at the core node, have been considered for switching. The core node initiates the switching and all other nodes cooperatively change their parents on the fly maintaining the DFS or BFS properties as required. However, the application remains transparent to the switching that assures the availability of the system at any instance of time. Simulation results show that each application message is delivered correctly to the destination without any loss or redundancy. The proposed scheme is scalable and the control message overhead for switching is linear with respect to the number of edges in the communication graph. Furthermore, there is no control message overhead to assure the delivery of application messages at the time of switching. © 2013, Springer Science+Business Media New York.

Journal of Network and Systems Management

Dynamic Tree Switching for Distributed Message-Passing Applications

Today, Information and Communication Technology (ICT) is ubiquitous. It has touched almost all spheres concerning human survival, growth and development. Agriculture is no exception to it. This paper presents the research initiatives towards improving irrigation efficiency through affordable ICT solutions. It presents experimental results of studies related to improving irrigation efficiency in Semi arid zones in India, using commercially available ICT solutions. It compares between simple microirrigation and microirrigation with ICT (irrigation on demand) and establishes that by using ICT in agriculture, we can increase the irrigation efficiencies by a considerable amount. The paper also presents the development of indigenous wireless sensor node and gateway made from open source hardware and software components. The efforts in this regard show, while maintaining comparable performance, a significant reduction in the cost of wireless sensor devices, when compared to commercially available wireless sensor devices, is achieved. © 2015 IEEE.

Department of Mechanical Engineering

2015 7th International Conference on Communication Systems and Networks, COMSNETS 2015 - Proceedings

Development of affordable ICT solutions for water conservation in agriculture

Real world applicability of many computer vision solutions is constrained by the mismatch between the training and test domains. This mismatch might arise because of factors such as change in pose, lighting conditions, quality of imaging devices, intra-class variations inherent in object categories etc. In this work, we present a dictionary learning based approach to tackle the problem of domain mismatch. In our approach, we jointly learn dictionaries for the source and the target domains. The dictionaries are partially shared, i.e. some elements are common across both the dictionaries. These shared elements can represent the information which is common across both the domains. The dictionaries also have some elements to represent the domain specific information. Using these dictionaries, we separate the domain specific information and the information which is common across the domains. We use the latter for training cross-domain classifiers i.e., we build classifiers that work well on a new target domain while using labeled examples only in the source domain. We conduct cross-domain object recognition experiments on popular benchmark datasets and show improvement in results over the existing state of art domain adaptation approaches. © Springer International Publishing Switzerland 2015.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Learning Partially Shared Dictionaries for Domain Adaptation

In this work, we apply the sliced biclique cryptanalysis technique to show 8-round collision attack on a hash function H based on 4-branch, Type-2 Generalized Feistel Network (Type-2 GFN). This attack is generic and works on 4-branch, Type-2 GFN with any parameters including the block size, type of round function, the number of S-boxes in each round and the number of SP layers inside the round function. We first construct a 8-round distinguisher on 4-branch, Type-2 GFN and then use this distinguisher to launch 8-round collision attack on compression functions based on Matyas-Meyer-Oseas (MMO) and Miyaguchi-Preneel (MP) modes. The complexity of the attack on 128- bit compression function is 256. The attack can be directly translated to collision attack on MP and MMO based hash functions and pseudocollision attack on Davies-Meyer (DM) based hash functions. When the round function F is instantiated with double SP layer, we show the first 8 round collision attack on 4-branch, Type-2 GFN with double SP layer based compression function. The previous best attack on this structure was a 6-round near collision attack shown by Sasaki at Indocrypt’12. His attack cannot be used to generate full collisions on 6-rounds and hence our result can be regarded the best so far in literature on this structure. © Springer International Publishing Switzerland 2015.

Collision attack on 4-branch, type-2 GFN based hash functions using sliced biclique cryptanalysis technique

This paper presents VIDeo Content Analysis and Retrieval (VIDCAR), an unsupervised framework for Content-Based Video Retrieval (CBVR) using representation of the dynamics in the spatio-temporal model extracted from video shots. We propose Dynamic Multi Spectro Temporal-Curvature Scale Space (DMST-CSS), an improved feature descriptor for enhancing the performance of CBVR task. Our primary contribution is in representation of the dynamics of the evolution of the MST-CSS surface. Unlike the earlier MST-CSS descriptor [22], which extracts geometric features after the evolving MST-CSS surface converges to a final formation, this DMST-CSS captures the dynamics of the evolution (formation) of the surface and is thus more robust. We have represented the dynamics of MST-CSS surface as a multivariate time series to obtain a DMST-CSS descriptor. A global kernel alignment technique has been adapted to compute a match cost between query and model DMST-CSS descriptor. In our experiments, VIDCAR was shown to have greater precision recall than the competitors on five datasets. © Springer-Verlag London 2014.

International Journal of Multimedia Information Retrieval

VIDCAR: An unsupervised CBVR framework for identifying similar videos with prominent object motion

Extraction of the moving foreground object from a given video shot is an important task for spatiotemporal analysis and content representation in many computer vision and digital video processing applications. We propose an iterative framework based on energy minimization, for segmenting the prominent moving foreground object efficiently from moving camera video (MCV) shots. The solution obtained using graph-cut for figure-ground classification is enhanced using features extracted over a set of neighboring frames. This is used to iteratively update the foreground and background probability (tri-) maps and hence the graph weights. The segmentation results from neighboring frames are integrated as constraints to iteratively guide the energy minimization process, for an efficient solution. The proposed framework is automatic and does not require any human interaction (neither initialization nor refinement). Our method outperforms recent state-of-the-art moving object segmentation techniques on benchmark datasets with MCV shots. © 2014, Springer-Verlag London.

Signal, Image and Video Processing

Prominent moving object segmentation from moving camera video shots using iterative energy minimization

This paper proposes an Autonomous Machine Vision system which grasps a textureless object from a clutter in a single plane, rearranges it for proper placement and then places it using vision. It contributes to a unique vision-based pose estimation algorithm, collision free path planning and dynamic Change-Over algorithm for final placement. © 2015 ACM.