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

Visual tracking is the process of estimating the position of an object in a video sequence and plays a very important role in the field of autonomous video processing. Recent work renders that the trackers developed using deep learning techniques such as the convolutional neural network (CNN) exhibits outstanding performances in terms of accuracy and robustness as compared to other state-of-the-art trackers. Multi-domain convolutional neural network (MDNet) is a deep tracker which uses the CNN for estimating the target in each frame of the video sequence. The majority of the tracking challenges could be very easily handled by the MDNet tracker due to its offline training and online tracking features. The offline training stage helps in capturing the target representations into the shared layers of the CNN, while the online tracking uses a large number of probable random samples of bboxes (bounding boxes) around the previous target for estimating the target in the current frame. Once the target is estimated, a process of fine-tuning is performed which will update the weights of shared layers of CNN. The large number of random samples used for target estimation and the huge number of random training samples generated for the fine-tuning during the online stage makes tracking by MDNet computationally complex and slow. The major contribution of this paper is to suggest using guided samples to the input of the CNN rather than random samples. Moreover, it generates a lesser number of highly efficient training samples for the fine-tuning which helps in decreasing the computational complexity of the tracker by half without much compromise on the performance and thus improves the speed of the tracking process. An extensive evaluation has been performed on the proposed Guided MDNet with different datasets like ALOV300++, OTB and VOT, and its performances are measured in terms of metrics like F-score, one-pass evaluation, robustness and accuracy. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH, DE part of Springer Nature.

World University root node

Visual Computer

Guided MDNet tracker with guided samples

The structural and magnetic properties of Sr0.67Ca0.33Fe9Al3-xCrxO19 submicrometric powders with sub-stitution levels ranging from x = 0 up to x = 3, prepared by sol-gel synthesis method, have been investigated. The powders were characterized by high resolution synchrotron X-ray diffraction, scanning electron microscopy, powder neutron diffraction and MPMS-SQUID magnetometer. The structural analysis shows that upon Ca-substitution of SrFe12O19 at the Sr site and Al-Cr at the Fe sites, the magnetoplumbite structure is preserved, even for large levels of substitutions. The X-ray powder diffraction peak broadening and Williamson-Hall plots analyses indicate an increase of crystallites size and relaxation of the microstrain with increasing amounts of Chromium. The magnetic hysteresis loops reveal high coercivity fields at room temperature, with the highest value of H-C = 1125 kA/m (1.41 T) obtained for Sr0.67Ca0.33Fe9Al2.5Cr0.5O19 powder interesting to develop free-Rare Earth high coercivity magnets. The saturation magnetization and remanence values increase monotonically with increasing Cr content while the switching field decreases. The structural Rietveld refinement of combined X-ray/neutron diffraction data shows the affinity of Al3+ and Cr3+ cations to migrate mainly towards (2 alpha) Fe-Oh2 and (12k) Fe-Oh3 sites and only minor amounts is found on the (4f1) Fe-Oh1 octahedral sites. The structural observations corroborate the magnetic properties, demonstrating a correlation between the structure and magnetic properties in Ca-Al-Cr substituted strontium hexaferrite. The role of the particle size to maximize the coercivity properties is discussed. (C) 2021 Published by Elsevier B.V.

JOURNAL OF ALLOYS AND COMPOUNDS

High magnetic coercive field in Ca-Al-Cr substituted strontium hexaferrite

AAAI Conference on Artificial Intelligence

Detection of Digital Manipulation in Facial Images ( Student Abstract )

With increasing usage of face recognition algorithms, it is well established that external artifacts and makeup accessories can be applied to different facial features such as eyes, nose, mouth and cheek, to obfuscate one&#x2019;s identity or to impersonate someone else&#x2019;s identity. Recognizing faces in the presence of these artifacts comprises the problem of disguised face recognition, which is one of the most arduous covariates of face recognition. The challenge becomes exacerbated when disguised faces are captured in real-time environment, with low resolution images. To address the challenge of disguised face recognition, this paper first proposes a novel multi-objective encoder-decoder network, termed as DED-Net. DED-Net attempts to learn the class variations in the feature space generated by both disguised as well non-disguised images, using a combination of Mahalanobis and Cosine distance metrics, along with Mutual Information based supervision. The DED-Net is then extended to learn from the local and global features of both disguised and non-disguised face images for efficient face recognition and the complete framework is termed as Disguise Resilient (D-Res) framework. The efficacy of the proposed framework has been demonstrated on two real-world benchmark datasets: Disguised Faces in the Wild (DFW) 2018 and DFW2019 competition datasets. In addition, this research also emphasizes on the importance of recognizing disguised faces in low resolution settings and proposes three experimental protocols to simulate the real-world surveillance scenario. To this effect, benchmark results have been shown on seven protocols for three low resolution settings (32&#x00D7;32, 24&#x00D7;24 and 16&#x00D7;16) of the two DFW benchmark datasets. The results demonstrate superior performance of the D-Res framework, in comparison with benchmark algorithms. For example, an improvement of around 3% is observed on the Overall protocol of the DFW2019 dataset, where the D-Res framework achieves 96.3%. Experiments have also been performed on benchmark face verification datasets (LFW, YTF and IJB-B), where the D-Res framework achieves improved verification accuracy.

IEEE Transactions on Circuits and Systems for Video Technology

Disguise Resilient Face Verification

Fulltext

IEEE Transactions on Biometrics, Behavior and Identity Science

TBIOM Special Issue on 'Best Reviewed Papers from IJCB 2020 - Editorial'

The security of Bitcoin heavily relies on the distributed consensus algorithm. There are many consensus algorithms used in different cryptocurrencies to provide security for transactions in the blockchain. Along with the consensus algorithm, the blockchain structure is also an important aspect that affects the performance of the cryptocurrency. However, most of the state-of-the-art schemes have some serious concern for scalability issues such as less throughput, high consensus delay, and high block interval time. The consensus algorithm and structure of blockchain have direct implications on these performance parameters. Therefore, we proposed a scheme called BlockTree, a sharding-based secure, scalable, and decentralized ledger scheme. Unlike Bitcoin, BlockTree is a nonlinear data structure based scheme, which partitioned the miners with fixed group size whenever the number of miners becomes more than the set limit called ``m{''}, and the blockchain is forked from that instance. Due to this partitioning, the BlockTree supports the parallel execution of the transaction, which has improved the throughput and consensus delay while keeping the high level of security. The BlockTree used the leader selection algorithm to select a leader and the data block verification algorithm to reach consensus among miners on the next set of the transactions to append blocks in the blockchain. The BlockTree is analyzed based on real-world parameters such as throughput, consensus delay, different block sizes, block generation intervals, and network propagation compared to existing technologies.

CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS

BlockTree: a nonlinear structured, scalable and distributed ledger scheme for processing digital transactions

Vehicular ad-hoc networks (VANETs) suffer from dis-connectivity due to very high mobility, node sparseness, and lossy link. Thus, the broadcasting algorithms in the vehicular multi-hop scenario have to operate in all extreme scenarios as dense, sparse, and disconnected topologies that are changeable dynamically due to different traffic conditions. The store-and-carry-forward scheme for disconnected topologies with dynamic nodes can be model as a Steiner-tree problem due to link weight change dynamically. The Steiner tree has to be recomputed to obtain the new Steiner tree, making it more difficult for broadcasting algorithms in VANETs. We show that broadcasting in VANETs (i.e., given graph G(V1, E1) with the terminal set, T1⊆ V1), modeled as to find as many (maximum) disjoint Steiner trees as possible that are disjoint on the non-terminal nodes and the edges are APX-hard even for small T1. If all the nodes in G are terminal nodes, i.e., T1= V1, the trouble is finding a maximum cardinality set of edge-disjoint spanning trees and encountering a considerable lot of edge-disjoint spanning tree. We then propose an O(log n) -Approximation Algorithm for Broadcasting in VANETs via bipartite graph (G= ((T1, V1/ T1) , E1)) and using projection. We compared the performance of the proposed scheme (i.e., EASBVN) with existing state-of-the-scheme and found it to be better than the state-of-the-art schemes (i.e., DV-CAST, UVCAST, DPPEMB, MoZo, and MobiVNDN) in terms of end-to-end delay and re-healing time. © 2020, Springer Science+Business Media, LLC, part of Springer Nature.

Wireless Networks

EASBVN: efficient approximation scheme for broadcasting in vehicular networks

In a permutation graph, vertices represent the elements of a permutation, and edges represent pairs of elements that are reversed by the permutation. In the Permutation Vertex Deletion problem, given an undirected graph G and an integer k, the objective is to test whether there exists a vertex subset S ⊆ V (G) such that |S| ≤ k and G−S is a permutation graph. The parameterized complexity of Permutation Vertex Deletion is a well-known open problem. Bożyk et al. [IPEC 2020] initiated a study towards this problem by requiring that G − S be a bipartite permutation graph (a permutation graph that is bipartite). They called this the Bipartite Permutation Vertex Deletion (BPVD) problem. They showed that the problem admits a factor 9-approximation algorithm as well as a fixed parameter tractable (FPT) algorithm running in time O(9k|V (G)|9). And they posed the question whether BPVD admits a polynomial kernel. We resolve this question in the affirmative by designing a polynomial kernel for BPVD. In particular, we obtain the following: Given an instance (G, k) of BPVD, in polynomial time we obtain an equivalent instance (G′, k′) of BPVD such that k′ ≤ k, and |V (G′)| + |E(G′)| ≤ kO(1) © Lawqueen Kanesh, Jayakrishnan Madathil, Abhishek Sahu, Saket Saurabh, and Shaily Verma; licensed under Creative Commons License CC-BY 4.0

Leibniz International Proceedings in Informatics, LIPIcs

A polynomial kernel for bipartite permutation vertex deletion

Parametric path problems arise independently in diverse domains, ranging from transportation to finance, where they are studied under various assumptions. We formulate a general path problem with relaxed assumptions, and describe how this formulation is applicable in these domains. We study the complexity of the general problem, and a variant of it where preprocessing is allowed. We show that when the parametric weights are linear functions, algorithms remain tractable even under our relaxed assumptions. Furthermore, we show that if the weights are allowed to be nonlinear, the problem becomes NP-hard. We also study the multi-dimensional version of the problem where the weight functions are parameterized by multiple parameters. We show that even with 2 parameters, this problem is NP-hard. © 2021 37th Conference on Uncertainty in Artificial Intelligence, UAI 2021. All Rights Reserved.

37th Conference on Uncertainty in Artificial Intelligence, UAI 2021

Generalized Parametric Path Problems

Trees are an integral part of many outdoor scenes and are rendered in a wide variety of computer applications like computer games, movies, simulations, architectural models, AR and VR. This has led to increasing demand for realistic, intuitive, lightweight and easy to produce computer-generated trees. The current approaches at 3D tree generation using a library of trees lack variations in structure and are repetitive. This paper presents an extended grammar-based automated solution for 3D tree generation that can model a wide range of species, both Western and Indian. For the foliage, we adopt a particle system approach that models the leaf, its size, orientation and changes. The proposed solution additionally allows control for individual trees, thus modelling the tree growth variations, changes in foliage across seasons, and leaf structure. This enables the generation of virtual forests with different tree compositions. In addition, a Blender add-on has been developed for use and will be released. © 2021 ACM.