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

Synthesis of long-term human motion skeleton sequences is essential to aid human-centric video generation [8] with potential applications in Augmented Reality, 3D character animations, pedestrian trajectory prediction, etc. Long-term human motion synthesis is a challenging task due to multiple factors like, long-term temporal dependencies among poses, cyclic repetition across poses, bi-directional and multi-scale dependencies among poses, variable speed of actions, and a large as well as partially overlapping space of temporal pose variations across multiple class/types of human activities. This paper aims to address these challenges to synthesize a long-term (> 6000 ms) human motion trajectory across a large variety of human activity classes (> 50). We propose a two-stage activity generation method to achieve this goal, where the first stage deals with learning the long-term global pose dependencies in activity sequences by learning to synthesize a sparse motion trajectory while the second stage addresses the generation of dense motion trajectories taking the output of the first stage. We demonstrate the superiority of the proposed method over SOTA methods using various quantitative evaluation metrics on publicly available datasets. © 2021 IEEE.

World University root node

Proceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021

GlocalNet: Class-aware long-term human motion synthesis

arXiv e-prints

DILF-EN framework for Class-Incremental Learning

Improving Few-Shot Learning using Composite Rotation based Auxiliary Task

Diabetic retinopathy (DR) is a significant reason for the global increase in visual loss. Studies show that timely treatment can significantly bring down such incidents. Hence, it is essential to distinguish the stages and severity of DR to recommend needed medical attention. In this view, this paper presents DRISTI (Diabetic Retinopathy classIfication by analySing reTinal Images), where a hybrid deep learning model composed of VGG16 and capsule network is proposed, which yields statistically significant performance improvement over the state of the art. To validate our claim, we have reported detailed experimental and ablation studies. We have also created an augmented dataset to increase the APTOS dataset’s size and check how robust the model is. The five-class training and validation accuracy for the expanded dataset is 99.21 % and 75.50 %. The two-class training and validation accuracy on augmented APTOS is 99.96 % and 97.05 %. Extending the two-class model for the mixed dataset, we get a training and validation accuracy of 99.92 % and 91.43 % , respectively. We have also performed cross-dataset and mixed dataset testing to demonstrate the efficiency of DRISTI. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.

Fulltext

Signal, Image and Video Processing

DRISTI: a hybrid deep neural network for diabetic retinopathy diagnosis

An automatic system for analysis and retrieval of building floor plans images is helpful for the architects while designing new projects and providing recommendations to the buyers. For such systems, query by example is preferred over query by keyword, for which user's requirements must be available in digital image form. Floor plans are converted to digital form by scanning and often get rotated slightly by a certain degree of angle during digitization. In this paper, we have proposed a geometric feature-based approach for floor plan image retrieval and our key contribution is to handle different kinds of rotation and scale while retrieving similar floor plan from the database. Our framework is divided into three phases, namely outer shape feature extraction, internal object feature extraction, followed by matching and retrieval. For our experimentation, we have rotated images of ROBIN dataset as currently no rotated floor plan dataset was available. Our experiment shows that the proposed methodology outperforms recent competing methods. © 2021 Elsevier B.V.

Pattern Recognition Letters

A rotation and scale invariant approach for multi-oriented floor plan image retrieval

Recently, Jain et al. [IJCAI, 2019] studied the effect of project interactions in participatory budgeting (PB) by assuming an existing partition of the projects to interaction structures, namely a grouping of the projects into substitution and complementarity groups. Motivated by their study, here we take voter preferences to find such interaction structures. In our model, voters submit interaction structures, and the goal is to find an aggregated structure. Formally, given a set P of m projects, and n partitions of P, the task is to aggregate these n partitions into one aggregated partition. We consider this partition aggregation task both for substitution structures and for complementarity structures, studying several aggregation methods for each, including utility-based methods and Condorcet-based methods; we evaluate these methods by analyzing their computational complexity and their behavior with respect to certain relevant axiomatic properties. © 2021 International Foundation for Autonomous Agents and Multiagent Systems (www.ifaamas.org). All rights reserved.

Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS

Partition aggregation for participatory budgeting

Liquid Democracy (LD) uses transitive delegations in voting. In its simplest form, it is used for binary decisions, however its promise holds also for more advanced voting settings. Here we consider LD in the context of Participatory Budgeting (PB), which is a direct democracy approach to budgeting, most usually done in municipal budgeting processes. In particular, we study Knapsack Voting, in which PB voters approve projects, such that the sum of costs of projects each voter approves must respect the budget limit. We observe possible inconsistencies, as the cost of voter-approved projects may go over the budget limit after resolving delegations. We offer ways to overcome them by studying the computational complexity of updating as few delegations as possible to arrive - after following all project delegations - to a consistent profile. © 2021 International Foundation for Autonomous Agents and Multiagent Systems (www.ifaamas.org). All rights reserved.

Preserving consistency for liquid knapsack voting

Recent advances in swarm inspired optimization algorithms have shown its extensive acceptance in solving a wide range of different real-world problems. Particle Swarm Optimization (PSO) is one of the most explored nature-inspired population-based stochastic optimization algorithm. In this paper, a Multi-level Particle Swarm Optimization (MPSO) algorithm is proposed to find the architecture and hyperparameters of the Convolutional Neural Network (CNN) simultaneously. This automated learning will reduce the overhead of human experts to find these parameters through manual analysis and experiments. The proposed solution uses multiple swarms at two levels. The initial swarm at level-1 optimizes architecture and multiple swarms at level-2 optimize hyperparameters. The proposed method has used sigmoid like inertia weight to adjust the exploration and exploitation property of particles and avoid the PSO algorithm to prematurely converge into a local optimum solution. In this paper, we have explored an approach to suggest the best well-conditioned CNN architecture and its hyperparameters using MPSO in a specified search space. The complexity and performance of MPSO-CNN will depend on the dimension of the search space. The experimental results on 5 benchmark datasets of MNIST, CIFAR-10, CIFAR-100, Convex Sets, and MDRBI have demonstrated one more effective application of PSO in learning a deep neural architecture. © 2021

Swarm and Evolutionary Computation

Hybrid MPSO-CNN: Multi-level Particle Swarm optimized hyperparameters of Convolutional Neural Network

Skeletal Bone age assessment is a routine clinical procedure carried out by paediatricians and endocrinologists for investigating a variety of endocrinological, metabolic, genetic and growth disorders in children. Skeletal maturity advances with change in structure and size of the skeletal bones with respect to age. This is commonly done by radiological investigation of the left hand due to its non dominant use. Dissent in the skeletal age and bone age values indicates abnormality. In this study, a bone-age assessment model using triplet loss for children in 0–3 years of age is proposed. Furthermore, this is the first automated bone age assessment study on lower age groups with comparable results, using one tenth of the training data samples as opposed to conventional deep neural networks. We have used small number of radiographs per class from Digital Hand Atlas Database System (DHA), a publicly available comprehensive x-ray dataset. Model trained achieves an AUC of 0.92 for binary and 0.82 for multi-class classification with visible separation in embedding clusters; thereby resulting in correct predictions on test data set. © 2021, Springer Nature Switzerland AG.

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

Bone Age Assessment for Lower Age Groups Using Triplet Network in Small Dataset of Hand X-Rays

This article defines a new methodology for pre-recognition of events with object motion analysis in a video without any prior knowledge. This unsupervised application is named as ‘conceptualization’. This conceptualization technique is also tested with real-time video data in an internet of things (IoT) architecture. The merits of rough sets in the framework of granular computing are explored to execute the task. The proposed method is designed for the video sequences that are acquired by simple static RGB sensors. Here the video sequences are granulated with our newly defined ‘motion granules’ and then those are modeled as rough sets over this granulation for moving object/ background estimation. Video conceptualization is performed afterwards by quantifying the approximation with a new measure, namely, motion entropy. The values obtained by this measure reflect the amount of uncertainty present in the motion of each individual moving object which enables precognition of events. The effectiveness of the proposed method is verified with extensive experiments in identifying the different motion patterns present in a video sequence. The frames with possibilities of events present therein are identified with this analysis. Both offline and real-time sequences are used for this verification. An IoT architecture is formed to test the proposed algorithm with physical devices in identifying the frames containing possible events. © 2020