University of Illinois at Urbana Champagn
Brief Introduction : Professor Narendra Ahuja is a Research Professor in the Dept. of Electrical and Computer Engineering, Beckman Institute, and Coordinated Science Laboratory at University of Illinois at Urbana-Champaign (UIUC). During 2013-2019, he was the Founding Director of Information Technology Research Academy (ITRA), a national initiative started by the Ministry of Electronics and Information Technology, Government of India, to pilot a new model of enhancing the quality of advanced technical and related education across India (https://itra.digitalindiacorporation.in/). He has also served as the Founding Director of the first IIIT (IIITs are dedicated to IT and its applications), at Hyderabad. He is a fellow of IEEE, American Assoc for Artificial Intelligence, Intl. Association for Pattern Recognition, Assoc for Computing Machinery, American Assoc for the Advancement of Science, and Intl. Society for Optical Engineering. His research is in Artificial Intelligence fields of computer vision, pattern recognition, machine learning, image processing and their applications, including on problems in developing societies.
Title of the talk : Human Activity Recognition - Dance Modelling and Analysis
Abstract: Dance experts often view dance as a hierarchy of information, spanning low-level (raw images, image sequences), mid-levels (human poses and bodypart movements), and high-level (dance genre). We propose a Hierarchical Dance Video Recognition framework (HDVR). HDVR estimates 2D pose sequences, tracks dancers, and then simultaneously estimates corresponding 3D poses and 3D-to-2D imaging parameters, without requiring ground truth for 3D poses. Unlike most methods that work on a single person, our tracking works on multiple dancers, under occlusions. From the estimated 3D pose sequence, HDVR extracts body part movements, and therefrom the dance genre. The resulting hierarchical dance representation is explainable to experts. To overcome noise and interframe correspondence ambiguities, we enforce spatial and temporal motion smoothness and photometric continuity over time. We use an LSTM network to extract 3D movement subsequences from which we recognize dance genre. For experiments, we have identified 154 movement types, of 16 body parts, and assembled a new University of Illinois Dance (UID) Dataset, containing 1143 video clips of 9 genres covering 30 hours, annotated with movement and genre labels. Experimental results demonstrate that our algorithms outperform the state-of-the-art 3D pose estimation methods, which also enhances our dance recognition performance.
Research Interests : Computer Vision, Robotics, Image Processing, Sensors, Pattern Recognition, Virtual Environments, Intelligent Interfaces
School of Computing, National University of Singapore
Brief Introduction : Mohan Kankanhalli is Provost's Chair Professor of Computer Science at the National University of Singapore (NUS). He is also the Dean of NUS School of Computing. Before becoming the Dean in July 2016, he was the NUS Vice Provost (Graduate Education) during 2014-2016 and Associate Provost during 2011-2013. Mohan obtained his BTech from IIT Kharagpur and MS & PhD from the Rensselaer Polytechnic Institute. Mohan's research interests are in Multimedia Computing, Computer Vision, Information Security & Privacy and Image/Video Processing. He has made many contributions in the area of multimedia & vision - image and video understanding, data fusion, visual saliency as well as in multimedia security - content authentication and privacy, multi-camera surveillance.
He directs N-CRiPT (NUS Centre for Research in Privacy Technologies) which conducts research on privacy on structured as well as unstructured (multimedia, sensors, IoT) data. N-CRiPT looks at privacy at both individual and organizational levels along the entire data life cycle. N-CRiPT, which has been funded by Singapore's National Research Foundation, works with many industry, government and academic partners. He earlier directed the SeSaMe (Sensor-enhanced Social Media) Centre during 2012-2018. SeSaMe did fundamental exploration of social cyber-physical systems with applications in social sensing, sensor analytics and smart systems.
Mohan is a Fellow of IEEE.
Title of the talk : Privacy-aware Multimedia Analytics
Abstract :In this talk, we present our research on privacy-aware multimedia analytics. We will present three works covering different aspects of multimedia analytics. The first work is about privacy protection against machines. Utilizing machine learning and big data, algorithms often act as a tool for privacy violation, by automatically selecting content with sensitive information, such as photos that contain faces or vehicle license plates. The key idea is to perturb images using adversarial machine learning to protect image attributes privacy, while ensuring the images are not degraded. We conducted an experimental study to explore factors that influence human sensitivity to visual changes, which led to the concept of a human sensitivity map. Using this map, a human-sensitivity-aware image perturbation model is developed that can subtly alter an image such that sensitive attributes like gender are misclassified. The second work concerns privacy-preserving analytics on images. Attributes such as emotions, gender and age in images and videos are important for many applications. Existing methods extract this information from faces in the images. However, faces raise serious privacy concerns as they reveal people's identity. We first did an eye-tracking based human study of age, gender, and emotion prediction of people in images under various identity preserving scenarios - obfuscating eyes, lower face, head or the full face. Motivated by this study, we successfully developed a deep learning model for attributes prediction under privacy-preserving conditions and we present its results. The third work concerns training machine learning models where data sets cannot be shared due to privacy regulations (e.g., from medical studies). A simple yet unconventional approach for anonymized data synthesis can enable third parties to benefit from such valuable data. We propose learning implicitly from visually unrealistic, task-relevant stimuli, which are synthesized by exciting the neurons of a trained neural network. Neuronal excitation serves as a pseudo-generative model, and can be extended to inhibit representations that are associated with specific individuals, thus providing privacy. The stimuli data is then used to train new classification models. Experiments on MNIST and sleep apnea data show that these models offer protection against adversarial association and membership inference attacks. We will end with a general discussion on privacy concerns related to multimedia analytics.
Indian Institute of Technology, Jodhpur
Brief Introduction : Professor Santanu Chaudhury, Professor, Department of Electrical Engineering, IIT Delhi, has assumed charge as Director, IIT Jodhpur, on 10 December 2018. Professor Chaudhury holds B.Tech. (Electronics and Electrical Communication Engineering) and Ph.D. (Computer Science & Engineering) Degrees from IIT Kharagpur.
Professor Chaudhury joined as Faculty Member in the Department of Electrical Engineering, IIT Delhi, in 1992. He was Dean, Under-Graduate Studies at IIT Delhi. He has served as Director of CSIR-CEERI, Pilani, during 2016-18. Professor Chaudhury is a recipient of the Distinguished Alumnus award from IIT Kharagpur.
Professor Chaudhury is a Fellow of Indian National Academy of Engineers (INAE) and National Academy of Sciences (NAS). He is a Fellow of International Association Pattern Recognition (IAPR). He was awarded the INSA (Indian National Science Academy) Medal for Young Scientists in 1993. He received ACCS-CDAC award for his research contributions in 2012.
A keen researcher and a thorough academic, Professor Chaudhury has about 300 publications in peer reviewed journals and conference proceedings, 15 patents and 4 authored/edited books to his credit.
Norwegian University of Science and Technology (NTNU), Norway
Brief Introduction : Raghavendra Ramachandra obtained his Ph.D. in Computer Science and Technology from the University of Mysore, Mysore India and Institute Telecom, and Telecom Sudparis, Evry, France (carried out as a collaborative work) in 2010. He is currently appointed as a Full Professor at Institute of Information Security and Communication Technology (IIK), Norwegian University of Science and Technology (NTNU), Gjøvik, Norway. He was a Researcher with the Istituto Italiano di Tecnologia, Genoa, Italy, where he worked with video surveillance and social signal processing. His main research interests include Deep learning, applied machine learning, statistical pattern recognition, data fusion schemes, and random optimization, with applications to biometrics, multimodal biometric fusion, human behavior analysis, and crowd behavior analysis. He has authored several papers, holds patents and is a reviewer for several international conferences and journals. He was/is also involved in various conference organizing and program committees and serving as an associate editor for various journals. He was/is participating (as PI/Co-PI/contributor) in several EU projects, IARPA USA and other national projects. He is serving as an editor for ISO/IEC 24722 standards on multimodal biometrics.
He has received several best paper awards, and he is also a senior member of IEEE.
Title of the talk :Face Morphing Attacks on Face Biometric Systems : Vulnerability and Detection
Abstract :Face morphing is the process of seamlessly combining the face images from different data subjects to generate the composite (or morphing) image that can share the facial characteristics from the contributing data subjects. Face morphing attacks have shown high vulnerability on the commercial face recognition system. In this keynote, deep learning-based based morphed attack generation will be presented. Further, 3D face morphing generation will also be discussed. Lastly, part of the talk will focus on the detection of face morphing attacks.
Indian Institute of Technology, Delhi
Brief Introduction :Sumantra Dutta Roy is a B.E. (Computer Engineering) from D.I.T., Delhi (1993), and completed his M.Tech and Ph.D. degrees at the Department of Computer Science and Engineering, I.I.T. Delhi, in 1995 and 2001, respectively. He started his career in teaching and research in the Department of Electrical Engineering at I.I.T. Bombay, where he worked from 2001 to early 2007 as an Assistant Professor. From 2007 to 2018, he was an Associate Professor in the Department of Electrical Engineering at I.I.T. Delhi. Since 2018, he has been a Professor in the Department of Electrical Engineering at I.I.T. Delhi. He is a recepient of 2004 INAE Young Engineer Award (Indian National Academy of Engineering), and the 2004 - 05 BOYSCAST Fellowship of the Department of Science and Technology, Government of India. He has been an Associate Editor of the Pattern Recognition Letters since 2011. His research interests are in Computer Vision and Image Analysis, Video and Image Coding, Biometrics, Music Information Retrieval, and Medical Informatics.
Title of the talk :Millenial Computer Vision
Abstract :The talk will focus on some developments in the field of Computer Vision starting around the Millenium, which have entered technologies in our day-to-day life. The talk will feature developments in geometry, getting more out of ordinary cameras, and some applications of machine learning.
Research Interests : Computer Vision and Image Analysis, Pattern Recognition, Music Information Retrieval and Analysis, Biometrics, Bioinformatics
Indian Institute of Technology, Indore
Brief Introduction : Dr. Somnath Dey is currently working as an Associate Professor in the Department of Computer Science & Engineering at the Indian Institute of Technology Indore (IIT Indore). He is also holding the position of Head of Department of Computer Science & Engineering at IIT Indore. Earlier, he was Associate Dean Administration at IIT Indore.
He received his B. Tech. degree in Information Technology from the University of Kalyani in 2004. He completed his M.S. (by research) and Ph.D. degree in Information Technology from the School of Information Technology, Indian Institute of Technology Kharagpur, in 2008 and 2013, respectively.
He has published over 40 research articles (including papers in international journals, conferences and book chapters). He is a recipient of the Young IT Professional Award (Eastern Region Level) of Computer Society of India. He is actively involved in academic and industrial collaboration for research and development. His research interest includes biometric security, image processing and pattern recognition. He has a vast experience of more than 7 years and he has served as TPC chair/member for several reputed conferences.
Title of the talk : Feature Extraction in Fingerprint Liveness Detection
Indian Institute of Technology, Patna
Brief Introduction : Dr. Jha completed his M. Tech. and Ph. D. Degree from Indian Institute of Technology (IIT) Kharagpur in the area of Image Processing in January 2003 and January 2010 respectively.
M.Tech Thesis Title: "A Novel Approach for Watermarking of Digital Images".
PhD Thesis Title: “Improvement of Image Enhancement Segmentation and Watermark Detection Using Stochastic Resonance”. He has published more than 20 top Journals and 23 top conferences papers till date and many papers are in review process.
Visiting Scientist in the Department of Information and Communication Engineering, The Faculty of Engineering, The University of Tokyo Japan.
Research Interests: Image and Video Processing, Medical Image Processing. Computer Vision, pattern recognition and multimedia systems. Stochastic resonance.
Lead Speech Scientist, Uniphore
Brief Introduction : Bidisha Sharma a post-doctoral research fellow at Electrical and Computer Engineering Department, National University of Singapore, Singapore. Area Of research interest is specifically in the area of speech processing including text-to-speech synthesis, speech enhancement, speech recognition and language processing. Speaker Is also experienced in different applications of music and singing voice technology and also enthusiastic for active involvement in applied science and audio related projects.
Research Engineer, IBM Research AI
Brief Introduction : Saneem Chemmengath is a research scientist at IBM Research. His interests are in core machine learning, computer vision and natural language processing. In IBM he worked on both fundamental and specific applications of machine learning including recommender systems, satellite images, solar panels, question answering systems, credit card analytics, etc. Saneem has been working in the areas of AI and ML for the past 10+ years with papers in top ML conferences such as NeurIPS, UAI, EMNLP, NAACL. Prior to joining IBM he did his Master's by research from IISc, Bangalore.
Title of the talk : Question Answering Systems
Abstract : Question Answering is a specialized form of information retrieval. Given a collection of documents in the form of passages or tables, a Question Answering system attempts to retrieve correct answers to questions posed in natural language. In this session, I will briefly talk about the important concepts in question answering, their applications and state of the art models using transformers. I will also cover our recent works on answering questions over tables and how to make them work on domain specific documents.
Assistant Professor, LNMIIT
Brief Introduction : Dr. Saurabh Kumar has received his Bachelor of Engineering (B.E.) degree in Computer Science and Engineering from Chhattisgarh Swami Vivekanand Technical University, Bhilai, India, in 2010, M. Tech in Computer Science from Birla Institute of Technology, Mesra, Ranchi, India, in 2013, and PhD in Computer Engineering from Sardar Vallabhbhai National Institute of Technology, Surat, Gujarat, India, in 2019. He has been working on the problems of optimized deployment strategy and localization of nodes and events using IoT for post disaster management. He has successfully published 13 conference papers and 4 journals with leading publication houses like ACM, IEEE, Springer, Elsevier, and Oxford University Press in these fields. He has been an active researcher in the fields of Internet of Things, Collaborative data processing, multi-agent systems and wireless sensor networks. Recently, he was conferred with 2020 Wilkes Award for best paper (runner-up) by Oxford University Press for his novel work on localization of events for post disaster management. In the past, he served as an IT Officer for two years with mining business of the Aditya Birla Group. He is currently serving as an Assistant Professor in the Department of Computer Science and Engineering at The LNM Institute of Information Technology, Jaipur, Rajasthan, India.
Title of the talk : Multiagent Systems: Learning Entities and Implementation
Assistant Professor, LNMIIT
Brief Introduction : Dr. Ram Prakash Sharma is currently working as an Assistant Professor in the Department of Computer Science & Engineering at the Laxmi Niwas Mittal Institute of Information technology (LNMIIT), Jaipur. He received his Ph.D. degree in Computer Science and Engineering from Indian Institute of Technology Indore (IITI) in 2020. He completed his masters from University of Hyderabad and Bachelors from Rajasthan Technical University. His research interest includes biometric security, image processing, machine learning, and pattern recognition. He has published various research articles in international journals and conferences.
Title of the talk : Feature extraction of fingerprint images using NBIS tool
Assistant Professor, LNMIIT
Brief Introduction : Indra Deep Mastan is an Assistant Professor in the Department of Computer Science and Engineering at LNMIIT Jaipur. He has completed his Ph.D. from CSE at IIT Gandhinagar under the guidance of Dr. Shanmuganathan Raman. He is working on Computer Vision using Deep Learning. His interests are in Image Restoration and Image Synthesis problems.
Title of the talk : Generative Adversarial Networks for Image Synthesis
Abstract : Generative Adversarial Networks (GANs) are powerful deep learning tools that have attracted exciting applications such as synthesis of realistic faces and videos. In this talk, we provide an overview of GANs and various famous GAN models for image synthesis tasks.