Abhinav Shrivastava

Abhinav Shrivastava is an Associate Professor in the Department of Computer Science at University of Maryland, College Park, with a joint appointment in the Institute of Advanced Computer Studies (UMIACS). Previously, he served as an Assistant Professor at the same institution from August 2018 to June 2024, and spent one year as a Visiting Research Scientist at Google Research from September 2017 to August 2018.

His educational background includes:

• PhD in Robotics and Artificial Intelligence from Carnegie Mellon University (2017), advised by Abhinav Gupta, with thesis titled 'Discovering and Leveraging Visual Structure for Large-scale Recognition'
• MS in Artificial Intelligence from Carnegie Mellon University (2011), supervised by Alyosha Efros and Martial Hebert
• BTech in Computer Science and Engineering from Jaypee Institute of Information Technology (2010)

Professor Shrivastava's research focuses on computer vision and machine learning, with particular expertise in object detection, image recognition, and neural representations. His work bridges theoretical advances with practical applications, exploring how visual systems can discover and leverage structure in large-scale recognition problems. He has made significant contributions to understanding the role of supervision in vision transformers, developing novel approaches for object-state composition recognition, and creating efficient neural representations for videos and 3D scenes. His research often addresses fundamental challenges in visual recognition, including handling novelty in open-world environments and improving the efficiency of visual systems.

An analysis of his recent publications reveals a strong emphasis on neural representations, particularly for dynamic content like videos and 3D scenes. His work demonstrates increasing sophistication in handling open-world vision problems, with research spanning object discovery, localization, and representation learning. The publications show a clear progression toward more efficient and scalable models, with recent work focusing on model compression, sparse representations, and addressing the challenges of working with limited annotations.

His scientific contributions have been recognized with several prestigious awards:

• Best Paper Award (Applications) at IEEE Winter Conference on Applications of Computer Vision (2020)
• Microsoft Research PhD Fellowship (2014-2016)
• Best Student Paper Award at IEEE Winter Conference on Applications of Computer Vision (2014)
• Outstanding Reviewer Award at IEEE CVPR (2015)

Professor Shrivastava has successfully mentored numerous graduate students, many of whom have become prominent researchers in computer vision. His Amazon Research Awards (2020 and 2023) have supported innovative projects including 'The pursuit of knowledge: discovering and localizing new concepts using dual memory' and 'Audio-conditioned Diffusion Models for Generating Lip-synchronized Videos.' He has served as Area Chair for major conferences including ICCV, CVPR, and AAAI, demonstrating his leadership in the computer vision community. His research has attracted significant funding from both academic and industry sources, supporting his exploration of fundamental questions in visual recognition and representation learning.