Advancing Visual Intelligence: Innovations Across Images, Videos, and Point Clouds

Mrigank Rochan, Assistant Professor, Department of Computer Science, University of Saskatchewan, Canada

– 5 December 2023

Talk summary: As the demand for advanced computer vision applications continues to grow, there is a pressing need to improve the understanding and interpretation of visual data. In this talk, Mrigank Rochan presented his group’s efforts to push the boundaries of visual intelligence across multiple modalities, including images, videos, and point clouds, enabling more accurate and efficient analysis of diverse visual content. Firstly, he introduced their method, which can automatically localise the object in an image associated with a user-generated textual tag. Secondly, he described their work towards the automatic creation of a short visual summary or highlight of a long input video, allowing users to easily preview, search, and edit ever-growing video data. Thirdly, he discussed their research on robust visual perception systems in autonomous driving, focusing specifically on LiDAR point cloud semantic segmentation. Finally, he concluded with some interesting future directions.

Speaker bio: Mrigank Rochan is an Assistant Professor in the Department of Computer Science at the University of Saskatchewan, Canada. He earned his PhD in Computer Science from the University of Manitoba in 2020. His research interests lie in computer vision and machine learning. He has published papers in top-tier computer vision and robotics conferences and journals, including CVPR, ICCV, ECCV, ICRA, and TPAMI. He received the 2020 Canadian Image Processing and Pattern Recognition Society (CIPPRS) John Barron Doctoral Dissertation Award for his doctoral dissertation on deep learning models for video abstraction, a prestigious national award given annually to the top PhD thesis defended in computer/robot vision at a Canadian university. 

[Talk organised in collaboration with the Department of Computational and Data Sciences]