Paritosh
Parmar


I am a Computer Vision Scientist at A*STAR, Singapore. Prior to that, I was a Postdoctoral Research Fellow at the University of British Columbia, Vancouver. University of Nevada, Las Vegas was a great place to do my PhD. My research interests primarily lie in video understanding/video analysis (action recognition, action quality assessment, sports performance analysis, video captioning/description/QA, action anticipation, etc.).

I am always open to collaboration, so feel free to contact me if you are looking to collaborate.

                   

paritosh parmar profile photo
Research
CausalChaos! Dataset for Comprehensive Causal Action Question Answering Over Longer Causal Chains Grounded in Dynamic Visual Scenes
Ting En Lam, Yuhan Chen, Elston Tan, Eric Peh, Ruirui Chen, Paritosh Parmar, Basura Fernando
Preprint, 2024  
arxiv | dataset (coming soon) | code (coming soon)

TL;DR: We capitalize on cartoons and their properties to create a novel, challenging, focused dataset for causal video question answering; study the performance of various models; show improvements on real-world datasets.

Hierarchical NeuroSymbolic Approach for Action Quality Assessment
Lauren Okamoto, Paritosh Parmar
CVPR 2024 CVSports, 2024   (🏆 Best Paper Award; Oral Presentation; 3/3 Strong Accepts; Selected for CVPR 2024 Demos)  
arxiv | dataset | code

TL;DR: We propose a Hierarchical Neurosymbolic approach for comprehensive and explainable action analysis.

Learning to Visually Connect Actions and their Effects
Eric Peh, Paritosh Parmar, Basura Fernando
Preprint, 2024  
arxiv | dataset (coming soon) | code (coming soon)

TL;DR: Video understanding has made tremendous progress. However, it has not been explored if video understanding can connect actions and their effects. In this pilot study, we introduce the concept of connecting actions and their effects in video understanding literature.

Domain Knowledge-Informed Self-Supervised Representations for Workout Form Assessment
Paritosh Parmar, Amol Gharat, Helge Rhodin
ECCV, 2022  
arxiv | dataset | code

TL;DR: We proposed three self-supervised approaches to learn pose-sensitive representations for workout form assessment: Pose Contrastive learning; Disentangling Erroneous Motion from Regular Motion; Disentangling Pose and Appearance.

Win-Fail Action Recognition
Paritosh Parmar, Brendan Morris
WACV Workshops, 2022  
arxiv | dataset

TL;DR: Trying to shift focus from just recognizing patterns to gaining truer understanding of videos by teaching CNNs to differentiate between successful and failed attempts at various tasks.

Piano Skills Assessment
Paritosh Parmar, Jaiden Reddy, Brendan Morris
MMSP, 2021  
arxiv | dataset | media coverage

TL;DR: Automated, data-driven approach to assessing piano skills.

HalluciNet-ing Spatiotemporal Representations Using a 2D-CNN
Paritosh Parmar, Brendan Morris
Signals, 2(3), 2021  
arxiv | code

TL;DR: Approximating a 3DCNN with a 2DCNN, and the benefits of doing so.

What and How Well You Performed? A Multitask Learning Approach to Action Quality Assessment
Paritosh Parmar, Brendan Morris
CVPR, 2019   (Oral Presentation)
app | arxiv | dataset | code

TL;DR: Learning to mimic commentators can help better mimic judges.

Action Quality Assessment Across Multiple Actions
Paritosh Parmar, Brendan Morris
WACV, 2019
arxiv | dataset | code

TL;DR: It is better to simultaneously learn to mimic human judges from multiple sports.

Learning to Score Olympic Events
Paritosh Parmar, Brendan Morris
CVPR Workshops, 2017   (Oral Presentation; 🏆 People's Choice-3rd Place)
arxiv | dataset | code | external coverage

TL;DR: Mimic human judges.

Measuring the Quality of Exercises
Paritosh Parmar, Brendan Morris
EMBC, 2016
arxiv

TL;DR: To facilitate the collection of physiotherapy assessment dataset without a physiotherapist, we treat AQA as a classification problem, instead of a regression problem. Further, we conduct a feasibility study on the same.




Professional Service
Conferences CVPR ECCV WACV ICCV NeurIPS AAAI IROS International Conference on 3D Vision (3DV) Conference on Robots and Vision (CRV) International Conference on Intelligent Transportation Systems (ITSC) Intelligent Vehicles Symposium (IV) Midwest Symposium on Circuits and Systems (MWSCAS)
Journals TPAMI IJCV Machine Vision and Applications (MVA) Transactions on Circuits and Systems for Video Technology (TCSVT) Journal of Electronic Imaging (JEI) Robotics and Automation Letters (RA-L) Sensors Applied Sciences Sports Engineering Journal of New Music Research



Team
University of British Columbia Ting En Lam (Intern, NTU CS Undergrad),     Jan'23 - Jul'23

Yuhan Chen (Intern, NUS EE Undergrad),     Jan'23 - Jun'23

Elston Tan (Intern, SP AI Undergrad),     Dec'22 - Dec'23

Eric Peh (Research Engineer),     Aug'22 - Ongoing
University of British Columbia Deyu (Jerry) Liu (CS Undergrad),     Fall'21, Spring'22

Wei Zhi (Matthew) Tang (CS Undergrad),     Summer'21



Teaching
University of Nevada, Las Vegas Spring'19:     Signals & Systems I - Discussion Circuits I - Discussion
Fall'18:     Digital Signal Processing - Lab Electronics I - Lab
Spring'18:     Circuits II - Lab
Fall'17:     Digital Signal Processing - Lab
Spring'17:     Signals & Systems I - Discussion
Spring'16:     Computer Logic Design II - Discussion
Fall'15:     Computer Logic Design I - Lab

based on a template by Jonathan Barron