Vida Adeli

I'm a Computer Science PhD candidate at University of Toronto, supervised by Dr. Babak Taati and Dr. Andrea Iaboni. I am also a Faculty Affiliate Researcher at Vector Institute and a Research Assistant at KITE - University Health Network. Previously, I was a Researcher at Vision & Learning for Autonomous AI (VL4AI) Lab, Monash University. I'm currently doing an internship at Pickford AI, focusing on video generation and real-time 3D animation generation.

Research

My research focuses on generative models applied across a range of computer vision tasks, with particular focus on 3D human motion understanding, realistic animation generation, and clinical gait analysis.

Selected publications

	PickStyle: Video-to-Video Style Transfer with Context-Style Adapters Soroush Mehraban, Vida Adeli, Jacob Rommann, Babak Taati Kyryl Truskovskyi * Equal contribution arXiv Project page / arXiv PickStyle is a diffusion-based video style transfer framework that preserves video context while applying a target visual style. It uses low-rank style adapters and synthetic clip augmentation from paired images for training, and introduces Context-Style Classifier-Free Guidance (CS-CFG) to independently control content and style, achieving temporally consistent and style-faithful video results.
	CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson's Disease Gait Assessment Vida Adeli, Soroush Mehraban, Ivan Klabucar, Javad Rajabi, Benjamin Filtjens, Soroush Mehraban, et al. Neural Information Processing Systems (NeurIPS), 2025 Project page / arXiv / Dataset CARE-PD is the largest publicly available archive of 3D mesh gait data for Parkinson’s Disease, collected across 9 cohorts from 8 clinical centers. It provides standardized, anonymized SMPL representations and benchmark protocols for clinical motion analysis on PD.
	GAITGen: Disentangled Motion-Pathology Impaired Gait Generative Model Vida Adeli, Soroush Mehraban, Majid Mirmehdi, Alan Whone, Benjamin Filtjens, Amirhossein Dadashzadeh, Alfonso Fasano, Andrea Iaboni, Babak Taati arXiv project page / arXiv GAITGen is a generative framework that synthesizes realistic gait sequences conditioned on Parkinson’s severity. Using a Conditional Residual VQ-VAE and tailored Transformers, it disentangles motion and pathology features to produce clinically meaningful gait data. GAITGen enhances dataset diversity and improves performance in parkinsonian gait analysis tasks.
	Benchmarking Skeleton-based Motion Encoder Models for Clinical Applications: Estimating Parkinson's Disease Severity in Walking Sequences Vida Adeli, Soroush Mehraban, Irene Ballester, Yasamin Zarghami, Andrea Sabo, Andrea Iaboni, Babak Taati IEEE International Conference on Automatic Face and Gesture Recognition (FG), 2024 Code / arXiv Evaluating recent motion encoders for the task of parkinsonism severity estimation (UPDRS III gait)
	MotionAGFormer: Enhancing 3D Pose Estimation with a Transformer-GCNFormer Network Soroush Mehraban, Vida Adeli, Babak Taati Winter Conference on Applications of Computer Vision (WACV), 2024 Code / video / arXiv Estimating 3D locations of 17 main joints from a monocular video.
	Ambient Monitoring of Gait and Machine Learning Models for Dynamic and Short-Term Falls Risk Assessment in People With Dementia Vida Adeli, Soroush Mehraban, Navid Korhani, Andrea Sabo, Sina Mehdizadeh, Alastair Flint, Avril Mansfield, Andrea Iaboni, Babak Taati IEEE journal of biomedical and health informatics (JBHI), 2023 arXiv A machine learning system using ambient gait monitoring to predict short-term fall risk in people with dementia.
	TRiPOD: Human Trajectory and Pose Dynamics Forecasting in the Wild Vida Adeli, Mahsa Ehsanpour, Ian Reid, Juan Carlos Niebles, Silvio Savarese, Ehsan Adeli, Hamid Rezatofighi IEEE/CVF International Conference on Computer Vision, (ICCV 2021) Project Page / arXiv / Video / Benchmark / ICCV2021 Workshop TRiPOD jointly forecasts human body pose dynamics and global trajectories in the wild by modelling human-human and human-object interactions.
	Socially and Contextually Aware Human Motion and Pose Forecasting Vida Adeli, Ehsan Adeli, Ian Reid, Juan Carlos Niebles, Hamid Rezatofighi IEEE Robotics and Automation Letters (RA-L) and IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2020 arXiv/ Video A unified end-to-end model that predicts both human global motion (trajectory) and detailed body pose jointly, using social and scene context to improve forecasting.
	A component-based video content representation for action recognition Vida Adeli, Ehsan Fazl-Ersi, Ahad Harati, Image and Vision Computing, 2019 Paper Proposes a component-based video content representation for human action recognition, decomposing videos into meaningful parts to improve classification in complex scenes.

Benchmarks and Workshops

CARE-PD Benchmark
SoMoF: Created SOcial MOtion Forecasting Benchmark (SoMoF), a key resource for human motion forecasting research.
ICCV2021 Workshop, Workshop and Challenge on Human Trajectory and Pose Dynamics Forecasting in the Wild.

Honors and Awards

Best Presentation Award, AI and Demenia International Workshop, AGE-WELL (2025)
Beatrice "Trixie" Worsley Graduate Scholarship In Computer Science (2025)
Outstanding Reviewer, IEEE Conference on Computer Vision and Pattern Recognition (CVPR2022)
Mount Sinai Hospital Graduate Scholarship in Science and Technology (2022)

Teaching Assistant

Introduction to Image Understanding (CSC420)
Foundations of Computational Vision (CSC2503)
Introduction to Visual Computing (CSC320)
Applied Fundamentals of Deep Learning (APS360)

Academic Service

Member of Reviewing Committee: CVPR, ICCV, ECCV, IROS, JBHI, ICLR, WACV, NeurIPS

Website Template source code.