MV-Fashion: Towards Enabling Virtual Try-On and Size Estimation with Multi-View Paired Data
Abstract
Existing 4D human datasets often fall short for fashion-specific research, lacking either realistic garment dynamics or task-specific annotations. To bridge this gap, we introduce MV-Fashion, a massive multi-view video dataset engineered for domain-specific fashion analysis.
MV-Fashion captures complex, real-world garment dynamics across 80 diverse subjects wearing multiple layered outfits. Crucially for Virtual Try-On (VTON) applications, it provides paired data: synchronized multi-view captures of worn garments alongside their corresponding flat, catalogue images.
Dataset Highlights
Explore the diversity and quality of the captured data, designed to push the boundaries of human-centric rendering and virtual try-on.
Diverse Poses
Capturing a wide range of natural and complex human motions.
Layered Outfits
Intricate details of multi-layered clothing combinations.
Multi-view Consistency
Synchronized capture ensuring perfect alignment across all views.
Paired Data
Catalogue domain image pairs for the multi-view recordings for VTON.
Challenging Garments
Includes difficult items like loose dresses and transparent fabrics.
Robust Tracking
Accurate SMPL-X fitting and tracking.
80 Subjects
Diverse pool of participants (50.6% male, 45.7% female) across various BMI and age distributions.
754 Garments
Spanning 14 distinct fashion categories, comprising single, double, and triple-layered outfits.
Paired VTON Data
Unique paired data featuring synchronized multi-view captures of worn garments with corresponding flat catalogue images.
68 Synchronized Cameras
60 RGB global shutter and 8 Depth/4K cameras capturing real-world dynamic deformations.
Rich Annotations
Features precise SMPL-X fits, 3D point clouds, text descriptions, and segmentation masks.
3,273 Sequences
Extensive multi-view video database yielding over 72.5 million high-fidelity frames.
Get Started
Download the dataset and run the initial evaluation scripts with just a few commands.
# Clone the repository
git clone https://github.com/HunorLaczko/MV-Fashion.git
cd MV-Fashion
# Install dependencies
pip install -r requirements.txt
# Download the sample dataset
bash scripts/download_sample.sh
Citation
If you find our work useful in your research, please consider citing:
@InProceedings{Laczko_2026_CVPR,
author = {Laczk'o, Hunor and Jia, Libang and Truong, Loc-Phat and Hern'andez, Diego and Escalera, Sergio and Gonzalez, Jordi and Madadi, Meysam},
title = {MV-Fashion: Towards Enabling Virtual Try-On and Size Estimation with Multi-View Paired Data},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2026},
pages = {42810-42823}
}
