Fashion & ApparelFashion Retailer, KSA
High return rates from sizing uncertainty were eroding margins and customer confidence.
45%
Return Reduction82%
Satisfaction28%
Revenue GrowthImpact Stories
Representative outcomes from retailer deployments across fashion, activewear, and luxury.
0%
Average Return Reduction
Across deployed clients
0%
Customer Satisfaction
Reported fit confidence uplift
+0%
Revenue Impact
Average growth after adoption
< 2 weeks
Implementation Time
From kickoff to launch
Featured Success Stories
Measurable Results,
Fashion & Apparel
Sports & ActivewearActivewear Brand, UAE
Customers struggled to choose performance-wear sizes confidently online.
52%
Return Reduction91%
Satisfaction35%
Revenue GrowthMore Success Stories
Real Industries
Luxury FashionLuxury Fashion House, GCC
A premium online experience demanded accurate, trustworthy sizing with no room for error.
38%
Return Reduction88%
Satisfaction41%
Revenue GrowthResearch
Backed by Science
Our measurement engine is grounded in published research on body pose estimation, computer vision, and e-commerce sizing behaviour.
Computer Vision2023
Accurate Full-Body Measurement from a Single Monocular Image
We present a lightweight convolutional approach to extracting 12 body measurements from a single front-facing photograph, achieving sub-centimetre accuracy on a held-out validation set of 4,200 subjects across diverse body types and clothing conditions.
MIQYAS Research Team
E-Commerce2023
Reducing E-Commerce Return Rates Through AI-Assisted Size Recommendation
A controlled study across three apparel retailers shows that integrating automated body-measurement-based size recommendations reduces size-related return rates by an average of 43%, with the strongest effect observed in fitted categories such as outerwear and performance wear.
MIQYAS Research Team
Machine Learning2022
Robust 2D Pose Estimation for Garment Sizing in Unconstrained Environments
We propose a pose-estimation pipeline optimised for real-world sizing use cases: variable lighting, occlusion by clothing, and non-studio backgrounds. Benchmarked on our proprietary dataset of 11,000 annotated images, the model outperforms general-purpose estimators on sizing-relevant keypoints.
MIQYAS Research Team
Impact Stories
Ready to Transform Your Business?
Deploy MIQYAS with your catalog and sizing logic.