Professor Pai recognized for developing sustainable AI-powered fashion solution
Dr. Dinesh Pai, a professor in UBC Computer Science and graduate student, Megha Shastry, have won a competition for their research on AI within the fashion industry.
The Grand Challenge competition was themed around Generative AI and was organized by the 3D Retail Coalition (3DRC) and the IEEE.
Fashion, including wearable electronics, tremendously affects the environment. Estimates from a 2020 report1 state that the fashion industry produces four to five billion tonnes of global greenhouse gas emissions and 190,000 tonnes of oceanic microplastics per year, while using 79 trillion litres of water and vast quantities of textile waste. Upwards of 92 million tonnes per year, including unsold merchandise, ends up in landfills or is burnt.
Online shopping has dramatically improved the individual clothing shopping experience but contributes additional negative environmental impacts due to free and easy returns, estimated at two to four times that of in-store returns. Even if worn once to test fit, the product cannot be resold.
“Fit, in the garment industry, is a fundamental problem,” said Dr. Pai.
Sustainable solution
Pai, who directs UBC’s Sensorimotor Systems Lab, leverages AI to make adjustments to garment patterns, ultimately to reduce waste.
The critical task of adjusting a garment pattern to achieve good fit, such as removing pressure points, is currently done by human trial-and-error. Designers must iteratively adjust and estimate 2D pattern edits to achieve desired fit in 3D. This can be a frustrating and time-consuming process.
With a background in engineering, robotics, and physics, Pai applies his skills to develop real-world models that can make a difference.
“By applying principles of engineering and physics to computer graphics, we can simulate soft objects, like the human body, to improve garment fit.”
By using physics simulation research, Pai and his team can see how a garment will fit and drape on a human body by inputting measurements. They run simulations for thousands of garments by training the model to predict how a pattern should fit.
Helping AI fit into fashion
Pai’s aim is to take the guess work out of pattern development, improve fit, and reduce garment-industry waste.
Fit can be quantified using physical attributes such as ease and pressure on the body, that can be computed by 3D virtual fit testing software. Currently Pai and his research team are in the validation-testing phase of “uFit,” a novel fit-testing software that uses real garments draped on a mannequin. However, in as little as six months, mass customization using AI could be available.
“Why should we only have four or eight sizes available when the human body comes in innumerable shapes and sizes? Mass customization has completely new implications as we can now make clothing made-to-measure,” says Pai.
“Understanding how the human body works, and turning our skills to modeling provides a remarkable opportunity to vastly reduce waste in the garment industry,” says Pai.
It’s a grand challenge, for which he and his research team have earned several awards including the IEEE 3DRC Grand Challenge for 2023.
More about Dinesh Pai
More about Megha Shastry
Read the IEEE announcement
[1] Niinimäki, et al., The environmental price of fast fashion, Nat. Rev. Earth Environ 1, 189–200 (2020). https://doi.org/10.1038/s43017-020-0039-9