Scientists from Cornell have created a novel form of intelligent attire capable of monitoring someone's posture and workout regimen. This garment appears, feels, and functions exactly like an ordinary shirt when worn and laundered.
SeamFit, the latest innovation, incorporates bendable conductive threads stitched into the neck, sleeve, and side seams of a regular short-sleeve T-shirt. This eliminates the need for manual logging during workouts since an AI system recognizes motions, determines the type of exercise, and tallies repetitions automatically. Once finished, all you have to do is detach a control panel located at the rear neckline and throw your now-damp top in with the laundry.
The team imagines that SeamFit might prove beneficial for athletes, fitness lovers, and individuals undergoing physical therapy.
According to Catherine Yu, a doctoral candidate in information science who led the research, most current body-tracking attire tends to be either snug and limiting or laden with bulky sensors.
Yu mentioned that they were curious about creating intelligent clothing that remains slim and functional, adding that enhancing usability was crucial so that individuals could wear these garments just as they normally do with regular clothes.
Instead, athletes might opt for fitness trackers such as smartwatches or rings; however, these additional gadgets could be something individuals prefer not to wear during workouts and lack the capability to monitor movements throughout the whole body.
"Everyone isn’t ready to experiment with an innovative wearable design, yet they all wear clothing,” explained co-author Cheng Zhang, who serves as an assistant professor of information science at the Cornell Ann S. Bowers College of Computing and Information Science. “Our product offers a sleek and convenient design that stays with you constantly.”
Their research titled "SeamFit: Moving Toward Realistic Intelligent Garments for Automated Workout Tracking," published in March in the ACM Transactions on Interactive, Mobile, Wearable and Ubiquitous Technology Proceedings , and will be showcased at the UbiComp/ISWC 2025 The meeting scheduled for October in Espoo, Finland.
Many ready-made garments feature seams, an aspect that Yu recognized as potentially useful for crafting comfy and budget-friendly intelligent attire. To create her prototype, she made three SeamFit tops in varying sizes: small, medium, and large. Using a regular sewing machine at home, she added conductive threads along these seams. These different size options enabled wearers to opt for either a more relaxed or snug fit; however, this variety also introduced complexity when analyzing each individual’s motions.
To test the shirts' performance, the team recruited 15 volunteers, who did a series of 14 exercises—including lunges, sit-ups and biceps curls—while wearing SeamFit. Without any calibration or training for each user, SeamFit's model classified the exercises with 93.4% accuracy and successfully counted reps, with counts that, on average, were off by less than one.
SeamFit functions due to the fact that during physical activity, the threads' capacitance—their capacity to hold an electric charge—alters as they shift, change shape, and come into contact with the human body. At the rear collarline, there’s a circuit board designed to measure these changes in capacitance before sending this data via Bluetooth to a computer. Subsequently, a specialized, compact system of signal processing along with machine learning analyzes these signals to interpret the movements accurately.
Following his exercises, Yu cleaned and dried the shirts at home.
This project represents a fresh version of SeamPose , an earlier attempt to monitor body positions with conductive threads across eight seams of a long-sleeved shirt.
The team imagines that this kind of inconspicuous intelligent apparel would be particularly beneficial for athletes tracking their workout regimens and for physical therapists overseeing the advancement of their patients remotely from home.
On a larger scale, this kind of technology has the potential to enhance human-AI interactions. By monitoring human actions and behaviors, AI systems can determine optimal times for engagement and periods when they should remain passive—for instance, during meals or sleep hours.
The primary objective at Zhang's Sci-Fi Lab is to equip AI with an understanding of human behavior through the development of innovative, AI-driven wearable technology. These devices aim to help AI grasp daily human actions and intents, offering assistance as required.
"Although this study focused on a basic article of clothing, we think it has the potential to be modified for various types of apparel and could benefit from incorporating intricate seaming techniques found in high-performance athletic wear," explained co-author François Guimbretière, who serves as a professor of information science at Cornell Bowers CIS and the multi-college Department of Design Tech.
To develop SeamFit, Yu established a “miniature sewing facility” within the laboratory. Nevertheless, she is now investigating ways to cost-effectively scale up the production process by utilizing industrial overlock machines—capable of stitching with three or four threads at once—as well as stronger conductive threads.
Just by substituting a single strand within this large-scale production method, every piece of attire could effortlessly gain smart capabilities including motion-tracking features," Yu explained. "I envision a future where you look into your wardrobe, and there’s essentially no distinction between intelligent and conventional garments.
More information: Tianhong Catherine Yu et al., SeamFit: Moving Toward Practical Smart Garments for Automated Exercise Tracking ACM Transactions on Interactive, Mobile, Wearable and Ubiquitous Technology Proceedings (2025). DOI: 10.1145/3712287
Provided by Cornell University
This tale was initially released on Tech Xplore . Subscribe to our newsletter For the most recent updates on science and technology news.