SeamPose: Repurposing Seams as Capacitive Sensors in a Shirt for Upper-Body Pose Tracking

TL;DR

SeamPose introduces a minimally obtrusive approach to upper-body pose tracking by repurposing existing shirt seams as capacitive sensors. A proof-of-concept long-sleeve shirt with eight conductive seams feeds a tailored deep-learning pipeline that estimates 3D joint positions relative to the pelvis, achieving an MPJPE of 6.0 cm in a 12-person study. The method preserves the garment’s aesthetics and comfort while delivering competitive tracking performance against prior wearable systems. Key contributions include a fabrication workflow for conductive seams, a compact 8-channel sensing board, and a two-stage training regime (user-independent plus user-adaptive) to handle inter-user variability. The work lays groundwork for scalable, everyday wearable pose tracking, with future work addressing broader garment patterns, real-world deployment, and washable, manufacturable designs.

Abstract

Seams are areas of overlapping fabric formed by stitching two or more pieces of fabric together in the cut-and-sew apparel manufacturing process. In SeamPose, we repurposed seams as capacitive sensors in a shirt for continuous upper-body pose estimation. Compared to previous all-textile motion-capturing garments that place the electrodes on the clothing surface, our solution leverages existing seams inside of a shirt by machine-sewing insulated conductive threads over the seams. The unique invisibilities and placements of the seams afford the sensing shirt to look and wear similarly as a conventional shirt while providing exciting pose-tracking capabilities. To validate this approach, we implemented a proof-of-concept untethered shirt with 8 capacitive sensing seams. With a 12-participant user study, our customized deep-learning pipeline accurately estimates the relative (to the pelvis) upper-body 3D joint positions with a mean per joint position error (MPJPE) of 6.0 cm. SeamPose represents a step towards unobtrusive integration of smart clothing for everyday pose estimation.

Figures (10)

• Figure 1: The seams and patterns of our proof-of-concept prototype with a long-sleeve T-shirt. The black fabric pieces represent the patterns of the T-shirt, while the dotted lines are the repurposed seams, also indicating where the stitches are when joining the fabric pieces. The color of the dotted lines can be mapped to the lines, or seams, on the constructed T-shirt.
• Figure 2: Machine Sewing Conductive Thread. (A) Home sewing sewing machine setup. (B) The off-the-shelf insulated conductive thread in use. (C) An illustration of machine-sewn conductive thread traces over existing seams, overlapping areas formed when stitching (by the original seam stitches) pieces of fabric together. (D) A side-view illustration of how the TPU-insulated conductive thread with silver nylon core is stitched onto the fabric.
• Figure 3: The battery-powered customized sensing board is housed inside a 3D-printed PLA case and hot-glued onto the prototype below the neck.
• Figure 4: Example SeamPose Signals. We show 19 seconds of continuous SeamPose the left and right seam signals (4 each) as the wearer performs symmetric (both arm raises) arm movements, asymmetric arm movements (right arm curls, left front raises, alternate punches, confused gesture), head movements, and standing still. The colors map the signals and their corresponding electrode placements, illustrated on the left. The numerical value of signal on $y$-axis is median normalized.
• Figure 5: Customized Deep Learning Pipeline.