TY - GEN
T1 - MagFace
T2 - 2026 CHI Conference on Human Factors in Computing Systems, CHI 2026
AU - Wang, Guanyun
AU - Zhang, Yifu
AU - Zheng, Xianzhe
AU - Fu, Huaqian
AU - Qi, Fanke
AU - Ye, Zhenxuan
AU - Zhai, Ruoyu
AU - Zhu, Yinzhen
AU - Fan, Yitao
AU - Yang, Yue
AU - Wang, Qi
AU - Tao, Ye
AU - Song, Weitao
N1 - Publisher Copyright:
© 2026 Copyright held by the owner/author(s).
PY - 2026/4/13
Y1 - 2026/4/13
N2 - Facial interaction provides a safe, hands-free input method for cyclists. However, existing wearable facial gesture recognition suffers from severe interference in real-world conditions such as lighting, vibration, sweat, noise, and temperature changes. We present MagFace, an interference-resistant recognition system for cycling glasses using energy-efficient magnetic sensing. MagFace employs four pairs of magnetic silicone and magnetometers on the frame to capture subtle facial skin movements, operating at 30 Hz with a peak power of 150 mW. A tailored deep learning pipeline effectively learns magnetic signals for gesture classification. An evaluation (N=15) shows that MagFace required only one minute of training data to recognize six gestures across different cycling scenarios with high accuracy. A controlled conditions evaluation (N=8) shows MagFace's robustness against strong lighting, wind, bumpy roads, and uphills. Finally, an in-the-wild evaluation (N=14) shows the stable performance of MagFace's real-time system and demonstrates promising usability of MagFace.
AB - Facial interaction provides a safe, hands-free input method for cyclists. However, existing wearable facial gesture recognition suffers from severe interference in real-world conditions such as lighting, vibration, sweat, noise, and temperature changes. We present MagFace, an interference-resistant recognition system for cycling glasses using energy-efficient magnetic sensing. MagFace employs four pairs of magnetic silicone and magnetometers on the frame to capture subtle facial skin movements, operating at 30 Hz with a peak power of 150 mW. A tailored deep learning pipeline effectively learns magnetic signals for gesture classification. An evaluation (N=15) shows that MagFace required only one minute of training data to recognize six gestures across different cycling scenarios with high accuracy. A controlled conditions evaluation (N=8) shows MagFace's robustness against strong lighting, wind, bumpy roads, and uphills. Finally, an in-the-wild evaluation (N=14) shows the stable performance of MagFace's real-time system and demonstrates promising usability of MagFace.
KW - eye-mounted wearable
KW - facial gesture recognition
KW - facial gestures
KW - magnetic sensing
KW - user-defined gestures
UR - https://www.scopus.com/pages/publications/105038690804
U2 - 10.1145/3772318.3790675
DO - 10.1145/3772318.3790675
M3 - Conference contribution
AN - SCOPUS:105038690804
T3 - Conference on Human Factors in Computing Systems - Proceedings
BT - CHI 2026 - Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems
A2 - Oliver, Nuria
A2 - Shamma, David A.
A2 - Candello, Heloisa
A2 - Cesar, Pablo
A2 - Lopes, Pedro
A2 - Bozzon, Alessandro
A2 - Kosch, Thomas
A2 - Liao, Vera
A2 - Ma, Xiaojuan
A2 - Artizzu, Valentino
A2 - Draxler, Fiona
A2 - Lopez, Gustavo
A2 - Reinschluessel, Anke V.
A2 - Tong, Xin
A2 - Toups Dugas, Phoebe O.
PB - Association for Computing Machinery
Y2 - 13 April 2026 through 17 April 2026
ER -