View publication

The abundance of wrist-worn heart rate measuring devices enables long term cardiovascular monitoring through photoplethysmography (PPG). Such signals contain unique identifiable information that can help in biometric authentication. In this work, we propose Fusion-ID, which use wrist-worn PPG sensors fused with motion sensor data as a way to do bio authentication on wrist worn devices. We conducted a user study using a PPG and motion sensor enabled wrist-worn device and collected data from 247 users. We then propose a novel sensor fusion deep Siamese network architecture for feature embedding. Specifically, Fusion-ID fuses information from multiple channels of PPG readings with information from motion sensors to authenticate a user. Our architecture only needs a few seconds of sample data (shots) from new users, and it is the first PPG-based bio-authentication method that is capable of adapting to new users without requiring on-device training or fine-tuning of the model. Our evaluations confirm the effectiveness of proposed siamese network with sensor fusion with an average accuracy of 95% and 12% false rejection rate at the 1% false acceptance rate operating point.

Related readings and updates.


Apple sponsored Acoustics, Speech, and Signal Processing (ICASSP), which was held in a hybrid format. The virtual event took place on May 7 to 13, and the hybrid main conference on May 22 to 27. ICASSP is the IEEE Signal Processing Society’s flagship conference on signal processing and its applications.

See event details

Video Frame Interpolation via Structure-Motion based Iterative Feature Fusion

Video Frame Interpolation synthesizes non-existent images between adjacent frames, with the aim of providing a smooth and consistent visual experience. Two approaches for solving this challenging task are optical flow based and kernel-based methods. In existing works, optical flow based methods can provide accurate point-to-point motion description, however, they lack constraints on object structure. On the contrary, kernel-based methods focus on…
See paper details