Person identification improved using daily activities from millimeter-wave data
Beyond Gait: Person Identification from Millimeter-Wave Point Clouds Across Activities of Daily Living
Computer Vision and Pattern Recognition
Summary
Most systems that identify people using millimeter-wave sensors do so by analyzing the way they walk. But walking indoors is often short and stops frequently. The authors show that other common activities also contain useful information to recognize people. They collected a new dataset showing 11 people performing seven daily activities and developed a method that uses knowledge of the activity to help identify each person better. Their method significantly improves accuracy compared to just using walking data.
What this means in practice
- •For indoor security teams: Implement more accurate person identification systems using millimeter-wave sensors by employing multiple daily activities rather than just gait patterns.$Commercial implications: Enables development of enhanced indoor security systems that identify individuals across varied daily activities, improving tracking in controlled environments.
- •For smart building operators: Use activity-conditioned frameworks to improve occupant identification for personalized services and access control in indoor environments.
Authors
Xilai Wang, Zixiong Han, Saad Rhanmouni, Chenzhe Zhao, Yunze Lu, Miodrag Bolic
Abstract
Person identification from millimeter-wave (mmWave) point clouds has mainly relied on gait. Indoor walking, however, is often brief and interrupted, while other activities of daily living (ADLs) may provide complementary identity information. We investigate identification across seven ADLs using mm-ADL, a new point-cloud dataset collected from 11 subjects under a controlled protocol. This extension introduces heterogeneous states and transitions whose spatial and temporal characteristics vary with activity. We therefore study whether activity can provide useful context for learning identity representations. We propose an activity-conditioned framework in which a human activity recognition router dispatches each clip to an activity-specific identity expert. The framework is implemented as a supervised mixture of experts, using a dual-stream static-dynamic PointNet (DS-SDPNet) to combine time-aggregated spatial structure with frame-to-frame information. We evaluate closed-set identification (ID) and subject-disjoint re-identification (ReID). With learned hard routing, ID accuracy increases from 62.1% to 68.0%. In a two-occupant ReID setting, hard routing increases mAP from 57.2% to 75.4% and Rank-1 accuracy from 59.1% to 82.1%. Under a matched gallery partition, activity-specific experts also outperform a shared embedding, showing that the gain extends beyond restricting the gallery. These results support the feasibility of using ADLs beyond gait for identification and the value of activity conditioning under controlled indoor conditions.