WiMANS: A Benchmark Dataset for WiFi-based Multi-user Activity Sensing

TL;DR

WiMANS addresses the gap in WiFi-based sensing by introducing the first benchmark dataset for multi-user activity sensing. It combines 11286 dual-band CSI samples with synchronized video across 3 environments, 5 locations, and 9 daily activities, with 0–5 users per sample and fine-grained annotations for identities, locations, and activities. The paper benchmarks a wide range of WiFi-based and video-based models on identification, localization, and HAR, revealing strong performance for identification/localization but room for improvement in HAR, and showing that video models, while accurate, are less efficient than WiFi-based models. By providing standardized data, baselines, and evaluation protocols, WiMANS enables reproducible research and paves the way for future work in areas such as multi-user pose estimation, dual-band augmentation, and cross-domain sensing.

Abstract

WiFi-based human sensing has exhibited remarkable potential to analyze user behaviors in a non-intrusive and device-free manner, benefiting applications as diverse as smart homes and healthcare. However, most previous works focus on single-user sensing, which has limited practicability in scenarios involving multiple users. Although recent studies have begun to investigate WiFi-based multi-user sensing, there remains a lack of benchmark datasets to facilitate reproducible and comparable research. To bridge this gap, we present WiMANS, to our knowledge, the first dataset for multi-user sensing based on WiFi. WiMANS contains over 9.4 hours of dual-band WiFi Channel State Information (CSI), as well as synchronized videos, monitoring simultaneous activities of multiple users. We exploit WiMANS to benchmark the performance of state-of-the-art WiFi-based human sensing models and video-based models, posing new challenges and opportunities for future work. We believe WiMANS can push the boundaries of current studies and catalyze the research on WiFi-based multi-user sensing.

Figures (6)

• Figure 1: Examples of WiFi CSI and videos in WiMANS, monitoring simultaneous activities performed by multiple users in various environments. (a) consists of 3 users in a classroom. (b) contains 4 users in a meeting room. (c) has 5 users in an empty room.
• Figure 1: Descriptions and examples of 9 daily activities in WiMANS.
• Figure 2: Layouts of environments in WiMANS, where a transmitter and a receiver collect WiFi CSI, and a monitor camera captures synchronized videos for reference.
• Figure 3: Scripts that instruct users to perform identical/different activities at varying locations independently yet simultaneously.
• Figure 4: Statistics of WiMANS regarding the distributions of user identities, locations, activities, and numbers of users per sample. In each distribution, all categories have almost equivalent proportions to construct WiMANS as a relatively balanced dataset.