WiOpen: A Robust Wi-Fi-based Open-set Gesture Recognition Framework

TL;DR

WiOpen tackles open-set gesture recognition using Wi‑Fi CSI by first reducing noise with CSI ratio preprocessing and then learning an uncertainty-quantification–driven OSGR network that leverages neighborhood relationships to define flexible, sample-based decision boundaries. The two-branch network minimizes an uncertainty-aware loss that promotes compact, semantically separable features and robust domain generalization, enabling effective rejection of unknown gestures while accurately recognizing known ones. Experiments on Widar3.0 and ARIL demonstrate superior AUROC and close-set accuracy, with strong cross-domain and small-sample performance, highlighting practical value for real-world Wi‑Fi sensing applications.

Abstract

Recent years have witnessed a growing interest in Wi-Fi-based gesture recognition. However, existing works have predominantly focused on closed-set paradigms, where all testing gestures are predefined during training. This poses a significant challenge in real-world applications, as unseen gestures might be misclassified as known classes during testing. To address this issue, we propose WiOpen, a robust Wi-Fi-based Open-Set Gesture Recognition (OSGR) framework. Implementing OSGR requires addressing challenges caused by the unique uncertainty in Wi-Fi sensing. This uncertainty, resulting from noise and domains, leads to widely scattered and irregular data distributions in collected Wi-Fi sensing data. Consequently, data ambiguity between classes and challenges in defining appropriate decision boundaries to identify unknowns arise. To tackle these challenges, WiOpen adopts a two-fold approach to eliminate uncertainty and define precise decision boundaries. Initially, it addresses uncertainty induced by noise during data preprocessing by utilizing the CSI ratio. Next, it designs the OSGR network based on an uncertainty quantification method. Throughout the learning process, this network effectively mitigates uncertainty stemming from domains. Ultimately, the network leverages relationships among samples' neighbors to dynamically define open-set decision boundaries, successfully realizing OSGR. Comprehensive experiments on publicly accessible datasets confirm WiOpen's effectiveness. Notably, WiOpen also demonstrates superiority in cross-domain tasks when compared to state-of-the-art approaches.

Figures (13)

• Figure 1: The comparison between different approaches. (a) Close-set methods; (b) Distance-based methods; (c) Prototype-based methods; (d) Wiopen. Colored circles represent known classes samples, gray "x" symbols, colored "x" symbols, and stars represents unknown samples, misclassified samples and prototypes, respectively.
• Figure 2: The multipath effect, variations in the environment, changes in the position and orientation of the performer, and interference all contribute to an increased randomness in the distribution of Wi-Fi CSI data.
• Figure 3: Examples of samples with different noise uncertainty, where $\Psi_{n}$ before filtering is 1.92, and $\Psi_{n}$ after filtering is 0.96.
• Figure 4: Domain uncertainty.(a) 0.0901; (b) 0.2659. Colored points represent samples from known classes, while gray points represent samples from unknown classes.
• Figure 5: WiOpen System.