# AMAR: Lightweight Attention-Based Multi-User Activity Recognition from Wi-Fi CSI

arXiv: https://arxiv.org/abs/2605.20649 · HTML: https://mlsanity.com/paper/2605.20649/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 34 references that are in the map.
- First posted on arXiv 20 May 2026.

## Builds on (most cited first, 13 of 13)

- [End-to-End Object Detection with Transformers](https://mlsanity.com/paper/2005.12872/) (2020)
- [Fast R-CNN](https://mlsanity.com/paper/1504.08083/) (2015)
- [Xception: Deep Learning with Depthwise Separable Convolutions](https://mlsanity.com/paper/1610.02357/) (2016)
- [Multi-Scale Context Aggregation by Dilated Convolutions](https://mlsanity.com/paper/1511.07122/) (2015)
- [SoundStream: An End-to-End Neural Audio Codec](https://mlsanity.com/paper/2107.03312/) (2021)
- [Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation](https://mlsanity.com/paper/2010.16061/) (2020)
- [MoMask: Generative Masked Modeling of 3D Human Motions](https://mlsanity.com/paper/2312.00063/) (2023)
- [EfficientFi: Towards Large-Scale Lightweight WiFi Sensing via CSI Compression](https://mlsanity.com/paper/2204.04138/) (2022)
- [WiMANS: A Benchmark Dataset for WiFi-based Multi-user Activity Sensing](https://mlsanity.com/paper/2402.09430/) (2024)
- [MUSE-Fi: Contactless MUti-person SEnsing Exploiting Near-field Wi-Fi Channel Variation](https://mlsanity.com/paper/2308.10234/) (2023)
- [Harvesting Ambient RF for Presence Detection Through Deep Learning](https://mlsanity.com/paper/2002.05770/) (2020)
- [RSCNet: Dynamic CSI Compression for Cloud-based WiFi Sensing](https://mlsanity.com/paper/2402.04888/) (2024)
- [Context-Aware Predictive Coding: A Representation Learning Framework for WiFi Sensing](https://mlsanity.com/paper/2410.01825/) (2024)

## Abstract (arXiv)

Wi-Fi-based human activity recognition (HAR) has emerged as a promising approach for contactless sensing, leveraging channel state information (CSI) collected from wireless transceivers. While existing studies have primarily concentrated on single-user scenarios, real-world deployments often involve multi-user settings where concurrent users' movements induce overlapping CSI patterns that challenge conventional classification methods. To address this limitation, this paper introduces an attention-based multi-user activity recognition (AMAR) framework that formulates HAR as a set prediction problem. The transformer-based architecture in AMAR leverages learnable query embeddings acting as specialized activity detectors, enabling the simultaneous identification of multiple activities from composite CSI representations. Moreover, to address deployment constraints, AMAR is designed in an edge-cloud split architecture form where lightweight convolutional networks on edge devices perform initial feature extraction, followed by residual vector quantization that achieves substantial bandwidth reduction while preserving activity-discriminative information. The cloud component performs final activity prediction through attention-based set matching, enabling the system to handle varying occupancy levels. Across classroom, meeting-room, and empty-room environments, on average AMAR nearly doubles the rate of perfectly predicting all concurrent activities compared to the best baseline. Moreover, it achieves an F1-score of 53.4% compared to 45.6% for the best benchmark, and reduces occupancy estimation error by 74%, while minimizing bandwidth substantially.