# WatchHAR: Real-time On-device Human Activity Recognition System for Smartwatches

arXiv: https://arxiv.org/abs/2509.04736 · HTML: https://mlsanity.com/paper/2509.04736/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 34 references that are in the map.
- First posted on arXiv 5 Sep 2025.

## Builds on (most cited first, 9 of 9)

- [Xception: Deep Learning with Depthwise Separable Convolutions](https://mlsanity.com/paper/1610.02357/) (2016)
- [Searching for MobileNetV3](https://mlsanity.com/paper/1905.02244/) (2019)
- [PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition](https://mlsanity.com/paper/1912.10211/) (2019)
- [Gated Multimodal Units for Information Fusion](https://mlsanity.com/paper/1702.01992/) (2017)
- [Recognizing Detailed Human Context In-the-Wild from Smartphones and Smartwatches](https://mlsanity.com/paper/1609.06354/) (2016)
- [MobilePoser: Real-Time Full-Body Pose Estimation and 3D Human Translation from IMUs in Mobile Consumer Devices](https://mlsanity.com/paper/2504.12492/) (2025)
- [Attend And Discriminate: Beyond the State-of-the-Art for Human Activity Recognition using Wearable Sensors](https://mlsanity.com/paper/2007.07172/) (2020)
- [CoughTrigger: Earbuds IMU Based Cough Detection Activator Using An Energy-efficient Sensitivity-prioritized Time Series Classifier](https://mlsanity.com/paper/2111.04185/) (2021)
- [ConvBoost: Boosting ConvNets for Sensor-based Activity Recognition](https://mlsanity.com/paper/2305.13541/) (2023)

## Built on by (most cited first, 3 of 3)

- [HARMES: A Multi-Modal Dataset for Wearable Human Activity Recognition with Motion, Environmental Sensing and Sound](https://mlsanity.com/paper/2605.02596/) (2026)
- [TransfHAR: Self-Supervised Wrist Representations for On-Demand Activity Recognition](https://mlsanity.com/paper/2608.15861/) (2026)
- [Smartwatch-Based Sitting Time Estimation in Real-World Office Settings](https://mlsanity.com/paper/2604.08808/) (2026)

## Abstract (arXiv)

Despite advances in practical and multimodal fine-grained Human Activity Recognition (HAR), a system that runs entirely on smartwatches in unconstrained environments remains elusive. We present WatchHAR, an audio and inertial-based HAR system that operates fully on smartwatches, addressing privacy and latency issues associated with external data processing. By optimizing each component of the pipeline, WatchHAR achieves compounding performance gains. We introduce a novel architecture that unifies sensor data preprocessing and inference into an end-to-end trainable module, achieving 5x faster processing while maintaining over 90% accuracy across more than 25 activity classes. WatchHAR outperforms state-of-the-art models for event detection and activity classification while running directly on the smartwatch, achieving 9.3 ms processing time for activity event detection and 11.8 ms for multimodal activity classification. This research advances on-device activity recognition, realizing smartwatches' potential as standalone, privacy-aware, and minimally-invasive continuous activity tracking devices.