arXiv:2510.22772v1 Announce Type: cross 
Abstract: Radar-based human activity recognition (HAR) is attractive for unobtrusive and privacy-preserving monitoring, yet many CNN/RNN solutions remain too heavy for edge deployment, and even lightweight ViT/SSM variants often exceed practical compute and memory budgets. We introduce Neural-HAR, a dimension-gated CNN accelerator tailored for real-time radar HAR on resource-constrained platforms. At its core is GateCNN, a parameter-efficient Doppler-temporal network that (i) embeds Doppler vectors to emphasize frequency evolution over time and (ii) applies dual-path gated convolutions that modulate Doppler-aware content features with temporal gates, complemented by a residual path for stable training. On the University of Glasgow UoG2020 continuous radar dataset, GateCNN attains 86.4% accuracy with only 2.7k parameters and 0.28M FLOPs per inference, comparable to CNN-BiGRU at a fraction of the complexity. Our FPGA prototype on Xilinx Zynq-7000 Z-7007S reaches 107.5 $\mu$s latency and 15 mW dynamic power using LUT-based ROM and distributed RAM only (zero DSP/BRAM), demonstrating real-time, energy-efficient edge inference. Code and HLS conversion scripts are available at https://github.com/lab-emi/AIRHAR.

Neural-HAR هو مسرع CNN مبتكر مصمم خصيصًا للتعرف على نشاط الإنسان في الوقت الحقيقي باستخدام الرادار. تعالج هذه الابتكار التحديات المتعلقة بنشر نماذج معقدة على الأجهزة ذات الموارد المحدودة، مما يجعلها نقطة تحول لحلول المراقبة غير المتطفلة. من خلال تحسين الأداء مع الحفاظ على الكفاءة، يفتح Neural-HAR آفاقًا جديدة للتطبيقات التي تحافظ على الخصوصية في مجالات متعددة، من الأمن إلى الرعاية الصحية.

Neural-HAR es un acelerador CNN innovador diseñado específicamente para el reconocimiento de actividades humanas en tiempo real basado en radar. Esta innovación aborda los desafíos de implementar modelos complejos en dispositivos con recursos limitados, convirtiéndose en un cambio de juego para soluciones de monitoreo discretas. Al optimizar el rendimiento y mantener la eficiencia, Neural-HAR abre nuevas posibilidades para aplicaciones que preservan la privacidad en diversos campos, desde la seguridad hasta la atención médica.

Neural-HAR est un accélérateur CNN révolutionnaire conçu spécifiquement pour la reconnaissance d'activité humaine en temps réel basée sur le radar. Cette innovation répond aux défis de déploiement de modèles complexes sur des appareils à ressources limitées, ce qui en fait un changement de donne pour les solutions de surveillance discrètes. En optimisant les performances tout en maintenant l'efficacité, Neural-HAR ouvre de nouvelles possibilités pour des applications préservant la vie privée dans divers domaines, de la sécurité à la santé.

Neural-HAR is a groundbreaking CNN accelerator designed specifically for real-time radar-based human activity recognition. This innovation addresses the challenges of deploying complex models on resource-constrained devices, making it a game-changer for unobtrusive monitoring solutions. By optimizing performance while maintaining efficiency, Neural-HAR opens up new possibilities for privacy-preserving applications in various fields, from security to healthcare.

Neural-HAR: A Dimension-Gated CNN Accelerator for Real-Time Radar Human Activity Recognition

One More Thing in AI – Your Shortcut to AI Mastery

Neural-HAR: A Dimension-Gated CNN Accelerator for Real-Time Radar Human Activity Recognition

Was this article worth reading? Share it

One More Thing in AI

LucidQuery AI

Airparser

Humanize AI

GPTHumanizer

FETCH HIVE

Ready to build your own newsroom?