Gesture recognition is a novel technology that aims to change the way people interact with machines. Existing solutions typically recognize gestures using camera vision, wearable sensors, or specialized signals (e.g., WiFi, sound, and visible light), but they have limitations such as high-power consumption or low signal-to-noise ratio (SNR) in comparison to their surroundings, making it difficult to accurately detect finger movements. In this research, we propose a device-free gesture identification system that recognizes different hand movements by processing through Edge Machine Learning (EML) algorithms the received signal strength indication (RSSI) and phase values from backscattered signals of a collection of Radio Frequency IDentification (RFID) tags mounted on a plastic plate. The performances of three algorithms, the Random Forest Classifier, the Support Vector Machine, and the Decision Tree Classifier, were compared giving very encouraging results with accuracy up to 99.4%.

Edge Machine Learning techniques applied to RFID for device-free hand gesture recognition / Merenda, M.; Cimino, G.; Carotenuto, R.; Corte, F. G. D.; Iero, D.. - In: IEEE JOURNAL OF RADIO FREQUENCY IDENTIFICATION. - ISSN 2469-7281. - 6:(2022), pp. 564-572. [10.1109/JRFID.2022.3185804]

Edge Machine Learning techniques applied to RFID for device-free hand gesture recognition

Merenda M.;Carotenuto R.;Iero D.
2022-01-01

Abstract

Gesture recognition is a novel technology that aims to change the way people interact with machines. Existing solutions typically recognize gestures using camera vision, wearable sensors, or specialized signals (e.g., WiFi, sound, and visible light), but they have limitations such as high-power consumption or low signal-to-noise ratio (SNR) in comparison to their surroundings, making it difficult to accurately detect finger movements. In this research, we propose a device-free gesture identification system that recognizes different hand movements by processing through Edge Machine Learning (EML) algorithms the received signal strength indication (RSSI) and phase values from backscattered signals of a collection of Radio Frequency IDentification (RFID) tags mounted on a plastic plate. The performances of three algorithms, the Random Forest Classifier, the Support Vector Machine, and the Decision Tree Classifier, were compared giving very encouraging results with accuracy up to 99.4%.
2022
Antenna arrays
Antenna measurements
Antennas
edge machine learning
Gesture recognition
hand gesture
Location awareness
Phase measurement
Plastics
RFID
RSSI
sensorless
File in questo prodotto:
File Dimensione Formato  
Merenda_2022_JRFID_edge_Post.pdf

accesso aperto

Descrizione: Articolo principale
Tipologia: Documento in Post-print
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 943.61 kB
Formato Adobe PDF
943.61 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12318/128709
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 11
  • ???jsp.display-item.citation.isi??? 6
social impact