Channel–Spatial Attention Enhanced Few-Shot Radar-Based Hand Gesture Recognition
IEEE Sensors Letters · 2026-04 · 10 (4) · 1-4
DOI: 10.1109/lsens.2026.3669424 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Sensors Letters, 2026-04. Verfasst von Krishnapriya S. und 2 weiteren Autor:innen. Die Arbeit nennt 23 Literaturangaben.
Abstract (Originalsprache)
Radar-based hand gesture recognition is a promising modality for human computer interaction due to its advantages over camera-based approaches. We propose a novel few-shot hand gesture recognition model, using range-Doppler images (RDI), that can generalize with a limited number of samples. The proposed model uses a prototypical network with a customized feature embedding network and channel–spatial attention to effectively capture the spatial and temporal features from the RDI frames. We carry out extensive experiments of the proposed prototypical network-based few-shot framework on four public radar-based hand gesture datasets. The results show promising gesture recognition accuracy and good generalizability, validating the proposed framework and its applicability in real-world scenarios.
Zitieren
APA
S., K., Srirangarajan, S., & Roy, S. D. (2026). Channel–Spatial Attention Enhanced Few-Shot Radar-Based Hand Gesture Recognition. IEEE Sensors Letters, 10(4), 1-4. https://doi.org/10.1109/lsens.2026.3669424
Vancouver
S. K, Srirangarajan S, Roy SD. Channel–Spatial Attention Enhanced Few-Shot Radar-Based Hand Gesture Recognition. IEEE Sensors Letters. 2026;10(4):1-4. doi:10.1109/lsens.2026.3669424
BibTeX
@article{S2026,
author = {S., Krishnapriya and Srirangarajan, Seshan and Roy, Sumantra Dutta},
title = {Channel–Spatial Attention Enhanced Few-Shot Radar-Based Hand Gesture Recognition},
journal = {IEEE Sensors Letters},
year = {2026},
volume = {10},
number = {4},
pages = {1--4},
doi = {10.1109/lsens.2026.3669424}
}
EUVIMED