TY - GEN
T1 - RFOOD
T2 - 23rd IEEE International Conference on Machine Learning and Applications, ICMLA 2024
AU - Kahya, Sabri Mustafa
AU - Yavuz, Muhammet Sami
AU - Sivrikaya, Boran Hamdi
AU - Steinbach, Eckehard
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - Out-of-distribution (OOD) detection is critical for the safe deployment of modern neural network architectures, as it aims to identify samples outside the training domain. In this paper, we introduce RFOOD, a novel OOD detection framework designed for real-time, privacy-preserving facial authentication using low-cost frequency-modulated continuous-wave (FMCW) radar. RFOOD employs both range-Doppler and micro range- Doppler images to enhance the detection accuracy. The architecture consists of a multi-encoder multi-decoder Body Part (BP) and Intermediate Linear Encoder-Decoder (ILED) components. This design allows the system to accurately classify a single individual's face as in-distribution (ID) while identifying all other faces as OOD. On our dataset collected with 60 GHz short-range FMCW radar, RFOOD achieves an Area Under the Receiver Operating Characteristic (AUROC) curve of 94.13 % and a False Positive Rate of 18.12% at a True Positive Rate of 95 % (FPR95). Additionally, RFOOD outperforms state-of-the-art OOD detection methods in common OOD detection metrics and operates in real-time.
AB - Out-of-distribution (OOD) detection is critical for the safe deployment of modern neural network architectures, as it aims to identify samples outside the training domain. In this paper, we introduce RFOOD, a novel OOD detection framework designed for real-time, privacy-preserving facial authentication using low-cost frequency-modulated continuous-wave (FMCW) radar. RFOOD employs both range-Doppler and micro range- Doppler images to enhance the detection accuracy. The architecture consists of a multi-encoder multi-decoder Body Part (BP) and Intermediate Linear Encoder-Decoder (ILED) components. This design allows the system to accurately classify a single individual's face as in-distribution (ID) while identifying all other faces as OOD. On our dataset collected with 60 GHz short-range FMCW radar, RFOOD achieves an Area Under the Receiver Operating Characteristic (AUROC) curve of 94.13 % and a False Positive Rate of 18.12% at a True Positive Rate of 95 % (FPR95). Additionally, RFOOD outperforms state-of-the-art OOD detection methods in common OOD detection metrics and operates in real-time.
KW - 60 GHz FMCW radar
KW - deep neural networks
KW - Facial authentication
KW - out-of-distribution detection
UR - http://www.scopus.com/inward/record.url?scp=105000864852&partnerID=8YFLogxK
U2 - 10.1109/ICMLA61862.2024.00077
DO - 10.1109/ICMLA61862.2024.00077
M3 - Conference contribution
AN - SCOPUS:105000864852
T3 - Proceedings - 2024 International Conference on Machine Learning and Applications, ICMLA 2024
SP - 528
EP - 533
BT - Proceedings - 2024 International Conference on Machine Learning and Applications, ICMLA 2024
A2 - Wani, M. Arif
A2 - Angelov, Plamen
A2 - Luo, Feng
A2 - Ogihara, Mitsunori
A2 - Wu, Xintao
A2 - Precup, Radu-Emil
A2 - Ramezani, Ramin
A2 - Gu, Xiaowei
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 18 December 2024 through 20 December 2024
ER -