Company News

Panasonic R&D Center Singapore Achieves No. 1 Accuracy of Face Recognition on the IJB-C Dataset

June 30, 2019 | Project

Panasonic R&D Center Singapore achieved the Best Accuracy of Face Recognition on the IJB-C dataset under three different protocols, namely 1:1 mixed verification, 1:N mixed identification and 1:1 covariate verification.

On top of that, a paper titled “Deep Face Recognition Model Compression via Knowledge Transfer and Distillation” was published on 3rd Jun 2019 on arXiv, which can be accessed by the following URL: https://arxiv.org/abs/1906.00619.

In the paper, we present a novel model compression approach based on a student-teacher paradigm for face-recognition applications. The proposed approach is simple and hardware-friendly, thus enabling the deployment of high-performing face-recognition deep models into resource-constraint devices.

Share

Related Posts

IP Week @ SG 2016

The Intellectual Property Group (IPG) of Panasonic R&D Center Singapore (PRDCSG) participated, as one of the exhibitors, in IP Ecosystem…

Read more

Paper accepted for publication in the IEEE Transactions on Systems, Man, and Cybernetics: Systems

Collaborating with Nanjing University of Science and Technology (NUST), Panasonic R&D Center Singapore co-authored a paper titled “Modular Lightweight Network…

Read more

Panasonic Tops the 2018 Nvidia AI City Challenge as First-Place Winner for Anomaly Detection

Panasonic R&D Center Singapore reprises the lead in Face Recognition Technology by notching up a stunning win for Track 2 (Anomaly Detection) in the 2018 Nvidia AI City Challenge .

Read more

Deep Learning Summit Asia @SG 2016

Members from PRDCSG and various Panasonic divisions and centres in Japan attended the Deep Learning Summit Asia, which presented a great opportunity for us to gain new perspectives on approaches to technical advances in deep learning and smart AI, as well as new forms of creativity in the new technological era. It also provided us with new opportunities for business networking with other participants.

Read more
Go top