The paper reports a novel cloud based approach for image matching between high-resolution images of faces and low resolution images of ID Cards. We design our application matching the mobile cloud computing design guidelines with the use of CUDA kernel invocation from regular mobile devices (devices that naively donât support CUDA GPGPUs) as a novel contribution. Face matching is performed by the OpenFace deep neural network, which evaluates pre-processed images in cloud, whilst pre-processing is done on mobile device. To test our system, we built an image dataset of 30 subject caputeres in 10 different poses, denoised to reduce any traces of stamps or watermark on the ID cards, mixed to the well known ORL and LFW datasets.
WhoAreYou (WAY): A Mobile CUDA Powered Picture ID Card Recognition System
Montella, Raffaele;Petrosino, Alfredo;Santopietro, Vincenzo
2017-01-01
Abstract
The paper reports a novel cloud based approach for image matching between high-resolution images of faces and low resolution images of ID Cards. We design our application matching the mobile cloud computing design guidelines with the use of CUDA kernel invocation from regular mobile devices (devices that naively donât support CUDA GPGPUs) as a novel contribution. Face matching is performed by the OpenFace deep neural network, which evaluates pre-processed images in cloud, whilst pre-processing is done on mobile device. To test our system, we built an image dataset of 30 subject caputeres in 10 different poses, denoised to reduce any traces of stamps or watermark on the ID cards, mixed to the well known ORL and LFW datasets.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
