Deep Learning algorithms have been widely used for different surveillance tasks in recent years, including people monitoring and counting, abnormal behavior identification, and video segmentation. In most situations, it is assumed that the input images are of high visual quality to provide good performance. When the input data is degraded by variables such as high noise or poor lighting conditions accuracy may degrade. We address the illumination issue in this paper by adapting a face recognition algorithm to near-infrared and thermal images. In this study, we propose a fine-tuning approach to allow deep CNN models to be applied to infrared face recognition (NIR and thermal spectrum). The obtained results with the proposed architecture and infrared images show promising results in deep face recognition with a VAR of 96.68% for the NIR dataset and a VAR of 94.57% for the thermal dataset.
An individual's face is a biometric feature that can be used in a computerized security system to identify or authenticate that particular person. The main challenge, while identifying a face through the use of a machine, is to match precisely the captured person's face with the image of the same individual's face already existing in the system's face database. Visual spectrum face images are affected by variations in lighting, head orientation, aging, and disguise resulting in poor visual face detection performance. Infrared imaging is used to help overcome some of these limitations. In this work, we propose a deep Deep Convolutional Neural Network architecture based on the FaceNet architecture and the MTCNN model to perform face recognition on a set of thermal data. Tests conducted on the USTC-NVIE dataset show promising results and the possibility of using deep learning in thermal face recognition.
These last years, we have witnessed considerable improvements in machine learning and deep learning. Many advanced techniques are now based on deep neural networks. Although many software libraries are available, the development of deep neural networks requires a good level of mathematical knowledge and high programming skills. In this work, we present a visual tool to help simplify the programming of deep learning networks. The developed framework DeepViP is comprised of a node editor that provides users with a toolbox representing different types of neural layers. It allows the connection between the different blocks and the configuration of important hyper parameters of each layer. Thus, speeding-up experimentation with different architectures. Additionally, the developed solution offers users the possibility to generate a python script of the designed network that can be run using specific libraries such as keras or tensorflow.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.