No abstract

Recently, monitoring of physiological signs such as heart rate and respiratory rate is very important, especially when we are talking about pandemics like Covid-19. In this paper we present a state of the art on the different techniques used for heart rate and respiratory rate extraction. These techniques presented will be based on image processing, were traditional sensor-based techniques creating a lot of problem at the contact level between patient and doctor. For this reason, we focus on non-contact techniques to avoid these problems. Generally, the literature review shows that non-contact monitoring techniques are based on RGB, thermal and multispectral cameras, the choice between these different cameras depends on the application that will be used. For example, thermal cameras are dedicated to the prediction of respiratory rate and temperature, while RGB and multispectral cameras are used for heart rate.

Convolutional Neural Networks (CNNs) have been incredibly effective for object detection tasks. YOLOv4 is a state-of-the-art object detection algorithm designed for embedded systems. It is based on YOLOv3 and has improved accuracy, speed, and robustness. However, deploying CNNs on embedded systems such as Field Programmable Gate Arrays (FPGAs) is difficult due to their limited resources. To address this issue, FPGA-based CNN architectures have been developed to improve the resource utilization of CNNs, resulting in improved accuracy and speed. This paper examines the use of General Matrix Multiplication Operations (GEMM) to accelerate the execution of YOLOv4 on embedded systems. It reviews the most recent GEMM implementations and evaluates their accuracy and robustness. It also discusses the challenges of deploying YOLOv4 on autonomous vehicle datasets. Finally, the paper presents a case study demonstrating the successful implementation of YOLOv4 on an Intel Arria 10 embedded system using GEMM.

No abstract