An Embedded Multicore Platform Exploration in Video Application Utilizing FPGA

Abstract

Multi-processor is not a new technology, but with the development of modern silicon technology, it is possible to integrate multiple cores in a single chip package, which is called multicore processor. Whether in the desktop personal machine, or embedded applications, multicore processor has been a general trend, due to the requirement of high performance and design problems in single-core processor. Surrounded multi-screen provides a better sense of reality, which is widely used in the surveillance, military, exhibitions, and so on. With the advantages in parallel processing, multicore technology has an important practical significance and a broad prospect in these applications. In this paper, an exploration on multicore architecture is mainly focused on, from the perspectives of processing elements, memory hierarchy, and on-chip interconnection. A basic platform for multi-screen display is implemented on the Xilinx field programmable gate array (FPGA), and it illustrates that there is a 3.6 times higher performance than the corresponding single-core design, which provides a helpful guidance and revelation to further researches.

Helmet wearing detection based on YOLOv5

“…Where, ( h w, ) and ( gt gt h w , ) are the width and height of the prediction box and the real box respectively, as shown in Equation (10), θ controls the attention to shape loss, avoids excessive attention to shape loss and reduces the movement of the prediction box, θ is set between [2,6].…”

“…Traditional machine learning object detection algorithms, such as CAI Limei [1] , realize the detection of hard hats by using the circular shape characteristics of hard hats and combining the histogram of the edge direction and orientation information as the feature map. Li Xiao [2] used Adaboost classifier and Har-like feature to detect whether to wear a helmet. Pathasu Doungmala [3] et al used haar features [4] for face detection and circular Hough transform for helmet detection.…”