Research and implementation of multi-object tracking based on vision DSP
- PDF / 1,339,077 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 93 Downloads / 152 Views
SPECIAL ISSUE PAPER
Research and implementation of multi‑object tracking based on vision DSP Xuan Gong1 · Zichun Le1 Received: 3 May 2019 / Accepted: 7 March 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract This paper applies a pragmatic approach to study the real-time performance effect of software design methods for multiple object tracking (MOT) based on vision digital signal processing (vision DSP). The MOT system in the paper combines target detection, the Hungarian algorithm and the Kernel correlation filter (KCF) tracker. In addition, the MOT system needs to support multiway video streams, so higher speed and storage requirements are necessary for target tracking. Therefore, we carried out some studies on how to improve tracker speed performance and reduce system resource consumption under limited system resources. In the paper, we achieved the goal in two respects. Regarding the data processing, we studied how to efficiently process tracking data by utilizing the parallel characteristics of iDMA (integrated direct memory access) and a DSP core; and regarding the data storage, we proposed a time-sharing strategy to solve the DSP local memory (data RAM) usage issue for multiple tracking objects. In addition, regarding the software design, we propose a new strategy, which includes two levels of parallel computations: the frame-level parallel computations and the tracking object-level parallel computations. The experimental results show that the KCF tracking algorithm based on vision DSP achieves not only the desired real-time tracking speed but also the expected goal of system resource utilization. Our research methods also provide a reference for algorithm embedded applications in the field of computer vision. Keywords KCF · Detection · Tracking · MOT · Hungary algorithm · Vision DSP
1 Introduction Target tracking is an important technology in computer vision. In recent years, tracking technology has continued to develop and evolve from the traditional tracking technology (e.g., using the optical flow, Kalman filter, and Mean-shift in target tracking, which are also called generative models) to the correlation filtering tracking technology developed in recent years [1–7] (e.g., MOSSE [1], KCF [2, 3], and DSST [4, 5]). Even deep learning has been applied to target tracking (the application of correlation filtering and deep learning to target tracking is commonly called the discriminant pattern method). As a result, tracking algorithms have achieved great improvement with respect to their complexity and accuracy. For example, the KCF algorithm with the HOG feature [8] achieved an accuracy of 73.2% and an average * Xuan Gong [email protected] 1
College of Science, Zhejiang University of Technology, Hangzhou 310023, China
frame rate of 172 fps on the OTB50 dataset. Although the frame rate of the MOSSE algorithm is 615 fps, the average accuracy is only 43.1%. The application of deep learning technology to target tracking has also become a major research hot spot in the field of targ
Data Loading...
