Improved Pedestrian Detection using Region based Convolutional Neural Networks for Advanced Driver-Assistance Systems

  • M. Angel Shalini, Dr.S. Vijayalakshmi
Keywords: Pedestrian Detection, CNN, Caltech, cluster.

Abstract

Pedestrian detection dependent on computer vision is a significant part of object detection, which is applied to insightful monitoring, smart driving, robot, etc. As of now, numerous pedestrian detection strategies are proposed. Be that as it may, in light of the intricacy of the foundation, pedestrian stance variety and pedestrian impediments, pedestrian detection is as yet a test which calls for exact implementations. In this paper, the Region-based Convolutional Neural Network is carried out. Right off the bat, image features were extracted by CNN. From that point onward, we developed a Region Proposal Network to extract regions that may contain pedestrians joined with group examination. Also, the region is recognized and grouped by detection network. At last, the technique was tried in the Caltech pedestrian detection dataset. The outcomes show that the strategy for pedestrian detection dependent on region-based CNN accomplishes the precision of 96.7%, performs better, contrasted and different algorithms.

Published
2021-11-23
How to Cite
Dr.S. Vijayalakshmi, M. A. S. (2021). Improved Pedestrian Detection using Region based Convolutional Neural Networks for Advanced Driver-Assistance Systems. Design Engineering, 15252-15259. Retrieved from http://thedesignengineering.com/index.php/DE/article/view/6660
Section
Articles