In recent yearsobject detection grabbed much research attention in the computer vision systems. The precisionlevels and efficiency of object detection has been increased drastically by introducing deep learning algorithms. A major challenge of many object detection algorithms is their dependence on other computer vision algorithms, which leads to deliberate and degraded performance. This paper aims to incorporate an innovative object detection technique with the goal of achieving high accuracy in real-time performance. This paper utilizes a deep learning approach which mainly based on Convolutional Neural Networks (CNN). CNNs are usually utilized, where speed & accuracy are most required parameters. In order to effectively manage these parameters, You Only Look Once (YOLO) algorithm which is based on CNN has been adopted. In this YOLO algorithm, the image is partitioned into grids, labeled and applied to a deep convolutional neural network. The experimental results prove that YOLO algorithm performs well in image level object detection and orientation estimation. YOLO provides novel memory mapping improvement which makes it much suitable in driving situation and makes it an optimal choice for the autonomous driving system.
Volume 11 | 02-Special Issue
Pages: 1232-1240