An Upgraded Object Detection Model for Enhanced Perception and Decision Making in Autonomous Vehicles

In Internet of Vehicles (IoV), accurate object detection is one of the basic requirements for Autonomous Vehicles (AV s) and Vision-based Self Driver Assistance System (VSDAS). With restricted processing power of sensor nodes and network bandwidth, enhanced object detection with low energy consumption and acceptable rate of accuracy is still a major issue that makes VSDAS untrust-worthy and unsustainable. Considering this situation, in this paper, the authors propose an upgraded Object Detection model for simulating the enhanced perception and decision making in autonomous vehicles. In this paper, we evaluate the performance of three methods namely Histogram of Ori-ented Gradients (HOG), Local Binary Pattern (LBP), HAAR utilizing KITTI dataset. The results reveal that Haar exhibits higher detection rate than the other two methods. For the enhanced object detection, we utilize a frame similarity difference technique for filtering out the duplicate frames and generating key frames. Finally, an upgraded Haar-cascade classification algorithm is proposed for accurate and fast object detection. Our comprehensive experimental outcomes on the eminent publicly available dataset (KITTI) showed that our model not only significantly improves the performance of object detection however, also saves the energy consumption of edee devices.