Advanced Driver Assistant Systems

EasyChair Preprint 5910

Abstract

In recent years, traffic is increasing exponentially. The queueing of the vehicles on the road or intersection has risen sharply and the this increase in traffic flow has led to the traditional traffic lights systems to failure resulting in widespread delays and congestion, resulting in need for vehicle and traffic sign detection. In this paper, a real – time vehicle detection and traffic signs detection is proposed to increase the safety on the roads and to reduce the delays. A object detection model that detects in real time and based on the deep convolutional Neural Networks referred as you only look once (YOLO). This paper is focused on analyzing the costs and benefits of using YOLO as an alternative model in detecting vehicles and traffic signs. Detecting data in a video stream is an object detection problem. An object detection problem can be approached as either a classification problem or a regression problem. In the classification approach, the image is divided into small patches, each of which will be run through a classifier to determine whether there are objects in the patch. The bounding boxes will be assigned to patches with positive classification results. This is all is done using the YOLOv4 approach which is combined with HOG and SVM approach. Linear SVM classifier algorithm is used to classify the car and traffic signs, Sliding Window and Heatmap algorithm are used to make it more robust.

Keyphrases: Advanced Driver Assistant System, Advanced Driver Assistant Systems, Deep Convolutional Neural Network, Grid cell, Traffic sign, Traffic sign detection, Traffic sign recognition, YOLO, anchor box, bounding box, image recognition, non max suppression, object detection, real-time, real-time object detection, vehicle recognition

@booklet{EasyChair:5910,
  author    = {Varun Goel and Harmandeep Singh Paul},
  title     = {Advanced Driver Assistant Systems},
  howpublished = {EasyChair Preprint 5910},
  year      = {EasyChair, 2021}}