Embedded Night-Vision System for Pedestrian Detection using Adaboosta Machine Learning Meta-Algorithm

Abstract

Operational vision-based strategies for the driver tomaximize human vision's capacity and ensure safe driving.Regrettably, their extensive service is practically limited tohigh-priced automobiles. Rather than the value of hardwareparts,expensiveismostlikelyaderivativeofthepriceacquired during the testing. This project aims to demonstratehow state-of-the-art algorithms can be used to build a mobilesystemforpedestriandetectioninpoor lighting conditions.Weusedacascade object detector to detect human detectioninthermalimageryandfoundthatthefindingswereinconsistentwiththecurrentstateofthedeeplearningapproach.