Optimization in Object Detection Model using YOLO.v3

Abstract

Object Detection is one of the important entities in the field of Computer Vision with a large number of applications. This project demonstrates Object detection using You Only Look Once (YOLO) Algorithm, version 3. YOLOv3 method is prominently used in object detection methods which are based on Deep Learning. It uses k-means cluster method for creating bounding boxes of specific height and width, which are used for predicting output. The model training is based on the Common Object in Context (COCO) Dataset. The dataset has around 164K images based on 80 categories, also called as classes. Thus, this object detection model takes an image from the user and then with the help of YOLO algorithm, predicts the types of objects present in that image and marks them accurately., the lower complex CNN model achieves an accuracy of 0.93.