Comparative Study On Various Architectures Of Yolo Models Used In Object Recognition

Abstract

In the last few decades, the deep learning paradigm has been widely used in the Machine Learning Community, thereby accounting for some of the most outstanding results on several complex cognitive results, performing on par or even better than human-level performance. One of these many complex tasks is Object Detection. This paper aims to do a comparative study on the YOLO model used in Object Detection, which would help the visually impaired understand their surroundings. With a wide use case in multiple industries and sectors, it has been a hot topic amongst the community for the past decade. Object detection is a method of finding instances of objects from an image of a certain class. Object Detection has been witnessing a brisk revolutionary change in recent times, resulting in many advanced and complex algorithms like YOLO, SSD, Fast R-CNN, Faster R-CNN, HOG, and many more. This research paper explains the architecture of the YOLO algorithm which is widely used in object detection and classifying objects. We have used the COCO dataset to train our model. Our aim of this research study is to trying to identify the best implementation of the YOLO Model.