A Practice for Object Detection Using YOLO Algorithm

Abstract

When we look at images or videos, we can easily locate and identify the objects of our interest within moments. Passing on this intelligence to computers is nothing but object detection - locating the object and identifying it. Object Detection has found its application in a wide variety of domains such as video surveillance, image retrieval systems, autonomous driving vehicles and many more. Various algorithms can be used for object detection but we will be focusing on the YoloV3 algorithm. YOLO stands for You Only Look Once . The YOLO model is very accurate and allows us to detect the objects present in the frame. YOLO follows a completely different approach. Instead of selecting some regions, it applies a neural network to the entire image to predict bounding boxes and their probabilities. YOLO is a single deep convolutional neural network that splits the input image into a set of grid cells, so unlike image classification or face detection, each grid cell in YOLO algorithm will have an associated vector in the output that tells us if an object exists in that grid cell, the class of that object, the predicted bounding box for that object. The model here is progressive so it learns more over time, increasing its prediction accuracy over time. The way the model works is that it makes many predictions in one frame and decides to use the most accurate prediction, thus discarding the other. The predictions are made randomly, so if the model feels like there is an object in the frame which is of a very small pixel it will take that also into consideration. To make it more precise and clearer, the model simply creates bounding boxes around everything in the frame, it would make predictions for each box and pick the one with the most confidence score. All this is done in a small-time frame, thus showing why this specific model is the best to use in a real time situation.