Object Detection Using Deep Learning Algorithm

Kilari Veera Swamy; Jayanth Sai Marisetty; Srujan Vare

Abstract

Object detection is a computer technology related to computer vision and image processing that deals with detecting images of semantic objects of a certain class (for example cars, buildings, etc.) in digital images and videos. To detect objects, there are many algorithms available, in this paper YOLO algorithm is explored for detection of objects on real-time data. You Only Look Once (YOLO) is a state-of-the-art real time object detection system whose architecture is based on Convolutional Neural Networks (CNN) and anchor boxes. In this work, YOLO V3 is considered. When it comes to the implementation, the input image is being divided into grids in which for each cell classification and localization is applied. After the localization, the algorithm draws bounding boxes for the detected objects. As in real time images there might be a probability of having objects present in more than one grid cell hence, we also use Non max suppression to detect the object more accurately. We also import OpenCV, NumPy libraries in PyCharm Environment for the better optimization of the output. In this paper, to detect the objects in an image using this YOLO algorithm is also tested apart from videos.