Object detection is a pivot and prime process in various applications and procedures such as surveillance, classification, recognition and prediction including image retrieval, computer visioning, video streams and many more. Real-time object detection requires identification of different kinds of objects specified in images, videos or live feed streams. It is basic and important to maintain the level of accuracy along with quick inference. This paper proposes an algorithm to perform the real-time object detection typically leverage machine learning, deep learning to produce effective results. The goal is to power machines to identify defined objects in a live feed, videos or images and achieve desired outcomes. The algorithm used is YOLO version3 (YOLOv3). It presents a fast and accurate object detection method with higher performance. To create reliable applications for resolving practical problems, computer vision techniques like tracking and counting are combined. It is the improved proposal to many machine learning algorithms like CNN, RNN, YOLO v1 and v2. Some of the applications are traffic surveillance, vehicle detection, face detection and recognition, number-plate detection, overspeed tracking and more. The factors included are segmentation, accuracy, precision, fastness, performance and efficiency. It is useful to meet the needs of growing technology.