Performance Comparison of Real Time Object Detection Techniques with YOLOv4

Computer vision is a recent technological advancement to digitally perceive the real world at an advanced level, through digital images and videos. Object detection is a subset of computer vision which is one of the prominent techniques used for object tracking, automatic driving, anomaly detection, etc. Object detection can be based on either machine learning or deep learning algorithms, it can be used for the localization of the image and classification of elements into diverse classes. This work provides a comparison of the object detection approaches such as Region with Convolutional Neural Network (R-CNN), Fast R-CNN, and You Only Look Once(YOLO) and Single Shot multibox Detector (SSD). The implementation of an object detection technique YOLOv4 and a custom model are done, which recognizes the objects from an input image, webcam image and live stream webcam video.