Real-Time Object Detection with Pre-eminent Speed and Precision using YOLOv4

Authors

  • Shraddha Sanjeev Pattanshetti Student, Department of Information Technology, D.K.T.E’s Society Textile and Engineering Institute, Ichalkaranji, India
  • Shabana Imam Nivade Student, Department of Information Technology, D.K.T.E’s Society Textile and Engineering Institute, Ichalkaranji, India

Keywords:

artificial intelligence, computer vision, deep learning, image processing, machine learning, neural network, object detection, pattern recognition, video processing

Abstract

A computer vision technique that has fascinated the world with its outstanding ability to localize and identify objects is Object recognition. It draws bounding boxes around the recognized objects and accurately labels them. Object detection is more intricate than classification as it not only recognizes the object but also the location of the object in the image. A widely known algorithm for precise and quick detection is YOLO. YOLO (You only look once) is an open-source and reliable real-time object recognition algorithm that can identify multiple objects in a single frame. Furthermore, it recognizes objects more rapidly and precisely than other recognition systems. It is one of the best and adaptable computer vision algorithms because it can process 45 frames per second and can estimate up to 9000 and more seen and unseen classes of objects. It is also seen that YOLO works faster than RCNN, due to its elementary architecture. Antithetical to R-CNN algorithms that use regions to circumscribe the entities in images, YOLO instead applies neural network to the complete image to anticipate bounding boxes and their probabilities. YOLOv4 is finer than YOLOv3 in terms of Adequate Speed and Accuracy of Object Recognition.

Downloads

Download data is not yet available.

Downloads

Published

2021-07-05

How to Cite

[1]
S. S. Pattanshetti and S. I. Nivade, “Real-Time Object Detection with Pre-eminent Speed and Precision using YOLOv4”, IJRESM, vol. 4, no. 7, pp. 26–31, Jul. 2021.

Issue

Section

Articles