Date of Award
5-2021
Culminating Project Type
Thesis
Degree Name
Computer Science: M.S.
Department
Computer Science and Information Technology
College
School of Science and Engineering
First Advisor
Julstrom Bryant
Second Advisor
Sarnath Ramnath
Third Advisor
Anda Andrew
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Keywords and Subject Headings
Object Detection; Image Recognition; OCR; AI; World Wide Web; YOLO
Abstract
The world in the 21st century is ever evolving towards automation. This upsurge seemingly has no decline in the foreseeable future. Image recognition is at the forefront of this charge which seeks to revolutionize the way of living of the average man. If robotics can be likened to the creation of a body for computers to live in, then image processing is the development of the part of its brain which deal with identification and recognition of images.
To accomplish this task, we developed an object detection algorithm using YOLO, and acronym for “You Only Look Once”. Our algorithm was trained on fifty thousand images and evaluated on ten thousand images and employed a 21 x 21 grid. We also programmed a text generator which randomly creates texts and URLs in an image. A record of useful information about the location of the URLs in the image is also recorded and later passed to the YOLO algorithm for training.
At the end of this project, we observed significant difference in the accuracy of URL detection when using an OCR software or our YOLO algorithm. However, our algorithm would be best used to specify the region of interest before converting to texts which greatly improves accuracy when combined with OCR software.
Recommended Citation
Ajala, John, "Object Detection and Recognition Using YOLO: Detect and Recognize URL(s) in an Image Scene" (2021). Culminating Projects in Computer Science and Information Technology. 37.
https://repository.stcloudstate.edu/csit_etds/37
