Skip to content

Real and Counterfeit Banknote Detection using YOLOv5 and Image Filters

Notifications You must be signed in to change notification settings

murtazahh/Counterfeit-Banknote-Detection

Repository files navigation

Real-and-Counterfeit-Banknote-Detection-using-YOLOv5-and-Image-Filters

Real and Counterfeit Banknote Detection using YOLOv5 and Image Filters

One of the projects that I have completed as a Data Science - Deep Learning Expert at Fiverr was to develop a program that can detect real and counterfeit banknotes using a mobile phone’s camera. I used the YOLOv5 model to train the program on a dataset of original banknotes, after applying Weiner and Gray scale filters to enhance the image quality. The program can take a new picture of a banknote and calculate the probability of it being real or counterfeit, without requiring any internet connection or external libraries. I also provided all the necessary files and code lines to ensure that the program can run smoothly and reliably on the client’s side. The program achieved an accuracy of 95% on the test set and received positive feedback from the client.

About

Real and Counterfeit Banknote Detection using YOLOv5 and Image Filters

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published