https://github.com/kailash-natarajan/inr-currency-denomination-recognition-using-transfer-learning

INR Denomination Recognition is an image classification project
https://github.com/kailash-natarajan/inr-currency-denomination-recognition-using-transfer-learning

cnn deep-learning efficientnetb0 feature-extraction fine-tuning image-classification keras keras-tensorflow tensorflow tensorflow2 transfer-learning transfer-learning-with-cnn

Last synced: about 2 months ago
JSON representation

INR Denomination Recognition is an image classification project

• Host: GitHub
• URL: https://github.com/kailash-natarajan/inr-currency-denomination-recognition-using-transfer-learning
• Owner: Kailash-Natarajan
• Created: 2022-05-27T05:28:29.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2022-08-23T05:50:06.000Z (over 2 years ago)
• Last Synced: 2024-11-09T13:24:33.539Z (3 months ago)
• Topics: cnn, deep-learning, efficientnetb0, feature-extraction, fine-tuning, image-classification, keras, keras-tensorflow, tensorflow, tensorflow2, transfer-learning, transfer-learning-with-cnn
• Language: Jupyter Notebook
• Homepage:
• Size: 14.1 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# INR-CURRENCY-DENOMINATION-RECOGNITION-USING-TRANSFER-LEARNING

INR Denomination Recognition is an image classification project built using Tensorflow Keras.

* In this project, Indian Rupees Notes are classified according to their currency value, using deep learning techniques. 

* This project leverages the power of transfer learning. The pretrained base model used is EfficientNetB0. 

* This project is a demonstration of image classification and transfer learning (feature extraction and fine tuning) using Tensorflow Keras 

* Performs significantly better than CNN models built and trained from scratch

* The dataset used for this project is sourced from https://www.kaggle.com/code/kerneler/starter-indian-currency-note-images-f2ab9d26-d/data

* Dataset contains roughly 450 training images for each class, namely, Rs 10, Rs 20, Rs 50, Rs 100, Rs 200, Rs 500, Rs 2000 and Background. The background class signifies there is no Indian Rupees Currency Note in the provided image. There are also roughly 45 images per class for validation set. There are 91 imaages for the purpose of testing

* Model checkpoint is also made available