While a typical human being has little trouble perceiving the differences between different banknotes, a person who is blind or has visual impairments would have a much more difficult time performing the same task. Those who are blind and mentally handicapped have a critical need for a system that can detect and identify currency in real time. Money is essential for conducting business, thus its prevalence in daily life is not surprising. An efficient and effective approach for detecting and recognising Indian currency relying on a CNN model is offered for use on mobile devices. Extreme Learning Machines were used in place of fully connected layers of pretrained trained MobileNeV2 to get more resilient result. According to the findings of the tests, the mobile net model-based approach that was recommended shows detection accuracy of 97.80%. This independent system performs its operations in real time.
Key words: Convolutional Neural Network, MobileNet, Extreme Learning Machine
|
