The applications of currency counting are vast in real-time. From the ATM to small entrepreneur business applications includes currency recognition. In contrary, currency recognition is one of the HCI applications that can be taken for any human-computer application. This technology limits the manpower and human supervision. The emerging technologies such as AI and Machine Learning were now part of every application. In this regard, the proposed methodology is a robust methodology built on deep learning technique using relatives less number of images in the database with high noise, including rotated notes, tilted notes, held by hand note images. The proposed methodology extracts denomination in 90 s. It uses a Convolution Neural Network (Inception using transfer learning) to extract currency denomination. We acquired the accuracy of 83% in predicting the denomination.