The validity and integrity of money transactions, and banknote categorization is essential in financial systems. The accuracy and effectiveness of banknote categorization systems have recently been greatly improved by the introduction of machine learning techniques. To offer a full overview of the developments achieved in the field of banknote categorization using machine learning, this study presents a thorough literature review on the subject. The survey discusses several machine learning algorithms, feature extraction methods, datasets, assessment standards, and difficulties in classifying banknotes. To extract useful information from banknote photos, feature extraction approaches, including texture analysis, colour analysis, and geometric characteristics, are investigated. For training and assessment purposes, the study makes use of databases on banknotes that are available to the public. The results of this study will help create reliable banknote categorization systems, improving the trustworthiness and integrity of financial transactions.