Philippine Coin Detection System Using Mask R-CNN Algorithm

This paper introduces a Philippine Coin Detection System utilizing the Mask R-CNN Algorithm for accurate classification and denomination of 1, 5, 10, and 20 peso coins. Image preprocessing techniques, such as Gaussian smoothing and Canny edge detection, are applied prior to classification. With a 70 % accuracy for multiple coin detection, the evaluation reveals specific coin accuracies: 1-peso (90%), 5-peso (87.5%), 10-peso (85 %), and 20-peso (92.5 %), with an overall accuracy and precision of 77.5 % and 78.33 % respectively. The study signifies a significant advancement in automated coin recognition systems with potential implications for financial automation and related applications.