Identifying Malaysian currency accurately, particularly for visually impaired individuals, presents significant challenges due to the variety and visual similarity of banknotes and coins. This study addresses these challenges by developing a sophisticated detection application utilizing the YOLOv8 model, designed to recognize and classify Malaysian currency with high precision. The research focuses on three primary objectives: analyzing recognition mechanisms in currency detection using deep learning, implementing a YOLOv8-based detection, and evaluating its performance through metrics such as confusion matrix, precision, recall, and mean Average Precision (mAP). A comprehensive dataset of Malaysian banknotes and coins was meticulously collected and subjected to rigorous preprocessing, including image resizing, data cleaning, and augmentation, to ensure high-quality inputs for the model. Various experiments were conducted using different dataset splitting ratios—70:30, 80:20, and 90:10—to identify the most effective configuration for accurate detection. The 90:10 split produced the most favorable outcomes, with a Box Precision of 0.84, Recall of 0.923, and mean Average Precision (mAP) values of 0.959 at IoU 0.5 (mAP@50) and 0.84 across IoU thresholds from 0.5 to 0.95 (mAP@50-95), indicating robust object localization. This research introduces an innovative approach by incorporating both banknotes and coins into the detection framework and applying the YOLOv8 model, which has not been widely used in similar studies. The findings highlight the model’s effectiveness in addressing the complexities of Malaysian currency recognition. Future research could explore expanding the dataset and integrating counterfeit detection capabilities to further enhance the application’s reliability and practical applications.