<![CDATA[Personal financial management still faces limitations in both manual recording and conventional applications, such as low consistency and bias in expense categorization. This study develops a mobile application for personal finance automation using the waterfall method, integrating Optical Character Recognition (OCR) and Deep Learning to automatically record and classify expenses. The dataset consists of 900 images of local transaction receipts with varying print conditions. Text extraction is performed using a Convolutional Recurrent Neural Network (CRNN) and compared with the baseline Tesseract OCR. For expense classification, a CNN model with EfficientNet fine-tuning is applied Evaluation results show significant improvements with a character accuracy of 97.05%, word accuracy of 92.1%, and an F1-score of 82%. Transaction input time was reduced by an average of 62% compared to manual recording. A usability test using the System Usability Scale (SUS) with 36 respondents yielded a score of 70.069. The main contribution of this study is the integration of adaptive OCR and deep learning–based classification in the context of Indonesia’s local financial management.]]>
Copyrights © 2026
