Application for archiving receipts and analysing expenses

The aim of the thesis was to develop a mobile application for archiving photographs of receipts and supporting expense analysis. The application implements automatic text detection and recognition from receipt images using convolutional (CNN) and recurrent neural networks (RNN), with a focus on a specific type of RNN, LSTM (Long Short-Term Memory). The system also allows manual data entry from existing image files and manual correction of any text recognition errors caused by inaccuracies of OCR technology.

The thesis included the analysis and selection of image processing technologies and OCR algorithms, with an emphasis on advanced character recognition techniques, including neural networks. The end result is a user-friendly, scalable mobile application that significantly simplifies expense tracking thanks to automated data entry and clear analytical features.