LIBL Bill Automation

Abstract: The project deals with day-to-day account settlement. It is a simple application used to make expense management quicker and easier. It uses OCR methodology to read and process receipts which can be converted to spread sheets, documents, or CSV. The user can take/upload a receipt image for recognition of the images. The bill images are captured through camera by image processing and then the data are extracted and also analysed by recognizing the text. All these actions are performed by using an OCR tool, which performs text detection and recognition on the pre-processed image using the Deep Learning (Two-Step CNN Framework) model. Once the image is clicked, it is then encoded and sent to the server. The extracted text data is used to extract the relevant information (such as date, company name, items, total, etc.). This data is then sent back in the desired format

Keywords: OCR, CNN, Text Recognition, Character Segmentation