At one point or another in our lives, we’ve all been frustrated by the time and effort it takes to split a bill at the end of a good meal. This is why we built Share—to share not only the food and the love, but also the responsibility to make paying easier for everyone.
What it does
An iOS app, Share allows users to share the bill information with others by processing images of receipts for individual items and prices. The bill can be split evenly among the group, or each person can select and pay for a subset of items on the bill.
How we built it
To analyze the receipt, we first used OpenCV libraries to detect edges in the picture. For each cluster of edges, we computed a bounding rectangle containing all of the edges in that cluster. Next, we linked Tesseract OCR to parse the text from each bounding rectangle. With the extracted text and their positions, we ran a line-sweep algorithm to merge nearby rectangles, matching the name of the item with its corresponding price.
Challenges we ran into
Since receipt formats vary from place to place, we had to take care of several edge cases when processing. This included merging multiple blocks of text to match with one price (i.e. keeping the entree and drink in the same order) and detecting noise during the text extraction phase with Tesseract.