Grocery Chum
Posted on Tuesday, April 11th, 2023
Client | BR[AI]YT |
Professor(s) | Dr. Hari M. Koduvely |
Program | Artificial Intelligence Software Development |
Students | Binulal Narayanan,, Xun Peng,, Esam Elsheh,, Thomas Sinclair., Shifeng Song |
Project Description:
BR[AI]YT is a leading artificial intelligence (AI) solutions provider that assists public and private organizations in achieving their business goals by applying advanced AI techniques. With extensive experience in designing, developing, deploying, and maintaining AI solutions across a diverse range of business domains, BR[AI]YT specializes in helping small and medium-sized organizations integrate Automation, Cloud, Data Analytics, Discovery, Ethics, and Strategy into their operations.
BR[AI]YT’s engineering teams are currently working on the development of various modules aimed at creating a fully autonomous, cashier-less retail experience. The primary objective of the project is to develop neural network modules capable of detecting and tracking individuals’ movements within a physical space using multiple cameras, all while sharing data simultaneously.
Our team is focusing on a specific portion of this ambitious project, aiming to digitize receipts and invoices from multiple stores. The system must be robust enough to process images with varying intensities, shapes, and orientations and extract text and numbers accurately. Essential information includes the store’s name, the date the receipt was issued, the items purchased, and the cost of each item. The system must be smart and flexible, as input images will be highly variable.
The goal is to accurately parse information from different stores, each with unique receipt formats. Additionally, the digitization process must account for images acquired using various devices and sensors, such as smartphone cameras with distinct post-process coloring and image qualities. The system should require minimal effort from the user in terms of image quality, accommodating less-than-optimal lighting and possibly slightly blurry images. A standard for image processing must be devised to create a flexible and robust system.
This project offers numerous opportunities and applications. One potential use case is market research into consumer habits, which stores can leverage to gain a foothold in the industry. By tracking individuals’ spending habits, companies can better understand their target demographics’ behavior. Another potential application is a smartphone app that enables the general public to monitor and manage their spending habits.
By digitizing receipts for easy processing, the project aims to handle various image qualities, store formats, and extract key information effectively. The development of this innovative system will revolutionize retail experiences, offering a seamless and efficient solution for businesses and consumers alike.