e-XCheck

Shop faster, Checkout smarter: Experience seamless checkout.

Comment

Inspiration

We thought about the current shopping process and how slow it was inefficient to scan an item one at a time. With this project, we wanted to allow an instant scanning of all items at once to speed up the shopping process. And in the process help to increase accessibility in the retail industry.

What it does

The project takes a camera, which after all items are under its view, there is a button to press to capture an image. The machine then processes the image and lists out all the items in a list with the corresponding price. After customer verification, they would press the checkout button to pay and be done with their shopping.

How we built it

An essential part of this project is a machine learning AI. We used images of various products and from various angles to train our machine-learning AI to recognize items with Azure. After which, tests are run on the AI to see the result of its training. Upon perfect success rate, AI is now running on a website, which has a camera connected to function as a scanner like one in a market.

The website is built in angular.js with ant design of angular for the front end and typescript for the back end. Data such as prices and dates would be stored upon the customer pressing the checkout button with firebase. These data can be accessed by the backend or employees to see how the business is doing and follow trends by self-generated graphs for analysis for the next step for the business.

Challenges we ran into

The machine learning process was long and tedious, requiring long hours and various data. An example is that some items needed to be in various angles for them to be able to fully recognize the item with no error.

Another challenge was that the graph implementation was messing up the code in the back end, which it almost destroyed the entire project, luckily a fix was applied now everything is working fine.

Accomplishments that we're proud of

Hooking up external hardware that would work with the program to bring out the full capabilities of this project. Another accomplishment that we are proud of is the machine learning AI and how we can train the AI to be mostly flawless in recognizing items.

What we learned

How to do machine learning with image recognition, different cloud native solutions to AI inferencing.

What's next for eXCheck

We would like to be able to station this technology on some sort of belt, like a conveyer belt. We want to have it so that the scanner would take a picture once every time to see what was being purchased as the items pass through on the conveyer belt. This sometimes would be set by the length covered by the scanner, and the speed of the conveyer belt to find the most accurate time to take a picture once in a while. This way customers can just dump their order on an empty conveyer belt, walk to the end of the conveyer belt, pay, pick up their items, and leave. That way, customers' time will be respected and allow smoother process at the checkout.

Built With

Share this project:

Updates