Scribd
Upload
173 views

INFRAMIND Season3 Round2 Businesscase EdgeComputing

Kirana, a retail store, wants to automate its checkout system using edge computing to reduce human workforce at billing counters. The system should allow automatic product detection using a camera based on size, type, and price to generate bills in real-time without cloud connectivity. An edge server with ML model would detect products from images, lookup prices, and add items to a bill of materials displayed in a "Kirana Product Billing" web app. The solution architecture would involve training the ML model on cloud, processing images on an edge server, and demonstrating the working app.

Uploaded by

amol chaudhari
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
173 views

INFRAMIND Season3 Round2 Businesscase EdgeComputing

Kirana, a retail store, wants to automate its checkout system using edge computing to reduce human workforce at billing counters. The system should allow automatic product detection using a camera based on size, type, and price to generate bills in real-time without cloud connectivity. An edge server with ML model would detect products from images, lookup prices, and add items to a bill of materials displayed in a "Kirana Product Billing" web app. The solution architecture would involve training the ML model on cloud, processing images on an edge server, and demonstrating the working app.

Uploaded by

amol chaudhari
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

EDGE COMPUTING

Requirement
Kirana is a retail store looking for more digitalized way of expanding their business. They want to use a
more systematic way of checkout system and reduce their human work force at billing counters.

Help Kirana devise a system that allows automatic detection of product using camera. The detection of
product must be wrt the size of the product, type of product and automatically take the cost of product
to make a bill of materials at checkout. This has to be done in real-time without sending the data to cloud
for processing as some of these stores can be in remote areas with intermi ttent connectivity and Kirana
do not want customers to wait due to latency issues for connectivity.

Example: If a toothpaste is placed in front of the camera, it should detect the product, according to the
size and weight of toothpaste it must understand what the price of the product is. The details of each
product scanned is then presented at the bill of materials for payment.

Please note, Kirana has Edge Servers that are not compute intensive.

Things to do:

 Build a ML model for product detection based on the mentioned parameters.


 Size of product
(When a customer keeps a product in front of the camera, it should detect the product
by determining the size of product. It should be able to distinguish between the products
of same brand, but of different size. Refer the toothpaste example above.)
 Train the ML Model.
 Create a web application named ‘Kirana Product Billing’ for Product Detection and Billing.
Integrate it with the trained ML model and show the details of the bill for products bought by the
customer.

Expected Output:

1. Open ‘Kirana Product Billing’ web application.


2. Upload image of a product for detection (You can also use desktop/Laptop camera for product
detection)
3. Based on the image, perform product detection through the ML model built for detecting product
based on size of product.
4. Once product is detected, directly add the corresponding Product ID, Product Name, Quantityand
Cost of detected product to the Bill of Materials.
5. Repeat the above steps for atleast 5-6 products. Amongst these products, include 2-3 images of
the same product, but in different size. Example: If a customer buys three toothpaste in small,
medium and large size each.
6. After images of all products are uploaded, show the final bill of materials. It should display the
following details
I. Sr. No.
II. Product ID(This is a unique ID which will be different for all the products regardless of the
brand and size
III. Name of Product
IV. Quantity
V. Cost of product
VI. Final Cost of Product(=Quantity*Cost of Product)
VII. Highlight ‘Final Cost’ by performing addition on the final cost of all the products bought

Sharing an example view of Bill of Materials (You can use your own details for products)

Sr. No Product ID Name of Product Quantity Cost of Final Cost


Product of Product
1 234 XYZ Toothpaste-Small 1 20 20
2 247 XYZ Toothpaste-Medium 2 30 60
3 351 XYZ Toothpaste-Large 1 40 40
4 114 ABC Potato Chips-Small 1 10 10
5 256 ABC Potato Chips-Large 1 50 50
180 Final
Cost

In the video, show the following:

 Details of technology and services used in the solution.


 Mention hardware/software utilized in the solution.
 Explain with a High-level Reference Architecture diagram, the proposed solution in a real Retail
Store scenario considering Edge Computing capabilities.
The Reference Architecture must cover
 How images from camera will be sent to Edge Server(this can be your
desktop/laptop/Raspberry Pi) for processing(any protocols/methods)
 How Cloud services can be used for model training(mention which Cloud Platform and Cloud
service)
 How Machine Learning Inference can be used in the solution
 Flow chart of the solution
 Show the working demo of “Kirana Product Billing’ web application.
 Show the developed ML code and code for web application.

Points to Remember:

 The Edge Server is not compute intensive, plan model training and application deployment
accordingly.
 ‘Kirana Product Billing’ is a web application which will run on the Edge Server(can be your
desktop/laptop/Raspberry Pi)
 You can use Python for development of code. If you are comfortable with any other development
language, feel free to use it.
POINTS TO REMEMBER
The below listed will be the upcoming events/ process next week onwards for Inframind Season 3 Round
2:
• For each tower, there will be a webinar session organized with the TCS SMEs to help you further
understand the problem statement. the Webinar announcements/links will be shared via emails
,Campus commune channels and text messages.
• For further more details regarding Inframind Season 3, students can visit the “Inframind Season 3 “and
“IT Infrastructure Services - Powering IT Infrastructure Globally - Be the Future!” channels on Campus
Commune.
• Pls utilize the webinars to help enhance your understanding of the case study; your queries will also be
addressed on this platform. Mentors will be assigned to you as required during these Webinars .

You might also like