Language selection

Search

Government of Canada Data Conference 2023: Indigenous Communities Food Receipts Crowdsourcing with Optical Character Recognition (DDN3-V10)

Description

Presenting a crowdsourcing web application using machine learning optical character recognition algorithms to collect and to extract food price information from pictures of grocery receipts from isolated Indigenous communities.

Duration: 00:02:44
Published: February 15, 2023

Event: GC Data Conference 2023: About the conference


Now playing

Government of Canada Data Conference 2023: Indigenous Communities Food Receipts Crowdsourcing with Optical Character Recognition

Transcript

Transcript

Transcript: Government of Canada Data Conference 2023: Indigenous Communities Food Receipts Crowdsourcing with Optical Character Recognition

Everyone deserves access to healthy, affordable food, no matter where they live. However, many Canadians living in northern and isolated communities face increased costs related to shipping rates and supply chains. As part of the Government of Canada's response to food insecurity in the North, the Nutrition North Canada subsidy program was established. This program helps make nutritious food like meat, milk, cereals, fruit and vegetables more affordable and more accessible. In order to better understand the challenges impacting food security, better price data is needed.

Hello, my name is Joanne Yoon and as a lead data scientist representing Statistics Canada's Data Science Division today, I'll introduce you to our crowdsourcing web application which aims to extract key information from food receipts uploaded by residents in isolated Indigenous communities. Using this method to collect food price and subsidy data will allow us to develop and share valuable insight into the cost of living in these communities.

Here is how it works: a user, likely a resident, will upload a photo of their grocery receipt to our secure web app. Our web app is built with machine learning algorithms, such as Optical Character Recognition and natural language processing. These algorithms will extract the location, date of purchase, price and subsidy information from the receipt image. Sensitive information such as personal identifiers will be redacted. Before submitting, the user can review and edit the extracted information. In our design, we're take full advantage of the latest advances in machine learning to develop a model that accurately extracts data from various types of receipts.

Not only will this application be available to residents, but so will the data—they'll be able to compare prices before making financially and nutritionally-informed choices.

The aggregate anonymized data we collect will provide better insight on issues associated with the high costs of food in isolated Indigenous communities and will improve the transparency and accountability of the Nutrition North Canada subsidy recipients to residents in the communities.

We are excited to build this web application and unlock an alternate data source to complement our current and ongoing data collection.

For more information, email us at datascience@statcan.gc.ca.

Related links


Date modified: