We're not sure if necessity is indeed the mother of invention, but when a health crisis shuts down your entire market (sports and entertainment) that your startup is trying to serve, it's definitely a setback!

Like most startup founders and entrepreneurs after the shock of the pandemic eased, our minds started turning to the problems ahead.

A lot of the stories we shared were about the lack of screening in Canadian airports: personal anecdotes, and that of friends and family.  Enhanced screening should go beyond a few questions on a touch screen. We decided that by combining our skills we could help the COVID-19 crisis and better equip society for subsequent waves and what might come next.

What it does

Combining thermal imaging cameras with machine learning our platform can detect and track sick people as well as predict where and when anomalies might be occurring. The platform uses both temperature as well as thermal image data to ascertain anomalies.

Our algorithm allows us to establish a baseline for individuals and groups of people to build a data set on thermal detection.

How we built it

Cleard AI combines thermal imaging cameras to a machine learning algorithm via a front end interface.  There are two main elements:

  1. backend engine built using Google platform for storage, AI, vision and machine learning.
  2. The front end user interface is javascript html running on node JS. Connected via Python.

Challenges we ran into

Currently, there are no thermal imaging mass data sets available on people in the public domain. There is a gap in the market so we used sample data from other industries to train the initial AI for proof of concept. We require data before, during and after health fluctuations.

Another major challenge is access to people during this crisis to take photos. We established a non-contact, socially distant method to capture thermal image photos of friends and family to establish some initial images to work with.

Accomplishments that we're proud of

Over the course of the hackathon we have:

  • built and trained our algorithm testing over 125K images from sample data
  • launched our website
  • built a front end interface for proof of concept
  • validated our idea with industry (health care and airport stakeholders), investors, and government agencies
  • conducted user research to validate end-user consent in thermal image screening in public locations such as airports, stadiums, etc.
  • created our plan for beta field testing

What we learned

  • We have the ability to make a massive contribution to thermal imaging data sets
  • There are multiple use cases for our technology including airport screening, triage in public health, providing essential services and businesses the data and confidence they require to re-open
  • 82% of respondents surveyed indicated that they are willing to submit their thermal images for research purposes
  • 70% of respondents agreed that pre-screening should become part of the check-in experience for travel and mass events

What's next for Detect, Track & Respond to Virus Threats Using IR Screening

  • Train the AI with real data. We are conducting a beta field test week of March 31st to capture baseline data on healthy and sick people to train our AI with real data.
  • Partner with an airport and or public or private health organization to conduct a screening beta test to determine if we can effectively screen large numbers of people and deliver meaningful triage data.
  • Raise capital to purchase hardware, staff the system in public spaces, support cloud storage fees and develop a real-time delivery mechanism.



auto-ml, google, google-cloud, javascript, machine-learning, mongodb, neural, node.js, python, redis, vision, vision-api

Devpost Software Identifier