Patriot One’s Xtract AI partners with Canadian Armed Forces, develops AI to detect COVID-19
Patriot One Technologies’ Vancouver subsidiary Xtract Technologies (Xtract AI) has signed a nearly CAD $200,000 contract with Canada’s Department of National Defense to work with the Armed Forces on hiding soldiers and vehicles with AI, as part of the Innovation for Defense Excellence and Security (IDEaS) Program, the company announced.
“We’re delighted that the Innovation for Defense Excellence and Security (IDEaS) program selected us to work on this project,” said Martin Cronin, CEO of Patriot One Technologies, in a prepared statement. “Utilizing Xtract AI’s knowledge of artificial intelligence, and data and video analysis we hope to be able to assist the Canadian Army in their ability to operate covertly, thus keeping our soldiers safer.”
The “Now you see me, now you don’t” project is seeking methods to change the visual and infrared signatures of Army platforms for covert operations. By leveraging deep learning and computer vision, Xtract AI will develop a system that evaluates adaptive and multi-spectral concealment and camouflage technologies or materials. It merges two deep learning components: a soldier and vehicle detector; and a soldier and vehicle concealer to identify people and vehicles from visual and infrared video streams. The concealer is a computer vision model that decides how to facilitate concealment based on real-time context information.
Xtract AI develops AI tech to detect COVID-19
Xtract AI is also working with Amazon Web Services (AWS), Vancouver General Hospital (VGH), the University of British Columbia (UBC), and SapienML to develop AI technology that radiologists can use to identify COVID-19 risks.
Project leads are Dr. Savvas Nicolaou and Dr. William Parker, who have the support of the UBC Community Health and Wellbeing Cloud Innovation Center (UBC-CIC), powered by Amazon Web Services (AWS).
The data used is collected from CTs and X-rays from multiple countries, with comments from 14 volunteering radiologists. The images are labelled to be categorized into background, normal lung, and Ground Glass Opacity (GGO). The percentage of lung volume affected by GGO is a signal for the disease.
The images are used to train 3D residual networks to automatically detect GGO volumes and compared them to the total volume of the lungs. The model will keep being improved until it reaches more than 90 percent diagnostic accuracy. The model is available for healthcare facilities around the world to use under an open source license.
“AI models are not magic, but this model we’ve developed is an open source gift to start answering questions,” said Dr. William Parker, project lead and radiology resident at UBC, in a prepared statement. “If we are going to get to the point of helping patients, we need to know the strengths and weaknesses of the models, and we need to have an understanding that not all models are created equal. The goal of our AI model is to drive feedback so that we can improve the model and make it available for clinical use in the fall of 2020.”