Sportlogiq is a Montreal, Canada-based startup that uses deep learning and NVIDIA GPUs to collect sports analytics from a standard video in real time. The company, a member of the NVIDIA Inception Program, recently showed off their technology at the annual Computer Vision and Pattern Recognition Conference in Salt Lake City, Utah this week.
“Sportlogiq is the intelligent sports analytics platform, it can work with broadcast feeds or any other visual information. The system understands the content of a sports game,” Mehrsan Javan, the company’s chief technology officer, said. “Computer vision is hard and everything in computer vision requires a lot of processing power so we need powerful hardware, and that’s why we selected NVIDIA,” Javan said.
Using NVIDIA Tesla V100 GPUs, Quadro P6000, and GeForce GTX 1080 Ti GPUs, the company trains their system on thousands of hours of real sports footage. With multiple cuDNN-accelerated deep learning frameworks, the company developed a neural network that can recognize players on the field and their actions, including analytics about their performance.
“Our software is able to transform raw video data into advanced sports analytics,” the team said. “These meaningful insights allow teams, broadcasters, analysts, and fans to experience the game in new ways.”
The company is focused on hockey but they are quickly expanding into soccer and other popular sports. The company says they will soon announce a new customer in a major sport in North America.