TVConal, a Singaporean startup participating in NVIDIA's Inception incubation program, is leveraging NVIDIA's Metropolis and TAO toolkits to build an artificial intelligence (AI) video analysis platform that automatically tags various events during sporting events.
TVConal stated that since sports events often generate large amounts of data, it is necessary to use artificial intelligence to analyze players' performance during the event from this large amount of data, and at the same time, to analyze players' performance in a quantitative way.
For popular sports like cricket, tennis, badminton, and football, TVConal is leveraging the NVIDIA Metropolis application framework to build an artificial intelligence visual image analysis platform. This platform can identify key events in a game, analyze or predict a player's next moves, and enable more accurate analysis of subsequent matches.
Masoumeh Izadi, General Manager of TVConal, explained that marking events during a match will facilitate the establishment of a timeline of events, serving as an important basis for event analysis. However, in the past, this often required the assistance of 20 recorders. However, with the assistance of artificial intelligence systems, time marking of events in a match can now be completed quickly. Furthermore, through artificial intelligence recognition and analysis, fouls that occur during a match can be corrected, and key moments that are difficult for the human eye to judge can be identified.
The NVIDIA solutions used by TVConal include the NVIDIA Metropolis toolkit for building an image analysis platform, as well as numerous SDKs and CUDA-X libraries. They are also optimized for execution on certified systems based on the NVIDIA EGX enterprise computing platform.
Other parts include using the NVIDIA DeepStream SDK to simplify the image processing process, accelerating artificial intelligence training efficiency with NVIDIA pre-trained models and the TAO toolkit, and using the NVIDIA TensorRT development toolkit to enhance inference capabilities.
In addition to TVConal, startups including Pixellot, Track160, and Veo also use the NVIDIA Metropolis toolset for sports analysis applications, significantly improving work efficiency through the assistance of artificial intelligence systems.
