Analysing Gemini and OpenAI performance for real-time sports feedback (github.com)

A recent project by Vision Agents showcased a real-time AI sports commentator that integrates Google's Gemini and OpenAI technologies for player identification and commentary. The system leverages Roboflow's RF-DETR model for immediate object detection, enabling it to identify players and the ball in fast-paced sports video footage. However, both Gemini and OpenAI's models struggled to deliver the necessary speed and accuracy for live commentary, rendering them insufficient for practical application in sports settings. This initiative is significant for the AI/ML community as it highlights the challenges of deploying advanced AI models in real-time environments, especially in applications requiring low latency and high precision, such as sports commentary. The findings underscore the need for further advancements in AI models to effectively handle dynamic visual data and deliver timely verbal output during live events. The project not only points to the current limitations but also sets a path forward for the development of more robust real-time models in the future.