US20250159309
2025-05-15
Electricity
H04N21/8146
The system introduces an augmented reality interface designed to enhance the experience of watching live sports games. By deploying a network of cameras around the game field, it captures real-time video footage and processes this data to create 3D models of players. These models are generated prior to the game based on the players' past performances. The system leverages deep learning techniques to identify players, extract skeletal and skin features, and animate 3D avatars that reflect real-time movements and poses.
The system comprises several components, including a memory for storing video footage and a processor-equipped object detection module. This module processes video frames to identify players and their positions on the field, continuously tracking their movements. The data is then transmitted to a spectator's device, allowing for the generation of a synthesized game field where 3D avatars are animated in real-time, providing a dynamic viewing experience.
Spectators can interact with the broadcasted game through a software application on their devices. This application allows users to manipulate various aspects of the game view, such as changing perspectives, adjusting zoom levels, replaying segments, and controlling playback. These interactions can be experienced on both 2D displays and through immersive technologies like VR goggles or smart glasses.
The system employs advanced deep learning models, including Convolutional Neural Networks (CNNs) and transformers, to map extracted features into skeletal and skin representations. It ensures smooth animation by filling gaps in player visibility across frames. The streaming architecture supports multiple protocols like HTTP Live Streaming (HLS) and WebRTC, ensuring efficient data transmission even with low bandwidth.
This technology provides sports enthusiasts with an enriched viewing experience by overcoming traditional broadcasting limitations. It allows spectators to view games from any angle with enhanced detail and interactivity. By reducing the need for extensive camera setups and computational resources, the system offers a scalable solution for real-time sports broadcasting.