Our client, a leading European technology company, specializes in a platform that analyzes football matches and training sessions to evaluate individual and team performance through data-driven insights. Serving professional clubs, sports federations, and broadcasters, they faced challenges automating object and event recognition within their existing video analytics infrastructure. To address this, we developed a comprehensive computer vision system featuring video stitching, 2D/3D camera calibration, lighting correction, player and ball tracking, event recognition, and pose estimation. Seamlessly integrated into their architecture via microservices and APIs, the solution enhanced accuracy up to 93%, reduced manual annotation by 80%, and tripled match coverage without added operators. Our system operates in real-time under varied lighting, directly on edge devices, significantly optimizing sports analysis and broadcast efficiency.