Does Camera Position and Capture Style Affect Tracking Quality?

Video Capture setups vary from stadium to stadium so we set out to test how much camera position and capture style affects tracking quality and physical performance data. More specifically we wanted to investigate SkillCorner’s capability to provide consistent and reliable data from a range of different capture scenarios - including high cameras, gantry cameras, low cameras, off-center cameras, clean broadcast, dirty broadcast, wide angle capture plus telescopic cameras used at academy and ‘grass routes’ levels.

To validate the robustness of our single-camera broadcast tracking solution, our Analysis team conducted a controlled multi-camera experiment at Sunderland AFC’s Stadium of Light, placing 10 cameras at different heights, angles, offsets and zoom levels to replicate the wide range of broadcast and tactical setups seen across stadiums and training grounds throughout global football. Outputs from each camera against ground-truth data from the official English Premier League tracking system, assessing four areas that matter most to clubs; 

• Player Detection and Identification
• Ball Tracking Accuracy
• Physical Performance Metrics
• Player Ranking Consistency

‍

The Results

The findings showed that SkillCorner data remains robust and reliable across real-world football environments, regardless of camera placement or feed format.

SkillCorner maintained high tracking accuracy and consistent physical outputs across all camera positions and video feed conditions tested. Player identification ranged from around 90% to 98%, ball tracking remained above 92% for most camera positions, and physical metrics such as PSV-99 stayed closely aligned with ground truth. Player rankings also remained highly consistent across setups.

‍

Technical Report 

Download the technical report for more information on the methodology and results. Download here.

‍

Video Explainer

Watch this short video for a walkthrough of the project and its headline findings.