Video Analysis
Upload a broadcast clip → detect players & ball → track across frames → project to pitch coordinates → spot events. Download JSON/CSV.
Drop a broadcast clip here
or click to browse. Supports MP4, WebM, MOV, AVI (max 500MB)
Broadcast footage has frequent camera changes. Calibration and tracking reset per shot — expect gaps in tracking during cuts.
Shirt number / re-ID is a separate task. Track IDs are consistent within a shot but not across camera changes.
Ball visibility is intermittent in broadcast. The ball tracker is the most fragile component — expect missing detections.
Full pipeline processing can take 2-10× the clip duration depending on resolution. Short clips (<30s) recommended.
Soccer-tuned YOLO model from Hugging Face. Detects players, goalkeepers, referees, and ball with bounding boxes.
ByteTrack for multi-object tracking. Assigns consistent IDs across frames within each camera shot.
SoccerNet camera calibration tooling. Estimates homography to map pixel space → pitch coordinates.
SoccerNet pretrained action spotting models. Detects goals, cards, substitutions, and 17 action classes.
SoccerNet BAS baselines for dense on-ball events: passes, dribbles, duels — closer to an event stream.
Tracking: frame, timestamp, track_id, class, confidence, bbox, pitch_x, pitch_y. Events: timestamp, label, confidence.