WHOSCORED & SOFASCORE PLAYER RATINGS DECODED -
While working for one of the major football academies in India, I developed my own 'Player Rating Algorithm'. During my research, I came across the fact that goals, assists, defensive actions sure have highest weightage but...
Top of the mind improvements - Using the center of mass as the line-crossing anchor improves accuracy, avoiding missed reps when only part of the ball overlaps the line.
Your inputs and suggestions are appreciated.
Thanks @skalskip92 and @roboflow team for the tutorials.
Automating Football Juggling Counter with Computer Vision
I was checking out the Roboflow Rapid : Build a Vision Model in minutes segment.
Came across a line counter workflow so decided to test it on a video to automatically count football juggling repetitions.
How it works:
How it works:
- Football detection with RF-DETR (medium)
- Tracking via ByteTrack
- Rep counting triggered when the complete ball crosses a virtual knee-height line
- Live overlays: bounding boxes, labels, and juggling counts
Ball recoveries, tackles, interceptions never told the full story about pressing sequences.
Pressing is blocking passing lanes, forcing bad passes, initiating pressure.
A new model captures and rewards them all.
Minho Lee & Geonhee Jo may have built a goldmine.
Breakdown๐
I thoroughly enjoyed dissecting this awesome piece of work. If you come across more such research papers regarding football analytics, please share them.
Thanks for reading.