What is it?
Computer vision engineers build software that can make sense of images and video. They train models to spot objects, follow them as they move, and turn what the camera sees into data people can use.
In football, that means tracking every player and the ball from match footage, measuring distances and speeds, spotting key moments and cutting highlights automatically — without anyone having to tag the video by hand.
What you'd actually do
Data
Collect match footage and label players, the ball and pitch lines so a model has examples to learn from.
Training
Train and fine-tune detection and tracking models, then compare versions to see which is more accurate.
Testing
Find the cases where the model gets it wrong — bad lighting, players overlapping, odd camera angles — and fix them.
Shipping
Make models fast and small enough to run on real footage, and work with app developers to put the results in front of users.
What to learn
Programming
Python is the everyday language; some C++ helps when speed matters.
Maths
Linear algebra, geometry and probability sit underneath every model.
Machine learning
Deep learning, especially convolutional networks and modern detection models.
Image fundamentals
How cameras, pixels, colour and perspective work — e.g. mapping a camera view onto a flat pitch.
Your roadmap
A step-by-step route into football computer vision. Each stage builds on the last, and ends with something you can actually do with match footage.
- Python
- NumPy
- OpenCV
- Video processing
- You can: Read and process football videos
Python Programming
- Linear algebra
- Probability
- Statistics
- Geometry
- You can: Calculate player positions and distances
Maths & Data
- Supervised learning
- Datasets
- Model evaluation
- Overfitting
- You can: Learn to distinguish players from the ball
Machine Learning
- Neural networks
- CNNs
- PyTorch
- Transfer learning
- You can: Understand how AI recognises visual patterns
Deep Learning
- Object detection
- Segmentation
- Tracking
- Homography
- Pose estimation
- You can: Detect and follow players across video frames
Computer Vision
- Pitch mapping
- Event detection
- Player identification
- Ball tracking
- Event recognition
- Performance metrics
- Deployment
- You can: Automatically produce stats, heatmaps and highlights
Football AI Systems
Tools you'll use
Ways in
University
A degree in computer science, engineering, maths or physics, often followed by a master's in AI or machine learning.
Apprenticeship
Software or data apprenticeships can lead into ML and vision roles as you build experience.
Self-taught
Free online courses plus a strong portfolio of vision projects can get you noticed.
Try it yourself
- Detect players in a single frame of match footage using a pre-trained model.
- Track the ball across a short clip and draw its path.
- Turn player positions from a broadcast angle into a top-down tactical map.