Back to projects
2024
AI Esports Coach
ML & Computer Vision Analytics
A machine-learning coach that analyzes gameplay through computer vision and delivers performance analytics and actionable feedback to help players improve.
CV
Analysis
ML
Insights
Live
Feedback
Machine LearningComputer VisionPythonAnalytics
The Problem
Improving at competitive games is hard without expert feedback — most players cannot afford a coach and cannot see their own patterns.
The Challenge
Turning raw gameplay footage into meaningful, actionable insights that a player can actually act on.
Architecture
- Computer-vision pipeline to extract in-game events from footage.
- Analytics layer that aggregates performance patterns over time.
- Feedback generation surfacing prioritized areas to improve.
Workflow
- 1Ingest gameplay footage.
- 2Detect events and extract performance signals.
- 3Aggregate into analytics and generate coaching feedback.
Engineering Decisions
- →Computer vision over API telemetry to stay game-agnostic.
- →Prioritized, human-readable feedback over raw stat dumps.
Results
Prototype pipeline converting footage into structured performance insights.
Lessons Learned
Actionable, ranked feedback beats exhaustive statistics — players need to know what to fix next.
Future Improvements
- Real-time overlay coaching.
- Support for more titles.