95% local AI + 5% GPT-6: a full basketball tracking pipeline breakdown
TheMoonMidas · x · 2026-09-11
Dev @skalskip92 shares a basketball analysis pipeline that runs 95% on local models, calling GPT-6 "Astra" for only 5% of the work: detecting ball and players, tracking and re-identifying players across plays, OCR of jersey numbers, recognizing the player in possession, detecting court keypoints, and mapping positions and trajectories. His takeaway: frontier models are powerful but expensive — use them only where they actually make a difference. A deep dive is linked.
More from coding & agent
- 'Model doesn't matter': is opinionated model routing in coding tools obsolete? — hugobowne · 2026-09-11
- Running agents and untrusted code in secure sandboxes: Google Cloud Run + Apps Script demo — rseroter · 2026-09-11
- opennews-mcp: open-source MCP server aggregating 85+ real-time sources into AI trading signals — tom_doerr · 2026-09-11
- GitHub's agentic workflows engine now runs your custom Pydantic AI agents to maintain repos — samuelcolvin · 2026-09-11
- Why chat assistants fail at delegation, and a security model to fix it — uriwa · 2026-09-11
- Claude Code vulnerability lets untrusted git repo escape macOS sandbox for RCE — ziv_ravid · 2026-09-11