Two-Frame Temporal Player Detection System
A custom 2-frame temporal person detection model for tennis player localization in difficult court-view footage.
Project Snapshot
| Signal | Details |
|---|---|
| Timeline | Jan 2026 to Mar 2026 |
| Organization | Enhanced Robotics / Tenniix |
| Focus | Two-Frame Temporal Player Detection System |
| Stack | YOLOv8 |
Architecture
- Modified YOLOv8 into a dual-frame input pipeline.
- Used one frame as background context and the other as the active foreground target.
- Leveraged motion cues between consecutive frames to improve player localization.
Impact
Improves detection of partially occluded or visually unclear players across the tennis net, supporting tracking, gameplay analysis, and sports vision workflows.