Interactive SAM3 Auto-Labeling for Tennis Ball Detection
An interactive SAM3-based auto-labeling pipeline for accelerating tennis ball detection dataset generation.
Project Snapshot
| Signal | Details |
|---|---|
| Timeline | Dec 2025 to Feb 2026 |
| Organization | Enhanced Robotics / Tenniix |
| Focus | Interactive SAM3-Based Auto-Labeling Pipeline for Ball Detection Dataset Generation |
| Stack | SAM3 |
Workflow
- The user clicks the ball in the first frame through a GUI.
- SAM3 tracks the ball across the remaining video frames.
- The pipeline generates annotations for the full video.
- The generated labels are used to train ball detection models.
Impact
Reduces manual labeling effort, improves annotation consistency, and speeds up training-data creation for ball detection.