Target Tracking
Robust detection and tracking of objects in dynamic, fast-paced environments across aerial and ground platforms.
Developing robust target detection and tracking pipelines for dynamic real-world environments, spanning aerial surveillance, robotic platforms, and sports video analytics.
Key contributions:
- Multi-object tracking across drone and ground-based platforms using deep learning models.
- Racquet detection, shuttle tracking, and hit detection in fast-paced sports video at Visist.AI.
- Aerial object detection with robustness across varying slant angles and weather conditions.
- Evaluated SOTA detection models for aerial imagery analysis as part of a 3-year collaborative Sim-to-Real project (2022–2025).