Research at Gabriella Systems
Movement intelligence, built on trustworthy foundations
We study how computer vision and responsible AI can turn everyday sports video into useful, measurable, and trustworthy movement intelligence.
Research pillars
Four questions driving our research
Each pillar is an open set of research questions — not a finished product.
Athlete Perception
Understanding athletes, equipment, and movement from ordinary video.
- Human pose and movement understanding
- Bat and ball tracking
- Temporal action understanding
- Event detection
- Occlusion and viewpoint robustness
Movement Intelligence
Moving beyond pose estimation toward understanding how movement unfolds through time.
- Batting sequence and timing
- Bowling phase analysis
- Hip–shoulder coordination
- Repetition consistency
- Session-to-session change
- Interpretable movement metrics
An active research direction: a structured representation of how batting or bowling movements unfold in sequence. Not a validated clinical or scientific metric — research in progress.
Trustworthy Sports AI
Sports AI should communicate when its measurements are reliable — and when they are not.
- View-aware measurement
- Analysis confidence
- Uncertainty estimation
- Metric reliability
- Camera-angle sensitivity
- Transparent limitations
- Reproducible measurement
Research toward structured confidence reporting — surfacing when measurements are based on high-quality detections versus limited or partial observations. Research in progress.
Youth Safety & Responsible AI
AI for young athletes should be designed differently.
Young athletes are not simply smaller versions of adult athletes. Age, developmental stage, coaching context, privacy, and how feedback is communicated can all affect whether AI is useful.
- What safeguards should exist for video of minors?
Research pipeline
From video to responsible feedback
Every stage is a distinct research problem. Quality and confidence are tracked through each step.
"Can't measure" is a valid, honest result — never a guessed number.
Research domain
Starting with cricket.
Cricket provides a rich environment — body movement, bat motion, ball motion, timing, repeated actions, and multiple camera viewpoints.
Current work: cricket batting and bowling. Intended to extend to baseball and softball.
Research publications are in development and will be announced following formal review.
Scientific Advisor
Prof. Mubarak Shah
Scientific Advisor — Computer Vision
UCF Center for Research in Computer Vision
Distinguished Professor of Computer Science at UCF, founding director of CRCV. Fellow of IEEE, AAAI, IAPR, and ACM. Research in computer vision, activity recognition, pose estimation, and video understanding.
UCF does not sponsor or endorse Gabriella Systems.
Research collaboration
Work with us
We welcome conversations with researchers, coaches, sports scientists, and organisations interested in computer vision, movement intelligence, trustworthy AI, and responsible technology for sport.
Discuss Research Collaboration