All competitions
- Hackathon
- 6 - 11 Jul 2026
GemmaClip
AMD Developer Hackathon: ACT II, team KMUTT Ma Laew
Google DeepMind Gemma Prize, Track 2
Evidence-first video captioning from team KMUTT Ma Laew, where I worked as AI Engineer.
Key numbers
- representative frames
- 6
- temporal anchors
- 4
- high-change moments
- 2
Highlights
- RoleAI Engineer on team KMUTT Ma Laew.
- FramesSix representative frames: four temporal anchors and two high-change moments.
- AudioBounded audio context alongside the visual evidence.
- TraceabilityAn inspectable flow from selected evidence to each generated caption.
- ResultWon the Google DeepMind Gemma Prize, Track 2: “Best Use of Gemma in Video Captioning.”
