From Raw Video to Real Physics: The Google Cloud AI Breakdown

Google Cloud · Intermediate ·🧠 Large Language Models ·9h ago
Watch Jason Davenport and Shaun White demonstrate how Google Cloud AI breaks down a split-second snowboard trick. By leveraging models built with Google DeepMind and Gemini, they analyze a complex "Switch Double Cork 1440" frame-by-frame without the need for physical sensors, instantly translating 2D video into a 3D pose. By tracking intricate metrics like flight dynamics, rotational velocity, and tuck compression with stunning skeletal and ribbon visual overlays, this showcases how Google Cloud effortlessly translates complex physical movements into engaging, understandable insights for fans.
Watch on YouTube ↗ (saves to browser)
Sign in to unlock AI tutor explanation · ⚡30

Related AI Lessons

⚡
GPT-5.5, Opus 4.7, DeepSeek V4: Frontier AI
Learn about the latest frontier AI models, including GPT-5.5, Opus 4.7, and DeepSeek V4, and how they are advancing the field of AI
Dev.to AI
 ⚡
Will AI Models Become Portable Like USB Drives?
Learn how portable AI models can revolutionize machine learning, making it possible to run inference on any device without relying on cloud services
Dev.to AI
 ⚡
Building SuperLex Skills: The Open-Source Legal Engineering Framework for AI Agents
Learn to build SuperLex skills, an open-source framework for legal engineering in AI agents, to improve their performance in the legal domain
Dev.to AI
 ⚡
How to Deploy Mixtral 8x7B MoE on a $12/Month DigitalOcean Droplet: Cost-Effective Mixture of Experts Inference
Deploy Mixtral 8x7B MoE on a $12/month DigitalOcean Droplet for cost-effective mixture of experts inference, saving on API costs
Dev.to AI
Up next
5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
Dave Ebbelaar (LLM Eng)
Watch →