Computer Vision Podcast Summaries
Computer Vision on Yedapo: 2 summarized podcast and YouTube episodes. Each includes key takeaways, core concepts and notable quotes with timestamps.
DeepSeek's Deleted Vision Paper Is Nuts...
bycloud
Jul 8, 2026
DeepSeek's new approach solves the 'reference gap' in multimodal models by allowing AI to 'point' at images using bounding boxes and coordinates. By interleaving visual primitives into its chain-of-thought, the model effectively grounds its reasoning in space rather than relying solely on ambiguous language descriptions, significantly outperforming frontier models in topological and counting tasks.
Key insight: For maze navigation, while most frontier models hover around 50% accuracy, this new architecture hits 66.9% by treating visual reasoning like a 'scratchpad' where the model draws points to track progress.
How Claude Code Created an AI Coach (Controls My Studio)
DesignCourse
Mar 24, 2026
A single Python script generated by Claude Code now manages 21 DMX light fixtures, overhead projectors, and real-time computer vision to create an interactive pool training coach. This experiment reveals how AI coding tools are enabling complex hardware-software integrations that were previously too intimidating for solo developers to attempt.
Key insight: The system uses an overhead camera to track balls in real-time, projecting live tangent lines and Q-ball deflection paths directly onto the table felt while an AI persona named "Buck Snipes" provides feedback.
