"V9 team on AI Infra Summit Hackathon Hackathon"

Team Idea

Recall — Ctrl+F for the physical world. A single camera on a SiMa.ai. Modalix DevKit that remembers everything happening in a room and answers questions out loud: "Where's my red mug?", "Who took the laptop?", "What happened at this desk while I was at lunch?" Instance segmentation tracks every object, pose estimation attributes pickups to the person whose hand was on it, and a vision-language model names things and answers questions. All running on one sub-10W chip with zero cloud. Unplug the network and it still works. That's the point. Nobody will stream 24/7 video of their home, lab, or hospital ward to a server, so visual memory can only exist on-device. We have the DevKit, the SDK running, and the perception pipeline underway.

Team Members

NarasingaMoorthy VeilukanthaPerumal

Vnmoorthy

Submission

Recall: Ctrl+F for the physical world

A camera on a SiMa.ai Modalix chip that remembers everything in a room and answers out loud: "Where's my red mug?" "Who took the laptop?" Segmentation, pose, tracking and a vision-language model on one sub-10 W chip, with no cloud at all.