Media
2 items
Description
On-device VLM / OCR in the browser. Images stay on-device; optional Google Drive text backup. VLM Camera runs vision-language models and OCR in your Chrome browser. It captures a camera frame on a schedule you choose, analyzes it on your computer, and can save a text result. Camera images are not stored locally or uploaded. What it does • On-device analysis with WebGPU (CPU/WASM fallback is slower) • Optional encrypted text history on the device • Optional backup of text responses to your own Google Drive (plaintext by default; encryption available) • Optional phone notifications when a condition you define is met (requires yourcut.net device registration) • Community templates for model + instruction sharing (not a one-click finished setup) What it does not do • It does not save or send camera images • It is not a guaranteed always-on or “zero false positive” system • Sleep, quitting Chrome, or disabling the extension stops scheduled analysis • If another app or site is using the camera, that cycle is skipped Requirements • Recent Chrome or Chromium with WebGPU recommended • A GPU with 6 GB VRAM or more is the practical target; integrated GPUs can work with smaller models and longer intervals • First model download can be hundreds of MB to about 2 GB This extension is free and provided as-is. Optional cloud backup and notifications are a separate yourcut.net web service. Accuracy depends on your prompts, model, and scene; you will need to tune settings for your use case. Privacy: https://yourcut.net/ext-apps/vlm-camera/privacy Terms: https://yourcut.net/ext-apps/vlm-camera/terms