Pool Ops Vision Console

Upload pool media. Get structured analysis back.

v1 is intentionally narrow: images go to Cloud Vision, videos go to Google Cloud Storage and then through Video Intelligence. This keeps the Cloudflare edge app simple while still giving you a live single-page console for field footage and site photos.

Cloudflare Pages + Worker
Cloud Vision for images
Video Intelligence for videos

Analyze Image

Upload a still image for labels, OCR text, landmarks, objects, and face signals.

Use JPG, PNG, WEBP, GIF, or BMP. iPhone HEIC/HEIF images are not supported by this Cloud Vision route yet.

Analyze Video

Upload a pool walkthrough or job-site video. The browser uploads straight to GCS, then the Worker starts Video Intelligence and polls the operation.

v1 Notes

Large videos are not sent through the Worker body. The Worker creates a resumable GCS upload session and the browser uses that directly.
Video jobs are asynchronous. You get an operation name first, then the UI polls until Google finishes processing.
This version is intentionally API-first and structured. It does not do Gemini summarization, Translation, or TTS.
You do not need to invent prompts. Pick a mode and the app uses built-in pool inspection goals and follow-up guidance automatically.

Result

Pool-aware structured output from the Worker, plus the underlying Google annotations for reference. 

Run an image or video analysis to see the response here.