I need a Chrome-based extension that watches a live YouTube stream, lets the viewer draw a bounding box on the video, and—every frame—grabs the characters inside that box, turns them into clean digital text, then instantly runs multiplication or division on the values it finds. The computed result must appear as a subtle but readable overlay positioned on top of the running video, all inside 1 second end-to-end. Here is what has to happen under the hood: • Real-time OCR on the selected area (Tesseract.js, OpenCV.js, or a comparable in-browser engine are fine as long as they keep latency under the one-second cap). • Simple UI to let the viewer draw / adjust the capture region at any time while the stream is playing. • Parsing logic that recognises alphanumeric strings, extracts the relevant numbers, and performs the requested multiplication or division automatically. • Overlay rendering that updates continuously, stays synced with the video, and doesn’t interfere with YouTube controls. Use Canvas, WebGL or CSS layers—whatever gives smooth 60 fps updates. • A clean settings pane so the user can toggle the extension, switch between multiplication and division, and choose text/overlay styling. Future-proofing: I may want to support several capture regions later, so structuring the code to allow additional boxes (even if only one is active for now) will earn extra points. Acceptance criteria 1. Extraction-to-overlay round-trip consistently ≤ 1000 ms on an average i5/8 GB laptop in Chrome. 2. OCR accuracy ≥ 95 % on high-contrast digits/letters sized 14 px or larger. 3. No console errors, no impact on normal YouTube playback controls. 4. Packaged as a standard Chrome extension (manifest v3) plus concise setup/usage guide. Send me a short note on your proposed tech stack, any latency figures you’ve hit before, and an estimated timeline for a working prototype.