Portfolio Project

Sheet Music Watermark Removal & Upscale

UNet & VDSR Pipeline

Machine Learning Python Computer Vision

STAR Summary

Situation
I needed clean, readable sheet music. Most of what I could find was low-res and watermarked.
Task
Owned the end-to-end build, from implementation through the final deliverable.
Action
  • Trained a UNet model on paired page images for watermark removal.
  • Upscaled low-resolution scans (e.g., ~612x792) to print-friendly resolution (e.g., ~1700x2200) with Very Deep Super-Resolution (VDSR).
  • Wrapped it in a simple GUI so I could run the full pipeline.
Result
  • Typical runs produce cleaner, more readable output in seconds per page (hardware- and batch-dependent).