name

BioHue - colorimetric readout app for a wound-dressing study

synopsis

context

Research group at BITS Hyderabad

period

2025 to 2026

stack

Python, FastAPI, OpenCV, Next.js, MongoDB

description

The readout app for a colorimetric wound-dressing study, built for the research group running it. Photograph a dressing, and the app extracts the sample region, measures its colour and classifies it against the thresholds calibrated for that substrate, so the reading does not depend on who is looking at it.

highlights

• Isolates the region of interest by thresholding saturation in HSV, refining the mask with morphological opening and closing, then taking the largest contour.
• Detects glare by brightness and repaints it from the average of the surrounding unaffected pixels, so a reflection cannot skew the reading.
• Crops that region to a circle, so every sample is measured over the same shape however the photo was framed.
• Applies a per-substrate formula to the separated RGB channels and classifies the result against calibrated thresholds.

links

• source
• the study this supported