A collaboration between Faculty of Pharmacy and IST, both from the University of Lisbon.
Given a photograph of an array of COVID-19 test samples, ColorCovid automatically detects each sample well, extracts its color characteristics, and classifies which tests are positive.
- Camera control — connect any camera to the computer and capture snapshots directly from the GUI
- Sample detection — automatically locates and uniquely indexes each well in the plate, handling variable array layouts, well shapes, and lighting conditions
- Color feature extraction — computes HSV and RGB channel averages per sample and exports them to CSV
- Classification — classifies each sample as positive or negative based on its color features
pip install opencv-python numpy matplotlib scikit-image scipy pillow imutils easyguiThe image processing pipeline detects and segments individual wells through five stages:
Processing steps
Original image |
Step 1 Detect the background |
Step 2 High-saturation threshold to broadly locate the wells |
Step 3 Remove the background |
Step 4 Euclidean distance mask |
Step 5 Watershed algorithm — final sample markers |
Detection works across a range of plate formats and lighting conditions:
Detection examples
| Original | Detected samples |
|---|---|
 |
 |
 |
 |
 |
 |
 |
 |
Each detected sample is uniquely indexed on the plate:
Color features (H, S, V, R, G, B averages) for all samples are exported to a CSV file:
A built-in visualization tool lets you inspect each sample individually and browse all color data at a glance:
Samples can be shown with their surrounding border or cropped tightly to the region of interest:
| With border | Cropped to ROI |
|---|---|
 |
 |
The list can be sorted by any parameter — color channel, test result, or sample index:
| By RGB value (red channel) | By test result | By sample index |
|---|---|---|
 |
 |
 |
Rafael Correia — LinkedIn