Smart-pH: Universal pH Estimation via Artificial Neural Networks Based on Multi-Source Colorimetric Analysis.
pH - ANN model demonstrated excellent predictive performance, achieving an R² > 0.997 for both training and validation, with very low error metrics (MSE = 0.10, MAE = 0.20).

>> pH paper types:
1- Type A: Universalindikator pH 1-14 (MERCK)
2- Type B: Indicator Paper pH 0-14 (UNI-CHEM)
3- Type C: DF Universal pH 0-14
4- Type E: pH-Fix 0-14 (MACHEREY-NAGEL (MN))
5- Type F: pH [0-14] (DF)
6- Type H: pH [5.5-9.0] (DF)

Short stepwise summary:
Home Screen: User selects from 6 commercial pH types.
Info Display: Shows manufacturer, range, and sample images.
Image Input: User captures or uploads pH strip image.
Bounding Boxes: Adjust boxes to isolate color blocks.
Feature Extraction: Compute RGB (mean, median, mode).
Prediction: ANN model predicts pH value.
Results: Display predicted pH with RGB stats.

Scientific validation and More info : "https://elsenety.github.io/ml.html"