A rapid detection model and device for pH value in water quality based on ratio fluorescence imaging

ABSTRACT

Summary: Met-GQDs and GSH-AuNCs form a ratiometric fluorescent probe (pH 2-12); a smartphone builds a 2,000-image dataset from 11 pH levels; YOLOv5n and CatBoost give 99.26% pH accuracy; a 3D-printed device and app read pH in 10 s (MAE 0.0440).

pH value is a critical physicochemical parameter in water quality assessment. Deviations from the optimal pH range in aquatic environments such as rivers, lakes, and oceans can exert profound and lasting adverse effects on ecosystem stability and human health. In this study, a pH detection system integrating a ratiometric fluorescence imaging strategy was developed. A ratiometric fluorescent probe based on graphene quantum dots and gold nanoclusters was first constructed. Fluorescence imaging data were subsequently captured using a smartphone camera, followed by data augmentation and expansion to establish a robust ratiometric fluorescence image dataset. A hybrid pH detection algorithm combining YOLOv5n-based object detection and CatBoost regression was proposed and implemented. Furthermore, a portable device for on-site water pH measurement was assembled. The proposed model achieved a detection accuracy of 99.26%, demonstrating excellent performance. The practical applicability of the device was validated through pH measurements of real water samples, with comparative analyses against conventional methods confirming its superior detection capability. Overall, the developed pH sensor offers distinct advantages in terms of portability and cost-effectiveness, rendering it suitable for rapid, on-site pH monitoring in diverse water bodies.