AquaFusionNet: Lightweight VisionSensor Fusion Framework for Real-Time Pathogen Detection and Water Quality Anomaly Prediction on Edge Devices

Summary

AquaFusionNet is a lightweight, cross-modal vision-sensor fusion framework designed for real-time pathogen detection and water quality anomaly prediction on edge devices, specifically addressing the limitations of fragmented monitoring in small-scale drinking water systems. It unifies microscopic imaging and physicochemical sensor data via a gated cross-attention mechanism, achieving high accuracy (94.8% mAP, 96.3% anomaly prediction) and robustness at low power (4.8W) in field deployments. The system significantly reduces common failure modes of unimodal detectors under challenging environmental conditions.

Medical Relevance

This technology directly impacts public health by enabling real-time, reliable detection of microbial contamination and water quality anomalies in drinking water, especially in vulnerable low- and middle-income regions. It facilitates prompt intervention to prevent waterborne diseases and improves the safety of decentralized water systems.

AI Health Application

AquaFusionNet is an AI-driven vision-sensor fusion framework that uses machine learning to detect microbial pathogens and predict water quality anomalies in drinking water in real-time. This application aims to prevent waterborne diseases and enhance public health by providing early warning systems for contamination events.

Key Points

• Addresses the challenge of rapidly fluctuating microbial contamination in small-scale drinking water systems in low and middle-income regions, where existing unimodal monitoring is unreliable for real-time decision-making.
• Introduces AquaFusionNet, a lightweight, edge-deployable cross-modal framework that unifies microscopic imaging (organism-level visibility) and physicochemical sensor data (water chemistry changes) into a single model.
• Utilizes a novel gated cross-attention mechanism to learn statistical dependencies between microbial appearance and concurrent sensor dynamics, optimized for low-power hardware.
• Trained on AquaMicro12K, a new dataset of 12,846 annotated micrographs curated for drinking water contexts, addressing the scarcity of publicly accessible microscopic datasets in this domain.
• Achieved high performance in a six-month field deployment across seven facilities in East Java, Indonesia, processing 1.84 million frames with 94.8% mAP@0.5 for contamination detection and 96.3% anomaly prediction accuracy, while operating at a low power consumption of 4.8W on a Jetson Nano.
• Demonstrated superior accuracy compared to representative lightweight unimodal detectors at comparable or lower power, with cross-modal coupling significantly reducing failure modes under conditions like fouling, turbidity spikes, and inconsistent illumination.
• All models, data, and hardware designs are released openly to facilitate replication and adaptation in decentralized water safety infrastructures globally.

Clinical Impact

The real-time, accurate, and robust detection capabilities of AquaFusionNet provide immediate actionable intelligence for public health authorities and water operators. This allows for rapid identification of microbial contamination and water quality issues, enabling prompt interventions to prevent outbreaks of waterborne diseases, particularly in resource-limited settings. Improved water safety monitoring directly translates to better community health outcomes and reduced disease burden.

Limitations

Not explicitly mentioned in the abstract regarding AquaFusionNet's own inherent limitations. However, the paper highlights how AquaFusionNet overcomes the limitations (e.g., failure modes under fouling, turbidity) of prior unimodal detection systems.

Future Directions

The open release of all models, data, and hardware designs facilitates replication and adaptation of AquaFusionNet in diverse decentralized water safety infrastructures globally. This implies future work in broader deployment, adaptation to different regional contexts, and potential integration into larger smart water management systems.

Medical Domains

Keywords

Abstract

Evidence from many low and middle income regions shows that microbial contamination in small scale drinking water systems often fluctuates rapidly, yet existing monitoring tools capture only fragments of this behaviour. Microscopic imaging provides organism level visibility, whereas physicochemical sensors reveal shortterm changes in water chemistry; in practice, operators must interpret these streams separately, making realtime decision-making unreliable. This study introduces AquaFusionNet, a lightweight cross-modal framework that unifies both information sources inside a single edge deployable model. Unlike prior work that treats microscopic detection and water quality prediction as independent tasks, AquaFusionNet learns the statistical dependencies between microbial appearance and concurrent sensor dynamics through a gated crossattention mechanism designed specifically for lowpower hardware. The framework is trained on AquaMicro12K, a new dataset comprising 12,846 annotated 1000 micrographs curated for drinking water contexts, an area where publicly accessible microscopic datasets are scarce. Deployed for six months across seven facilities in East Java, Indonesia, the system processed 1.84 million frames and consistently detected contamination events with 94.8% mAP@0.5 and 96.3% anomaly prediction accuracy, while operating at 4.8 W on a Jetson Nano. Comparative experiments against representative lightweight detectors show that AquaFusionNet provides higher accuracy at comparable or lower power, and field results indicate that cross-modal coupling reduces common failure modes of unimodal detectors, particularly under fouling, turbidity spikes, and inconsistent illumination. All models, data, and hardware designs are released openly to facilitate replication and adaptation in decentralized water safety infrastructures.

Comments

9Pages, 3 figure, Politeknik Negeri Banyuwangi