Real-Time Pond Water Assessment via Embedded Deep Learning and Visual Data Acquisition: A Practical Monitoring Approach for Aquaculture

Summary

We describe a protocol for real-time water quality classification using underwater image acquisition, a Raspberry Pi-based controller network, and an Artificial Neural Network (ANN). The method details system setup, data collection, preprocessing, model training, and deployment. This water-quality assessment system is particularly suitable for Koi breeders, helping ensure optimal conditions for healthy fish.

Abstract

Maintaining optimal water quality is essential in aquaculture, particularly for ornamental species such as Kois, where color clarity and environmental balance are directly linked to health and aesthetics. This article presents a complete protocol for designing, deploying, and operating a real-time water quality monitoring system based on underwater image classification using embedded deep learning. The proposed system integrates a low-cost underwater camera with a Raspberry Pi-based controller network in an autonomous water surface robot, enabling the capture, transmission, and classification of aquatic images through a lightweight neural network model. Five water condition categories -- ranging from clear to turbid -- are visually distinguished, with classification accuracy exceeding 99% on a custom dataset. The system is configured to run autonomously, offering continuous assessment and visual alerts for deteriorating conditions. This visual protocol covers device assembly, system calibration, model deployment, and real-time monitoring demonstrations. The method supports sustainable aquaculture practices by enabling non-invasive, continuous evaluation of aquatic environments using affordable, scalable technologies. The approach can be adapted to other aquaculture or environmental monitoring contexts with minimal modification.

Introduction

Koi fish (Cyprinus carpio var. koi) are ornamental varieties of the common carp, prized for their vibrant colors, distinctive patterns, elegant body shape, and graceful swimming. Their beauty and fluid movements make them a favorite choice for garden ponds and decorative water features worldwide.

Maintaining good water quality in ponds not only supports aquatic life but also contributes to reducing carbon emissions, thereby benefiting the environment^1,2. Maintaining water quality by following established standards for parameters such as pH, temperature, dissolved oxygen, sa....

Protocol

NOTE: The source codes used in this study are available at the following GitHub repository: https://github.com/prawit-chu/Real-Time-Pond-Water-Assessment-via-Embedded-Deep-Learning-and-Visual-Data-Acquisition

1. Equipment Setup

1. Set up the Raspberry Pi-based controller network (the robot and the action camera).
2. Set up a server running the Ubuntu v24.04 LTS operating system.

2. Action Camera Setup

1. Set up the default IP address (10.5.5.9) and network mask (255.0.0.0).
   NOTE: Table 1 shows the assigned IP addresses for con....

Results

The dataset was divided into independent training, validation, and test subsets. Model performance was monitored on the validation set during training, and the final evaluation was conducted using a held-out test dataset containing previously unseen images. Consistent performance across validation and test sets indicates stable generalization.

For real-time water quality classification, simplified CNN models and pre-trained networks (Inception V3, Xception, and ResNet50) were initially tested,.......

Discussion

A critical step in this protocol is Raspberry Pi-based controller network setting. Incorrect settings result in failure to communicate between the server and the action camera. Before testing the provided programs, communication between the server and the action camera should be verified by PING command. While the proposed model operates in real time on embedded hardware, quantitative measurements of inference time and energy consumption were not evaluated and are beyond the scope of this study.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
GoPro Hero 11 Black	GoPro		for taking photos and record videos
GoPro Hero 12 Black	GoPro		for taking photos and record videos
GoPro Hero 13 Black	GoPro		for taking photos and record videos
Raspbery PI	Raspberry PI	Zero 2W	Controller
Wireless Router	ZTE	HG8045x5-12
Ethernet adaptors	Tp-link	UE300
Server	Dell	T30
Battery Pack 12 VDC	-	-	Power Supply
DC pump	Thai Water	AW500S	robot movement
5V step Down	-	-	Step down from 12V to 5V
8-channel Relay	-	-	Controlling pumps
GoPro Quick application	GoPro	GoPro	for testing GoPro Wi-Fi connection