@article{Hongboonmee_Oonkham_2022, title={Classification of Water Quality Level using Deep Learning on Android Application}, volume={12}, url={https://ph02.tci-thaijo.org/index.php/JIST/article/view/246672}, DOI={10.14456/jist.2022.12}, abstractNote={<p>Water is crucial for all lives, yet recently its quality is increasingly affected by waste and pollution. As water quality inspection is important for ecological and water resource management, this work therefore proposed a water quality inspection technique based on deep learning to automatically classify water quality on an Android application. The process started from collection of 480 images which were divided into four grades of water: A, B, C and D. The images were used for training a model for water quality classification, and three algorithms were compared: MobileNet 0.5 MobileNet 0.75, and MobileNet 1.0. Model assessment was done using accuracy, precision and recall. The model was developed into an application, which was then finally tested. The test found that (1) the MobileNet 1.0 was the most suitable model with 100.00% accuracy, precision and recall, and (2) application test showed that it had classification success rate of 82.50%, with good performance. Application was assessed by experts and the result was 4.33 on average, or highly suitable. The test showed that this approach and application could be used for water quality inspection from images with high efficiency and ease of use.</p>}, number={2}, journal={JOURNAL OF INFORMATION SCIENCE AND TECHNOLOGY}, author={Hongboonmee, Nattavadee and Oonkham, Thiraphong}, year={2022}, month={Dec.}, pages={31–41} }