Development of Microalgae Detection via Artificial intelligence

Microalgae, often found in untreated water sources used in water production, pose a significant challenge due to their microscopic size, making it difficult to eliminate them using traditional filtration systems. Identifying and quantifying microalgae contaminants in raw water is time-consuming and labor-intensive. This research aims to develop a system for counting microalgae through automatic image detection using the fast and accurate YOLO (You Only Look Once) algorithm. The system involves training datasets, generating data weights for processing within programs developed in Microsoft Visual Studio, and evaluating the program's accuracy in detecting algae compared to expert analyses. The model achieved an impressive 93.54% accuracy in detecting all 11 types of algae, ensuring reliable results. Additionally, the system reduced the counting time by 94.44%, with an average image processing time of approximately 0.6 seconds, a significant improvement compared to the manual visual inspection time of 10.8 seconds per grid. The system's high accuracy demonstrates its effectiveness, making it a reliable tool for water quality management.