A Systematic Review of Deep Learning Microalgae Classification and Detection

• Published in: IEEE access Vol. 11; p. 1
• Main Authors: Madkour, Dina M., Shapiai, Mohd Ibrahim, Mohamad, Shaza Eva, Aly, Hesham H., Ismail, Zool Hilmi, Ibrahim, Mohd Zamri
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.01.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algae; Algae classification; Algae detection; Algorithms; Aquaculture; Classification; Computer vision; Datasets; deep architecture; Deep learning; deep network; Feature extraction; Image classification; Literature reviews; Machine learning; Machine learning algorithms; microalgae; Production; systematic literature review; Water quality
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3280410

Algae represent the majority of the diversity on Earth and are a large group of organisms that have photosynthetic properties that are important to life. The species of algae are estimated to be more than 1 million, they play an important role in many fields such as agriculture, industry, food, medicine. It is important to determine the type of algae to determine if it is harmful or useful, and to indicate the health of the ecosystem, water quality, health, and safety risks. The conventional process of classifying algae is difficult, tedious, and time-consuming. Recently various computer vision techniques have been used to classify algae to overcome challenges and automate the process of classification. This paper presents a review of research done on image classification for microorganism algae using machine learning and deep learning techniques. The paper focuses on three important research questions to highlight the challenges of classifying microalgae. A systematic literature review or SLR has been conducted to determine how deep learning and machine learning have improved and enhanced automatic microalgae classification rather than manual classification. 51 articles have been included from well-known databases. The outcome of this SLR is beneficial due to the detailed analysis and comprehensive overview of the algorithms and the architectures and information about the dataset used in each included article. The future work focuses on getting a large dataset with high resolution, trying different methods to manage imbalance problems, giving more attention to the fusion of deep learning techniques and traditional machine learning techniques.