Unveiling the Potential of Artificial Intelligence in Plant Phenotyping and Disease Detection: A Survey

In recent years, advanced plant phenotyping techniques have gained significant attention in the identification of plant diseases and the assessment of plant characteristics. These techniques involve machine vision, machine learning, and deep learning methods applied to non-invasive and non-destructive imaging. To accurately capture the plant's condition, efficient systems and equipment are required in the field of plant phenotyping. Unlike previous studies, which often focused on individual imaging methods; this work extensively investigates a wide range of imaging methods, such as visible imaging (RGB camera), multi or hyperspectral imaging, thermal imaging, chlorophyll fluorescence imaging (CFIM), hyperspectral imaging, and 3-dimensional imaging. This article provides an overview of these imaging techniques and their applications in plant phenotyping. The paper also presents a comprehensive survey of cutting- edge artificial intelligence for analyzing plant leaves or roots, including segmentation, trait estimation, and classification and using statistical analysis, machine learning, and deep learning algorithms Furthermore, it introduces an image pipeline that combines both analogical and numerical tools to detect visual plant traits and monitor environmental changes. This survey provides valuable insights into the advancements in imaging techniques for plant phenotyping and disease detection. It also encompasses all key aspects of machine learning, deep learning and statistical methods used in plant phenotyping including preprocessing, segmentation, feature extraction, classification and categorizing. Compared to earlier works, this survey provides a more exhaustive exploration of imaging techniques, combined with the latest advancements in AI-driven plant phenotyping, offering critical perspectives into both technological and practical developments in plant disease detection and phenotyping.