Smart or precision viticulture (PV) relies on technology-driven methods towards sustainable productivity. With recent advancements in artificial intelligence (AI), particularly to support computer vision tasks, PV has gained enhanced capabilities, allowing computational entities to complement human observations and insights, thus promoting cutting-edge decision support. Focusing on such context, this paper reviews and synthesizes recent advances in AI-powered grape cluster detection in modern viticulture, aiming at objective and scalable solutions to optimize crop-related processes. Primarily, some of the fundamental concepts of image analysis are surveyed, describing steps involved in image capture, both in 2D and 3D, and pre-processing to advanced feature engineering. Also, AI methodologies are discussed in detail, ranging from classical machine learning algorithms to more sophisticated and reliable deep learning (DL) methods, including convolutional neural networks (CNNs) to process data with spatial complexity. Additionally, contributions of AI for smart viticulture-related business processes are identified under relevant perspectives such as integrability with other technologies supporting distinct operational dimensions (e.g., enterprise resource planning, supply-chain management, robotics), resource efficiency, cost savings, scalability, and considerations regarding return on investment (ROI). Lastly, existing technical issues, such as environmental variability and challenging data needs, as well as the general industry-adoption obstacles, are covered along with a summary of future research directions that include multimodal data fusion and more generalizable AI models for a more sustainable and efficient globalized viticulture.
Chojka et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: