Determination of maturity and optimum harvest day in vegetables with artificial intelligence

As artificial intelligence (AI) and its applications become increasingly prevalent in daily life, it is being used in various fields, including military, logistics, and biology applications. The use of machine learning, deep learning, and convolutional neural networks-sub algorithms of deep learning-in decision support systems, and their integration into the system, particularly in agricultural applications, has enabled precise and data-driven decision-making. Thus, the use of data-based decision support mechanisms in product management can enable high yields during product harvesting with its innovative approach, reduce human error, and increase food quality and product yield. One of the distinguishing features of smart algorithms is their ability to process data with high accuracy, thereby benefiting vegetable production. The use of convolutional neural networks to detect ripeness levels in vegetables can minimize human error and eliminate the shortcomings of human observation. For all these reasons, timely and accurate ripeness detection is essential for determining the correct harvest time, as it increases the nutritional value and product quality of vegetables. Furthermore, it maximizes both the market value and nutritional value of the products for consumers, minimizing potential nutritional and financial losses that could arise from incorrect determination of the ripening period or incorrect determination of the harvest time. Systems supported by smart algorithms, smart sensors, and image analysis can track product development, accurately determine ripeness and harvest time, detect diseases in a timely and early manner, minimize losses, and play a useful role in optimizing resource management. Accurate determination of harvest timing by producers will increase efficiency in the logistics chain, enabling increased and sustainable food production for the rapidly growing population. For this reason, convolutional neural networks were utilized in this thesis study to determine ripeness and the optimal harvest time in vegetables. After examining intelligent algorithms used in deep learning and decision support mechanisms in the Methods section, the findings section will compare convolutional and traditional neural networks based on simulation studies for tomato fruit, while the Discussion section will compare them with studies conducted in recent years.