Computational models in Precision Fruit Growing: reviewing the impact of temporal variability on perennial crop yield assessment

Abstract

Abstract Early yield information of perennial crops is crucial for growers and the industry, which allows cost reduction and benefits crop planning. However, the yield assessment of perennial crops by computational models can be challenging due to diverse aspects of interannual variability that act on the crops. This review aimed to investigate and analyze the literature on yield estimation and forecasting modeling of perennial cropping systems. We reviewed 49 articles and categorized them according to their yield assessment strategy, modeling class used, and input variable characteristics. The strategies of yield assessment were discussed in the context of their principal improvement challenges. According to our investigation, image processing and deep learning models are emerging techniques for yield estimation. On the other hand, machine learning algorithms, such as Artificial Neural Networks and Decision Trees, were applied to yield forecasting with reasonable time in advance of harvest. Emphasis is placed on the lack of representative long-term datasets for developing computational models, which can lead to accurate early yield forecasting of perennial crops.