Deep Learning in Precision Agriculture: A Systematic Review of Methods, Datasets, and Applications

With the growing demand for food and water for the world population. Deep learning is one of the most powerful approaches for smart agriculture, offering the capacity for accurate and fast data manipulation and decision-making. To this end, this study contributes to the body of knowledge as follows. We present a comprehensive review of deep learning methods commonly applied in precision agriculture, categorized into three main categories: discriminative, generative, and hybrid models. Given the data-driven nature of DL, we provided a structured overview and tabulation of available datasets in precision agriculture, by categorizing them according to modality to guide dataset selection as well as benchmarking. Following, we taxonomize and review the potential of deep learning at different agricultural applications, ranging from early monitoring to post-harvest analysis. Finally, our research offers a visionary roadmap for addressing open challenges in precision agriculture using deep learning, giving insights into new research directions and future perspectives in this research era.