Crop detection is essential for many agricultural applications, such as crop management, crop identification etc. The development of deep learning methods and the availability of satellite imagery have made automated crop detection practical and effective. This work provides a thorough examination and analysis of current advancements in crop recognition utilizing deep learning techniques on satellite imagery. Planning the harvest type can give a premise to separating data on crop establishing design, and region and yield assessment. Acquiring enormous crop type planning by field examination is wasteful and costly. Conventional order strategies have low arrangement precision because of the discontinuity and heterogeneity of harvest planting. In any case, the profound learning calculation has major areas of strength for an extraction capacity and can really recognize and order crop types. This study utilizes GF-1 high-goal remote detecting pictures as the information hotspot for the otherworldly component informational collections are built through field inspecting and utilized for preparing and check, joined with essential review information of grain creation utilitarian regions at the plot scale. The outcomes show that the combination of multi-phantom data and vegetation record highlights further develops order precision. The profound learning calculation is better than the AI calculation in both order precision and grouping impact. Our model proposes a division of the yield over the chose region.