Machine learning occurs throughout the entire growth and collection cycle. It begins with the sowing of seeds planted in the ground, from the preparation of the soil, from the curve of the seeds and from the measurement of irrigation and ends when the robots harvest the crop identifying its maturity thanks to the use of computerized vision.The yield forecast represents one of the most important topics in the field of precision agriculture as it defines the mapping and estimation of the state of the plantations and the management of the crops. State-of-the-art approaches incorporate artificial vision technologies to provide moving data and a complete multidimensional analysis of crops, weather and economic conditions to maximize the yield for farmers and the population. The careful detection and classification of the quality characteristics of the crops can increase the productivity of the soil and reduce waste. Compared with human experts, machines can use seemingly insignificant data and interconnections to reveal new qualities that play a key role in overall crop quality and to detect them. Weeds are one of the most important threats to agricultural production and are difficult to detect and discriminate against from crops. Machine vision and machine learning algorithms can improve the detection and discrimination of weeds at low cost and without environmental problems and side effects. In the future, these technologies will drive robots that will destroy weeds, minimizing the need for herbicides. Deep learning algorithms require decades of data collected in the field to analyze crop performance in various climates and new features developed in the process. Based on these data it is possible to build a probability model that predicts which genes will most likely contribute to a beneficial trait for a plant. Machine learning can provide a more accurate classification of plants with faster results than traditional methods.