In any crop, the total production is definitely the most important criterion that shows the agricultural business’s success in revenue in relation to the expenses and simultaneously indicates if the producers achieved their targeted goals. Tree crops put some challenges in counting the fruits and as a consequence making a yield prediction or estimation. However, the development of AI and IoT with the embedded devices and small but powerful sensors in agriculture has given the potential to the producers to make use of these technologies and control their crop production. Computer vision contributes with the combined utilization of the cameras, software and the aforementioned tools to give precise and confident decisions. The output of the measurements are depicted on a map (Figure 1), known as yield map which can be configured according to the specific demand each time, in order to transfer the appropriate information.
Modern yield estimation systems with the form of a device (Figure 2), have strong capabilities in counting fruits, as can be easily plugged in the tractor, during any cultivation job without disrupting the user or bothering any of the tractor functionalities, in regards to the electronics assembly or operation. The fruit counting is associated with the specific coordinates taken from the GPS. The yield map depicts the distribution and the variability of the yield across the field and in this way guides the producer with the upcoming cultivation steps. With computer vision, another comparative advantage emerges which is the size of the fruits and the color for purposes like choosing the most suitable harvest time and in advanced systems even the texture can also be measured from the cameras and analyzed from the algorithms.
Putting all the above mentioned together, we understand that computer vision contributes to the yield estimation at both the cultivation and harvest and post harvest stages of the crop and brings direct and indirect benefits. It helps the producer with the decision making process and predicts the final yield months ago for better programming of the harvest and the sales contracts. It also saves money from unnecessary use of water and fertilizers. Moreover, it can reveal trees or zones of the field with lack of nutrients or unseen plant enemies in the soil and contributes significantly to the environmental protection and sustainability. Last but not least, it paves the way for other developing technologies and relevant activities like disease detection and precision spraying.