Yield estimation in citrus orchard using an UAV and computer vision

Estimating crop yield potential is important information for both farmers and agricultural cooperatives to be able to sell their products. This yield prediction is the key to predict the volume of stock necessary at the supermarkets and to organize harvesting operations. In many cases, visual estimates of yield are done but this is not accurate. The aim is to build an accurate, fast and reliable fruit detection system based on computer vision using the OpenCV library, for fruit yield estimation. The algorithm that allows the detection was developed and tested on 19 orange trees. Orange yield estimation and actual mass of the fruit per tree was compared. The errors showed very promising values, and therefore, a great potential of the algorithm is foreseen for the citrus yield estimation and probably of other fruits.