Defoliation by leaf-cutting ants in commercial forest plantations is one of the leading causes of biomass and productivity losses affecting all of Brazilian industrial forest. Thus, the development of monitoring tools that allow extracting the information below the surface in large areas, such as synthetic aperture radar (SAR) systems, is crucial. This work presents a method for ant nest size estimation in industrial forest based on SAR images. A field study is carried out using a drone-borne SAR system to survey a commercial eucalyptus forest by using a helical flight pattern and P band transmitting frequency and finally generating a ground tomography. A convolutional neural network (CNN) is employed for the ant nests size estimation from the tomograms. A mean error of 5 % and 21 % was achieved for a training a validation dataset, respectively.