A deep convolutional neural network is employed to simultaneously measure the beam-pointing and phase difference of sub-beams from a single far-field interference fringe for coherent beam combining systems. The amplitudes of sub-beams in the measurement path are modulated in order to prevent measuring mistakes caused by the symmetry of beam-pointing. This method is able to measure beam-pointing and phase difference with an RMS accuracy of about 0.2 μrad and λ/250, respectively, in a two-beam coherent beam combining system.