Contactless human hand gesture recognition has received significant attention in the preceding decade. This paper proposes a novel classification approach utilizing an advanced 77-GHz multiple-input-multiple-output (MIMO) frequency modulated continuous wave (FMCW) radar. The pre-processed range-Doppler images (RDIs) and range-angle images (RAIs) of this radar are fed into a dual-stream artificial neural network comprised of 2D convolutional neural network-gated recurrent units (2D CNN-GRU) for human hand gesture classification. According to the conducted experiments, the average accuracy of the proposed classification model with 8-fold cross-validation achieves 92.50%.