Here, we report the dynamic fracture toughness as well as the cohesive parameters of a bicontinuously nanostructured copolymer, polyurea, under an extremely high crack-tip loading rate, from a deep-learning analysis of a dynamic big-data-generating experiment. We first invented a novel Dynamic Line-Image Shearing Interferometer (DL-ISI), which can generate the displacement-gradient - time profiles along a line on a sample's back surface projectively covering the crack initiation and growth process in a single plate impact experiment. Then, we proposed a convolutional neural network (CNN) based deep-learning framework that can inversely determine the accurate cohesive parameters from DL-ISI fringe images. Plate-impact experiments on a polyurea sample with a mid-plane crack have been performed, and the generated DL-ISI fringe image has been inpainted by a Conditional Generative Adversarial Networks (cGAN). For the first time, the dynamic cohesive parameters of polyurea have been successfully obtained by the pre-trained CNN architecture with the computational dataset, which is consistent with the correlation method and the linear fracture mechanics estimation. Apparent dynamic toughening is found in polyurea, where the cohesive strength is found to be nearly three times higher than the spall strength under the symmetric impact with the same impact speed. These experimental results fill the gap in the current understanding of copolymer's cooperative-failure strength under extreme local loading conditions near the crack tip. This experiment also demonstrates the advantages of big-data-generating experiments, which combine innovative high-throughput experimental techniques with state-of-the-art machine learning algorithms.