DG-Net is a computer vision research system that performs person re-identification using joint discriminative and generative learning. It employs a deep neural network architecture to learn discriminative embeddings for matching person images across different camera views while simultaneously generating synthetic images for data augmentation. The system is implemented in PyTorch and achieves state-of-the-art results on standard re-identification benchmarks including Market-1501, DukeMTMC-ReID, and MSMT17.