DGCNN provides a re-implementation of the Dynamic Graph CNN architecture featuring EdgeConv, a differentiable neural network module designed for processing point clouds. The implementation supports high-level tasks including 3D object classification, semantic segmentation, and part segmentation. Both TensorFlow and PyTorch versions are provided, with the PyTorch implementation used for the original paper’s classification experiments. The model achieves state-of-the-art performance on point-cloud benchmarks and includes evaluations under corruption scenarios using ModelNet-C.