We address the challenging problem of continuous sign language recognition (CSLR) from RGB videos, proposing a novel deep-learning framework that employs spatio-temporal graph convolutional networks (ST-GCNs), which operate ...