Spatial-Temporal Graph Convolutional Networks for Sign Language Recognition

  • 2019-01-31 01:25:47
  • Cleison Correia de Amorim, David MacĂȘdo, Cleber Zanchettin
  • 46


The recognition of sign language is a challenging task with an important rolein society to facilitate the communication of deaf persons. We propose a newapproach of Spatial-Temporal Graph Convolutional Network to sign languagerecognition based on the human skeletal movements. The method uses graphs tocapture the signs dynamics in two dimensions, spatial and temporal, consideringthe complex aspects of the language. Additionally, we present a new dataset ofhuman skeletons for sign language based on ASLLVD to contribute to futurerelated studies.


Introduction (beta)



Conclusion (beta)