Skip to content

Latest commit

 

History

History
22 lines (13 loc) · 669 Bytes

README.md

File metadata and controls

22 lines (13 loc) · 669 Bytes

speech-animation

implementing "A deep learning approach for generalized speech animation" paper using Julia and Knet...

refrence: https://dl.acm.org/citation.cfm?id=3073699

the project is implemented on GRID dataset available at http://spandh.dcs.shef.ac.uk/gridcorpus/

Data collection

  • Used OpenCV and Dlib to extract landmark points on lower face

  • Filter out position, scale and rotational effect (shape alignment)

    • General Procrustes Analysis
  • shape alignment:

  • Shape model

    • apply PCA on shape data
  • training data preperation (sliding widow)