A Multimodal German Dataset for Automatic Lip Reading Systems and Transfer Learning

LREC - 13th Conference on Language Resources and Evaluation Palais du Pharo - Marseille, June 20-25, 2022, doi:- https://arxiv.org/abs/2202.13403 - May 2022.
Associated documents : GLipsPaper_.pdf [3.4Mo]   http://dx.doi.org/- https://arxiv.org/abs/2202.13403
Large datasets as required for deep learning of lip reading do not exist in many languages. In this paper we present the dataset GLips (German Lips) consisting of 250,000 publicly available videos of the faces of speakers of the Hessian Parliament, which was processed for word-level lip reading using an automatic pipeline. The format is similar to that of the English language LRW (Lip Reading in the Wild) dataset, with each video encoding one word of interest in a context of 1.16 seconds duration, which yields compatibility for studying transfer learning between both datasets. By training a deep neural network, we investigate whether lip reading has language-independent features, so that datasets of different languages can be used to improve lip reading models. We demonstrate learning from scratch and show that transfer learning from LRW to GLips and vice versa improves learning speed and performance, in particular for the validation set.

 

@InProceedings{SWQSW22a,
  author       = "Schwiebert, Gerald and Weber, Cornelius and Qu, Leyuan and Siqueira, Henrique and Wermter, Stefan",
  title        = "A Multimodal German Dataset for Automatic Lip Reading Systems and Transfer Learning",
  booktitle    = "LREC - 13th Conference on Language Resources and Evaluation Palais du Pharo - Marseille, June 20-25, 2022",
  month        = "May",
  year         = "2022",
  note         = "https://arxiv.org/abs/2202.13403",
  doi          = "- https://arxiv.org/abs/2202.13403",
  url          = "https://www2.informatik.uni-hamburg.de/wtm/publications/2022/SWQSW22a/GLipsPaper_.pdf"
}

» Gerald Schwiebert
» Cornelius Weber
» Leyuan Qu
» Henrique Siqueira
» Stefan Wermter