A Lip Reading Data Set for the Under-resourced Romanian Language
These data resources are designed to support research in the areas of computer vision and particularly automatic lip-reading algorithms. Two distinct collections are proposed: (i) wild LRRo data is designed for an Internet in-the-wild, ad-hoc scenario, coming with more than 35 different speakers, 1.1k words, a vocabulary of 21 words, and more than 20 hours; (ii) lab LRRo data, addresses a lab controlled scenario for more accurate data, coming with 19 different speakers, 6.4k words, a vocabulary of 48 words, and more than 5 hours. This is the first resource available for Romanian lip reading and would serve as a pioneering foundation for this under-resourced language. Nevertheless, given the fact that word-level models are not strongly language dependent, these resources will also contribute to the general lip-reading task via transfer learning.
Using the dataset:
If you plan to make use of the LRRo dataset, or refer to its results, please acknowledge the work of the authors by citing the following paper:
- A.C. Jitaru, Ș. Abdulamit, B. Ionescu, "LRRo: A Lip Reading Data Set for the Under-resourced Romanian Language", ACM Multimedia Systems - MMSys, June 8-11, Istanbul, Turkey, 2020 (6 pages, download draft PDF, link to ACM).
Acknowledgements:
Andrei Cosmin Jitaru, Șeila Abdulamit, Bogdan Ionescu, University "Politehnica" of Bucharest, Romania.
This dataset was supported by the SPIA-VA Technologies and Innovative Video Systems for Person Re-Identification and Analysis of Dissimulated Behavior project, owner University "Politehnica" of Bucharest, partners UTI Grup, Romanian Ministry of National Defence — Military Equipment and Technologies Research Agency, public beneficiary Protection and Guard Service, funded by UEFISCDI, Solutions Axis, grant 2SOL/2017.