Deep Learning Shape Trajectories for Isolated Word Sign Language Recognition
Sana Fakhfakh L3S Laboratory, El Manar University Tunis, Tunisia This email address is being protected from spambots. You need JavaScript enabled to view it. |
Yousra Ben Jemaa L3S Laboratory, El Manar University Tunis, Tunisia This email address is being protected from spambots. You need JavaScript enabled to view it. |
Abstract: In this paper, we propose an efficient trajectories analysis solution for the recognition of Isolated Word Sign Language (IWSL). The key technique innovation in this work is the shape trajectories analysis based on the deep learning method and achieved impressive results on different IWSL data sets: German: Rheinisch Westfälische Technische Hochschule(RWTH): RWTH-Boston-50 and RWTH-Boston-104(95.83%), Signer-Independent Continuous Sign Language Recognition for Large Vocabulary Using Subunit Models (SIGNUM: 98.21%) and new Tunisian Sign Language database (TunSigns: 98%).
Keywords: Sign language, isolated word recognition, shape trajectory analysis, deep learning, RWTH-Boston dataset and SIGNUM corpora.
Received January 17, 2020; accepted February 9, 2021