sign-language-recognition
Database
Downloaded from American Sign Language Lexicon Video Dataset (ASLLVD)
Use only few videos bellow for I/O test.
- Liz, Naomi and Tyler's scene1-camera1, scene2-camera1 and scene3-camera1 of
ASL_2008_01_18
Steps
- Split videos into images by FFmpeg.
- Generate label.txt for each videos by using csv2txt.py. The CSV file can be downloaded from the database.
- By using setup.py and label.txt, separate images into each label dir.
- Extract features by using extractor.py. Extracted features are in /dataset/results
- Run lstm.py
Reference
Source code in /reference dir is from aidiary/keras-examples: Kerasのサンプルプログラム.