sign-lanuage-datasets
sign-lanuage-datasets copied to clipboard
YAYAYru
Handshape datasets
Alexey, I'm the author of https://facundoq.github.io/unlp/sign_language_datasets/index.html.
We are currently developing https://github.com/midusi/handshape_datasets, a library to automate downloading and reading handshape datasets.
We are also plan on creating a similar library for sign language video data (https://github.com/midusi/sign_language_datasets).
If you are interested in these topics we are open to collaborations.
Facundoq, sorry, I didn’t answer you for a long time. Thank you for the issus. Yes, I khow the table of Sign Language dataset is your idea. I already wrote the references from your table in the github site. While I took your table to rewrite the table and study with the purpose of our creation phonological dataset of Russian Sign language (RSL) for machine learning using only a web camera or a mobile camera.
The phonological dataset of RSL is five parameters of SL in the dataset, including handshape, hand orientation, location, movement and non-manual parameter. I’ll also think about developing the correct dataset structure, first I create the dataset structure of only images on the Russian language library, including handshape, hand orientation and location, then create the video dataset. I would like to know your idea about my thought?
I read and understood your very cool software development from the link to automatically read the handshape dataset. I’ll also study your code to join your commit if I develop hand configurations for our dataset after creating our dataset. And I'm also interested the library for sign language video data.
I am ready to cooperate with you, I'm happy for you:))) You need my urgent answer, write me [email protected]
Hey Alex, first of all, don't get me wrong, there's no problem in taking the data from the table, it's all CC.
I was waiting to reply to you because we've just release our first version
of handshape-datasets on PyPi:
https://pypi.org/project/handshape-datasets/
You can get it just with pip install handshape-datasets. We are currently
supporting only 13 datasets but would like to add support for the
additional ones you listed. Also if you'd be interested we could add a
separate datasets.md file to the repo with the table you improved here.
Regarding the dataset approach, I also think that's the correct approach, but as I understand for many sign languages it's really hard. For example here in Argentina there is no consensus for many signs, and no official database that describes them. Therefore, in those cases a pure video approach may be more appropriate. However, if Russian Sign Language does has a standard sign dictionary that's great because you can use that metadata to improve the classifiers.
On Sat, May 2, 2020 at 1:47 AM Alexey Prikhodko [email protected] wrote:
acundoq, sorry, I didn’t answer you for a long time. Thank you for the issus. Yes, I khow the table of Sign Language dataset is your idea. I already wrote the references from your table in the github site. While I took your table to rewrite the table and study with the purpose of our creation phonological dataset of Russian Sign language (RSL) for machine learning using only a web camera or a mobile camera.
The phonological dataset of RSL is five parameters of SL in the dataset, including handshape, hand orientation, location, movement and non-manual parameter. I’ll also think about developing the correct dataset structure, first I create the dataset structure of only images on the Russian language library, including handshape, hand orientation and location, then create the video dataset. I would like to know your idea about my thought?
I read and understood your very cool software development from the link to automatically read the handshape dataset. I’ll also study your code to join your commit if I develop hand configurations for our dataset after creating our dataset. And I'm also interested the library for sign language video data.
I am ready to cooperate with you, I'm happy for you:))) You need my urgent answer, write me [email protected]
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/AlexeYAYAY/sign-lanuage-datasets/issues/2#issuecomment-622668503, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAMA6B5TN4NS6TASXTO2MXLRPOQVHANCNFSM4MHBSJRA .