search

SEACrowd / seacrowd-datahub

A collaborative project to collect datasets in SEA languages, SEA regions, or SEA cultures.
Apache License 2.0
66 stars 57 forks source link

Create dataset loader for FSL-105 #13

Closed SamuelCahyawijaya closed 10 months ago

SamuelCahyawijaya commented 1 year ago

Dataloader name: fsl_105/fsl_105.py DataCatalogue: http://seacrowd.github.io/seacrowd-catalogue/card.html?fsl_105

Dataset fsl_105
Description FSL-105 is a video dataset for 105 different Filipino Sign Language (FSL) signs. Each sign is categorized into one of 10 categories and is each represented by approximately 20 four-second video samples. Signs were performed by adult deaf FSL signers on a blank blue background and reviewed by an FSL expert.
Subsets -
Languages psp
Tasks Image-to-Text Generation, Optical Character Recognition
License Creative Commons Attribution 4.0 (cc-by-4.0)
Homepage https://data.mendeley.com/datasets/48y2y99mb9/2
HF URL -
Paper URL https://ieeexplore.ieee.org/document/10109284
IvanHalimP commented 1 year ago

self-assign

IvanHalimP commented 1 year ago

I'd like to know how I should implement this one. The data is video->text->category. Which schema/task it belongs to?

IvanHalimP commented 11 months ago

I guess video->text and video->category tasks are not implemented yet here.

sabilmakbar commented 11 months ago

The Video Text has been implemented per this PR. Apologies for super-late responses, @IvanHalimP :(

IvanHalimP commented 11 months ago

Thank you for your support! @sabilmakbar