ArSL21L - Dataset Ninja

Introduction #

Released 2022-02-11 ·Ganzorig Batnasan, Munkhjargal Gochoo, Munkh Erdene Otgonboldet al.

The authors present their collected and annotated ArSL21L: Arabic Sign Language Letter Dataset consisting of 14202 images of 32 letter signs with various backgrounds collected from 50 people. They benchmarked the ArSL21L dataset on state-of-the-art object detection models, i.e., 4 versions of YOLOv5.

Motivation

For individuals with hearing loss (PwHL), establishing social connections with the mainstream population can be challenging. This often necessitates interactive computer systems capable of understanding sign language. With the growing popularity of Metaverse applications employing augmented reality (AR) and virtual reality (VR), there’s an opportunity to facilitate remote sign language instruction through avatars that mimic gestures, powered by AI (Artificial Intelligence) systems. While numerous methods and datasets exist for English sign language, resources for Arabic sign language are limited. Thus, the authors introduce the Arabic Sign Language Letters Dataset (ArSL21L), comprising 14,202 images depicting 32 letter signs against various backgrounds. These images were collected from 50 individuals and annotated for clarity and consistency.

Arabic sign language.

Expand

Homepage

Research Paper

Kaggle

Summary #

ArSL21L: Arabic Sign Language Letter Dataset is a dataset for an object detection task. It is used in the educational domain.

The dataset consists of 14202 images with 14208 labeled objects belonging to 32 different classes including thaa, al, dha, and other: fa, ghain, ha, khaa, la, meem, saad, seen, sheen, ta, thal, toot, yaa, haa, ain, jeem, ya, aleff, bb, kaaf, gaaf, taa, laam, zay, dhad, ra, nun, waw, and dal.

Images in the ArSL21L dataset have bounding box annotations. All images are labeled (i.e. with annotations). There are 2 splits in the dataset: train (9955 images) and val (4247 images). The dataset was released in 2022 by the United Arab Emirates University.

Explore #

ArSL21L dataset has 14202 images. Click on one of the examples below or open "Explore" tool anytime you need to view dataset images with annotations. This tool has extended visualization capabilities like zoom, translation, objects table, custom filters and more. Hover the mouse over the images to hide or show annotations.

👀

Have a look at 14202 images

View images along with annotations and tags, search and filter by various parameters

Class balance #

There are 32 annotation classes in the dataset. Find the general statistics and balances for every class in the table below. Click any row to preview images that have labels of the selected class. Sort by column to find the most rare or prevalent classes.

Rows 1-10 of 32

Class ㅤ	Images ㅤ	Objects ㅤ	Count on image average	Area on image average
thaa➔ rectangle	451	451	1	14.06%
yaa➔ rectangle	450	450	1	15.95%
toot➔ rectangle	450	450	1	13.18%
thal➔ rectangle	450	450	1	14.94%
ta➔ rectangle	450	450	1	14.16%
sheen➔ rectangle	450	450	1	19.35%
seen➔ rectangle	450	450	1	15.56%
saad➔ rectangle	450	450	1	11.47%
meem➔ rectangle	450	450	1	13.37%
la➔ rectangle	450	450	1	15.07%

Images #

Explore every single image in the dataset with respect to the number of annotations of each class it has. Click a row to preview selected image. Sort by any column to find anomalies and edge cases. Use horizontal scroll if the table has many columns for a large number of classes in the dataset.

Object distribution #

Interactive heatmap chart for every class with object distribution shows how many images are in the dataset with a certain number of objects of a specific class. Users can click cell and see the list of all corresponding images.

Class sizes #

The table below gives various size properties of objects for every class. Click a row to see the image with annotations of the selected class. Sort columns to find classes with the smallest or largest objects or understand the size differences between classes.

Rows 1-10 of 32

Class	Object count	Avg area	Max area	Min area	Min height	Min height	Max height	Max height	Avg height	Avg height	Min width	Min width	Max width	Max width
thaa rectangle	451	14.06%	67.96%	1.75%	69px	16.59%	398px	95.67%	187px	44.96%	37px	8.89%	304px	73.08%
al rectangle	451	21.03%	82.78%	1.85%	68px	16.35%	402px	96.63%	192px	46.24%	47px	11.3%	380px	91.35%
yaa rectangle	450	15.95%	76.37%	1.83%	61px	14.66%	391px	93.99%	153px	36.72%	48px	11.54%	370px	88.94%
toot rectangle	450	13.18%	81.23%	1.62%	66px	15.87%	398px	95.67%	163px	39.16%	37px	8.89%	358px	86.06%
thal rectangle	450	14.94%	71.64%	2.36%	54px	12.98%	388px	93.27%	158px	37.96%	57px	13.7%	415px	99.76%
ta rectangle	450	14.16%	61.43%	1.37%	64px	15.38%	406px	97.6%	187px	45.05%	33px	7.93%	289px	69.47%
sheen rectangle	450	19.35%	83.37%	1.9%	70px	16.83%	408px	98.08%	191px	45.89%	47px	11.3%	376px	90.38%
seen rectangle	450	15.56%	72.41%	1.78%	73px	17.55%	414px	99.52%	187px	44.89%	40px	9.62%	351px	84.38%
saad rectangle	450	11.47%	71.99%	1.02%	45px	10.82%	358px	86.06%	128px	30.75%	36px	8.65%	369px	88.7%
meem rectangle	450	13.36%	75.83%	1.5%	59px	14.18%	391px	93.99%	161px	38.7%	37px	8.89%	391px	93.99%

Spatial Heatmap #

The heatmaps below give the spatial distributions of all objects for every class. These visualizations provide insights into the most probable and rare object locations on the image. It helps analyze objects' placements in a dataset.

Objects #

Table contains all 14208 objects. Click a row to preview an image with annotations, and use search or pagination to navigate. Sort columns to find outliers in the dataset.

Rows 1-10 of 14208

Object ID ㅤ	Class ㅤ	Image name click row to open	Image size height x width	Height ㅤ	Height ㅤ	Width ㅤ	Width ㅤ	Area ㅤ
1➔	nun rectangle	1201_23_M_nun_5.jpg	416 x 416	163px	39.18%	128px	30.77%	12.06%
2➔	seen rectangle	1020_22_M_seen_4.jpg	416 x 416	240px	57.69%	175px	42.07%	24.27%
3➔	ghain rectangle	1201_23_M_ghain_2.jpg	416 x 416	248px	59.62%	276px	66.35%	39.55%
4➔	yaa rectangle	202_18_M_yaa_0.jpg	416 x 416	269px	64.66%	339px	81.49%	52.69%
5➔	jeem rectangle	1056_19_F_jeem_1.jpg	416 x 416	212px	50.96%	174px	41.83%	21.32%
6➔	seen rectangle	1007_29_M_seen_7.jpg	416 x 416	104px	25%	65px	15.62%	3.91%
7➔	dal rectangle	229_20_M_dal_7.jpg	416 x 416	171px	41.11%	164px	39.42%	16.21%
8➔	thal rectangle	202_18_M_thal_0.jpg	416 x 416	298px	71.63%	249px	59.86%	42.88%
9➔	sheen rectangle	202_18_M_sheen_0.jpg	416 x 416	353px	84.86%	250px	60.1%	51%
10➔	toot rectangle	1020_22_M_toot_5.jpg	416 x 416	239px	57.45%	197px	47.36%	27.21%

License #

ArSL21L: Arabic Sign Language Letter Dataset is under CC BY 4.0 license.

Source

Citation #

If you make use of the ArSL21L data, please cite the following reference:

@dataset{ArSL21L,
  author={Ganzorig Batnasan and Munkhjargal Gochoo and Munkh Erdene Otgonbold and Fady Alnajjar and Timothy Shih},
  title={ArSL21L: Arabic Sign Language Letter Dataset},
  year={2022},
  url={https://data.mendeley.com/datasets/f63xhm286w/1}
}

Source

If you are happy with Dataset Ninja and use provided visualizations and tools in your work, please cite us:

@misc{ visualization-tools-for-arsl21l-dataset,
  title = { Visualization Tools for ArSL21L Dataset },
  type = { Computer Vision Tools },
  author = { Dataset Ninja },
  howpublished = { \url{ https://datasetninja.com/arsl21l } },
  url = { https://datasetninja.com/arsl21l },
  journal = { Dataset Ninja },
  publisher = { Dataset Ninja },
  year = { 2024 },
  month = { nov },
  note = { visited on 2024-11-21 },
}

Download #

Dataset ArSL21L can be downloaded in Supervisely format:

As an alternative, it can be downloaded with dataset-tools package:

pip install --upgrade dataset-tools

… using following python code:

import dataset_tools as dtools

dtools.download(dataset='ArSL21L', dst_dir='~/dataset-ninja/')

Make sure not to overlook the python code example available on the Supervisely Developer Portal. It will give you a clear idea of how to effortlessly work with the downloaded dataset.

The data in original format can be downloaded here.

. . .

Disclaimer #

Our gal from the legal dep told us we need to post this:

Dataset Ninja provides visualizations and statistics for some datasets that can be found online and can be downloaded by general audience. Dataset Ninja is not a dataset hosting platform and can only be used for informational purposes. The platform does not claim any rights for the original content, including images, videos, annotations and descriptions. Joint publishing is prohibited.

You take full responsibility when you use datasets presented at Dataset Ninja, as well as other information, including visualizations and statistics we provide. You are in charge of compliance with any dataset license and all other permissions. You are required to navigate datasets homepage and make sure that you can use it. In case of any questions, get in touch with us at hello@datasetninja.com.