Welcome to the LIVE American Sign Language Video Quality Assessment Database
LIVE-ASL Video Quality Assessment Database
Introduction
Technological advancements of communication systems have enabled the Deaf and Hard of Hearing community (DHH) to share and receive signed messages via video calls, video conferencing, and videos shared over the internet. Given the presence of data degradation in all these media, it is important to be able to measure the impact of distortions on the perceived quality and intelligibility of received and viewed sign language videos. For an ASL user, the presence of distortions on a video containing signed messages may significantly impact their ability to understand the information being conveyed in the video in the form of signing. Towards promoting study of this topic, we present the new LIVE-ASL VQA database, which is a publicly available video quality database that contains 358 high-quality compression-distorted videos computed from 149 unique ASL-video-message contents recorded with the help of 21 sign language speakers in a controlled setting. We conducted a large-scale human study on this data with the help of 42 paid volunteers from the DHH community, including but not limited to professional ASL experts, each of whom rated the intelligibility and perceptual video quality of all the 358 videos. We analyzed the outcomes of the human-study towards understanding the consistency and behaviour of the human responses. To demonstrate one aspect of usefulness of the new dataset, we trained and compared a variety of existing VQA models on their ability to predict both distorted ASL video intelligibility and video quality. We found that these models have significant scope for improvement when analyzing sign language videos.
We are making the LIVE-ASL Video Quality Assessment Database available to the research community free of charge. If you use this database in your research, we kindly ask that you to cite our paper and website listed below:
- S. Mishra, S. Gupta, R. Pahwa, M. H. Pinson, A. C. Bovik, "LIVE-ASL: Subjective and Objective Quality Assessment of American Sign Language Videos", arXiv:arXiv_ID [Arxiv]
- S. Mishra, S. Gupta, R. Pahwa, M. H. Pinson, A. C. Bovik, "LIVE-ASL Video Quality Assessment Database," Online: https://live.ece.utexas.edu/research/LIVE-ASL-VQA/index.html, 2024.
Download Link Here ! Please fill the Google Form to get access to the database.
Database Description
The LIVE-ASL VQA Database contains 358 video sequences generated from 149 high-quality (pristine) reference source videos. Each video was compressed using 2 or 3 resolution-bitrate protocols randomly chosen from an overall bank of 8 protocols. Our database comprised of videos of spatial resolution ranging from 144p to 1080p. The videos were encoded at bitrates in the range 100kbps to 800kbps. The compressed videos were then presented as stimuli in our subjective laboratory experiment to 42 human subjects who were verified as being fluent in ASL. The subjects rated the videos on the basis of both understandability (or intelligibility) and video quality. To demonstrate one aspect of usefulness of the new LIVE-ASL VQA dataset, we trained and compared a variety of existing VQA models on their ability to predict both distorted ASL video intelligibility and video quality. We found that these models have significant scope for improvement when analyzing sign language videos.
Investigators
The investigators in this research are:
- Sandeep Mishra ( sandy.mishra@utexas.edu ) -- Graduate student, Dept. of ECE, UT Austin.
- Shashank Gupta ( shashank.gupta@utexas.edu ) -- Graduate student, Dept. of ECE, UT Austin.
- Ramit Pahwa ( ramit@utexas.edu ) -- Graduate student, Dept. of ECE, UT Austin.
- Margaret H. Pinson ( mpinson@ntia.gov ) -- National Telecommunications and Information Administration
- Alan C. Bovik ( bovik@ece.utexas.edu ) -- Professor, Dept. of ECE, UT Austin
Copyright Notice
-----------COPYRIGHT NOTICE STARTS WITH THIS LINE------------
Copyright (c) 2023 The University of Texas at Austin
All rights reserved.
Permission is hereby granted, without written agreement and without license or royalty fees, to use, copy, modify, and distribute this database (the videos, the results and the source files) and its documentation for any purpose, provided that the copyright notice in its entirety appear in all copies of this database, and the original source of this database, Laboratory for Image and Video Engineering (LIVE,
http://live.ece.utexas.edu
) at the University of Texas at Austin (UT Austin,
http://www.utexas.edu
), is acknowledged in any publication that reports research using this database.
The following paper/website are to be cited in the bibliography whenever the database is used as:
- S. Mishra, S. Gupta, R. Pahwa, M. H. Pinson, A. C. Bovik, "LIVE-ASL: Subjective and Objective Quality Assessment of American Sign Language Videos", arXiv:arXiv_ID [Arxiv]
- S. Mishra, S. Gupta, R. Pahwa, M. H. Pinson, A. C. Bovik, "LIVE-ASL Video Quality Assessment Database," Online: https://live.ece.utexas.edu/research/LIVE-ASL-VQA/index.html, 2024.
IN NO EVENT SHALL THE UNIVERSITY OF TEXAS AT AUSTIN BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OF THIS DATABASE AND ITS DOCUMENTATION, EVEN IF THE UNIVERSITY OF TEXAS AT AUSTIN HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
THE UNIVERSITY OF TEXAS AT AUSTIN SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE DATABASE PROVIDED HEREUNDER IS ON AN "AS IS" BASIS, AND THE UNIVERSITY OF TEXAS AT AUSTIN HAS NO OBLIGATION TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
-----------COPYRIGHT NOTICE ENDS WITH THIS LINE------------
Back to Quality Assessment Research page