DATASET / COMPUTER VISION & ACCESSIBILITY
Sign Language Recognition Dataset – 2,000 Words
Compiled and published on Kaggle by Pamindu Karunadasa for sign language education, computer vision, and recognition research.
- Video files
- 17,668
- Folders with videos
- 1,947
- Class folders
- 2,000
01 / OVERVIEW
A resource for sign language research
I compiled and published this video collection as a Kaggle dataset. Videos are organized by word or class so researchers and students can explore sign recognition, gesture classification, and accessibility applications.
The Kaggle data card and dataset report provide the file inventory and collection details. This page describes the published collection; the dataset files are hosted on Kaggle.
02 / COVERAGE
What the published report contains
The dataset report lists 17,668 MP4 videos, 2,000 word or class folders, and 1,947 folders containing video data. The remaining 53 folders contain no videos. These are inventory counts reported with the dataset, rather than a claim that every word has usable examples.
Researchers should inspect the inventory, validate the samples, and choose training and evaluation splits appropriate to their task.
03 / SOURCE & ATTRIBUTION
Compilation and original video sources
Pamindu Karunadasa compiled and published the Kaggle collection. The original video samples were collected from SignASL and its contributors; publishing this collection does not make those original videos my own recordings.
The Kaggle data card specifies “Other” for the license and explains the original copyright holders’ rights. Consult its source and usage notes before using or redistributing the videos.