Deaf and Hard of Hearing

A dataset of video streams of German sign language isolated words and sentences is collected 450 basic signs and 740 different meanings were performed by 25 native signers

A dataset of video streams of Greek sign language phonemes and sentences is collected under a standard framework(EAGLES). The video corpora were performed by native signers under controlled conditions.

The dataset consists of videos of American Sign Language. The videos show the sign language gestures from multiple angles and are added with linguistic annotations.

A point cloud dataset of static Polish sign language fingerspellings is collected to advance research on automatic sign language recognition. Unknown number of users performed fingerspellings for 16 Polish sign language alphabets, this data was recorded using a 3D Kinect sensor.

Signers talked about a series of themes such as, making appointments, talking about school days and more. RGB camera recorded the data and edited and annotated. RGB camera recorded signed conversations.

This work aims to build a big parallel corpus of English written texts and American Sign Language glosses useful for developing an automatic translator. Experts in ASL contributed to the collection or correction of bilingual corpus applying grammatical dependencies rules.

This work collects images of fingerspelling letters of German Sign Language for appearance-based gesture recognition. The database consists of 1400 image sequences that contain gestures of 20 different signers recorded by two different cameras, one webcam and one camcorder under non-uniform daylight lighting conditions.

A dataset of video streams of American sign language sentences is collected Multiple cameras collect videos of ASL signers, and then the video is manually annotated. 201 ASL sentences are singed by 3 signers to generate 201 annotated videos.

Two large scale datasets are released for continuous and isolated sign language recognition, consisting of French Belgian Sign Language (LSFB) conversation videos. Since 2012, 90 hours of video conversations between both native and non-native signersn were gathered with an RGB camera at 50 FPS in a studio with a controlled environment. 25 hours are fully annotated, and the process is ongoing.

This dataset covers over 200 signers, signer independent sets, challenging and unconstrained recording conditions to advance the sign language recognition community. ASL vocabulary publicly accessible videos were obtained and identified with 222 signers using face recognition and clustering who are distinct across train, validation and test sets.