Database
The DEVISIGN database has been constructed under the sponsors of Microsoft Research Asia by VIPL group, ICT, CAS. The goals to create the Chinese Sign Language database include: providing the wordwide researchers of SLR community a large vocabulary Chinese SL database for training and evaluating their algorithms; advancing the state-of-the-art SLR technologies aiming at practical applications especially for the unknown signer situation.
Currently, the DEVISIGN database covers 4414 standard Chinese Sign Language vocabularies and totally contains 331,050 vocabulary data of 30 individuals(13 males and 17 females). Each vocabulary data is composed of RGB video(in “avi” format), the depth and skeleton information (in “bin” format). This database contains both inter-sign and intra-sign changes. For 15 signers, the data were collected by 4 times and for the remaining signers, the data were captured only one time.
In the first stage, this DEVISIGN database is partly made available (three subsets) for research purpose. VIPL is serving as the technical agent for distribution of the database and reserves the copyright of all the videos and binary files in the database.
Subsets
Three subsets of the whole DEVISIGN database have now made publicly available. The first subset is composed of 26 letters and 10 numbers and named as DEVISIGN-G. The second subset is composed of 500 daily used vocabularies (including the words in DEVISIGN-G) and named as DEVISIGN-D. The third subset is a large vocabulary dataset, which contains 2000 Chinese SL vocabularies (including the words in DEVISIGN-D) and is named by DEVISIGN-L.
For these three datasets, the only difference is the vocabulary size, while the signers and the record times are consistent. The data covers 8 different signers. Among them, the vocabularies are recorded twice for 4 signers (2 males and 2 females) and one time for the other 4 signers(2 males and 2 females). The following table summarizes the content of the three DEVISIGN subsets.
| Subset | Vocabulary | Signer | Record times | Total videos |
|---|---|---|---|---|
| DEVISIGN-G | 36 | 8 | 4 signers:2 4 signers:1 | 432 |
| DEVISIGN-D | 500 | 6000 | ||
| DEVISIGN-L | 2000 | 24000 |
Some example videos in DEVISIGN-G:
P04_0001_1(One)
P04_0001_2(One)
P07_0001_1(One)
Some example videos
DEVISIGN-G DEVISIGN-D DEVISIGN-LRequest
The DEVISIGN-G/-D/-L databases are released to universities and research institute for research purpose only. To request the access right to download the data resources, please do as follows:
(1) Download the DEVISIGN database Release Agreement, read it carefully, and complete it appropriately. Note that the agreement should be signed by a full-time staff member (that is, student is not acceptable). Then, please scan the signed agreement and send it to Mr. Yuecong Min (yuecong.min@vipl.ict.ac.cn ) and CC to Dr. Chai (chaixiujuan@ict.ac.cn). If you are from the mainland of China, please sign the agreement in Chinese rather than English. Then we will verify your request and contact you on how to get the database. It should be noted that it is not recommended to download the data directly from the internet for the huge amount of data. We would like to transfer the data by mailing a mobile HDD. You can choose to mail a free HDD to us or remit the corresponding amount to our appointed account.
(2) Before using the database, you are recommended to refer our Technical Report on DEVISIGN. You might also be interested in other papers we have published on sign language recognition by visiting here.
Any other questions on the sign language database, please contact:
Mr. Yuecong Min
Visual Information Processing and Learning (VIPL) group
Institute of Computing Technology, Chinese Academy of Sciences
Tel: +86 10 6260 0551
Fax: +86 10 6260 0548
Email: yuecong.min@vipl.ict.ac.cn