SignersMask
Digital mask for anonymous participation of deaf people in discussions and comments on the Internet
In social media, communication is usually in writing and most users use a pseudonym so that they remain unrecognized on the net. Although anonymity can also lead to people spreading hate comments or engaging in agitation, anonymity also offers protection for those who wish to participate openly in discussions and debates, for example, without enabling a platform for attacks on the person.
Unfortunately, deaf people do not have this choice, as they communicate primarily in sign language and post sign language videos on social media. They do this because access to reading and writing is challenging and therefore a barrier for them due to their hearing impairment. In a posted sign language video, they can be seen as a whole person and this contradicts informational self-determination and the right to pseudonymization of data (DSGVO/BDSG). Thus, deaf people cannot make anonymous comments and cannot make use of anonymous counseling, which would be important, for example, in the case of psychological crises.
Therefore, it would be necessary to alienate sign language comments so that signers* appear anonymous in their videos. In collaboration with the Visual Computing Institute (VCI, Prof. Leif Kobbelt, Isaak Lim), the Competence Center for Sign Language and Gestures (SignGes, Prof. Dr. Irene Mittelberg, Dr. Klaudia Grote) is therefore creating innovative software that generates a kind of masking of the signing person, so that an avatar instead of the person can be seen in real time on the posted videos. To make this possible, a digital neural network will be trained to map input videos of the signer* to pose information (face and hands) as output. This technology would not only enable more accessibility for the 70 million signers* worldwide, but can also be usefully employed by non-disabled individuals in other areas.
Cooperation
RWTH Visual Computing Institute (VCI)
Funding
RWTH Aachen Seed Fund – Open Seed Fund 2020
(OPSF598)
Links