Teacher dashboard (accessible online) that includes:
Student / classroom management
Reporting on student progress, scoring and usage
ASL lesson content is available for review
ASL lessons and games accessible on the system
Student access to lessons online (video only, no sign recognition)- to extend learning when away from the SignAll system
SignAll has the largest database of ASL vocabulary in the world
Thanks to our partnership with
In 2017, a strong partnership was forged with Gallaudet University (GU), the world’s leading university for the Deaf and Hard of Hearing (located in Washington D.C.). GU has been instrumental in moving the technology forward, providing guidance, and opening doors for SignAll.
In addition to this partnership, SignAll has been dedicated to having Deaf team members on board to provide necessary input and help us hold our vision in a way that celebrates Deaf culture and its heritage language.
SignAll has collected and annotated the largest ASL vocabulary database in the world, from a range of native signers.
Computer Vision and AI drive the technology
3D camera detects a “point cloud” image that has depth instead of colors.
Additional 3D (color) cameras detect the colored markers on gloves.
The information from both cameras are merged so we can locate where a signer is in 3D space.
As camera hardware improves, data quality will also improve. Over time we will be able to reduce the amount of hardware required, leading to a more portable solution.
3D Color Camera
3D Color Camera
Colored markers indicate space and time
Our users sometimes ask if the gloves are necessary. While we can extract large amounts of data without the gloves, the technology cannot accurately detect handshapes without them.
The system is calibrated to detect the subtle differences in colors, where they are located, and how fast they move.
Computational Linguistics: modeling human language for computer processing
Human language is astoundingly complex and diverse. We express ourselves in infinite ways: verbally, in writing, and signing. For example, there are hundreds of languages and dialects, and within those are unique sets of grammar and syntax rules, terms and slang.
The field of Computational Linguistics/NLP involves making computers to perform useful tasks with the natural languages humans use.
The fact that no written form of SL exists makes it quite difficult to research or implement statistical methods. Linguistically, many aspects are still under-researched, although high levels of details required for computational modeling are needed.
The necessary elements to capture
Other sign language technologies have failed because the necessary elements of the language were not captured.
The multiple data points that are captured and processed give context (both grammatical and meaning) to the language.
The skeleton shown is a representation of the body movements that are captured. Speed, posture, and movements relative to each other are all considered.
Challenges with signed languages + machine translation
Natural sign languages have a number of similarities to oral natural languages. However, the three dimensional nature of the space around a signer makes them even more difficult to model:
Non-manual features (multi-modal signs) carry additional information.:
Facial expressions associated with the position of eyebrows distinguish declarative (neutral brows), yes/no questions (raised brows) and wh-questions (furrowed brows).
Mouth patterns can provide adverbial information or help disambiguate manually similar signs.
Facial expression and body posture can indicate the signer's attitude to the accompanying proposition.
Syntactic - Agreement verbs incorporate within the sign information about person and number of the subject and indirect object.
The colored markers on the glove show where each hand is in space, and what shape it's making.
Also referred to as non-manual data (explained at right), they give context to ambiguous language.