Underlying DEEP is the insight of Daniele Raffa , CEO of Handy System, who recounted the genesis of the project, "My interest in sign language has been with me for many years. During the Covid-19 pandemic, I perceived even more clearly the difficulties of deaf people in communication, especially because of the masks. From there came the idea to develop a technology that would provide immediate access to services, without the need for specific devices."
Design Challenges and Future Prospects
The project required intensive research and development, explained Nicolas Tagliabue , scientific collaborator at SIF: "The main challenge was to create a technology that could capture the full expressiveness of LIS, which relies not only on hands, but also on facial expressions and body movements. We developed a non-invasive system that could track every detail and return an accurate translation. However, this very diversity made the translation process particularly challenging for an artificial intelligence system."A key aspect was the direct involvement of the hearing impaired community. "From the very beginning, we wanted the system to respond to the real needs of and for users," explained Elisa Colletti , scientific collaborator at SIF. "For this, the community contributed to the creation of a corpus of more than 62 hours of LIS videos, which was essential for training the AI." In the testing phase, participants evaluated the prototype in real-world scenarios, providing valuable feedback. "Their observations allowed us to improve DEEP, making it not only advanced, but really useful in everyday life."
Looking to the future, the project envisions an expansion of application areas, as indicated by Daniele Raffa: "In addition to municipal offices and pharmacies, it will be important to be able to implement DEEP in the emergency sector, to facilitate communication with emergency services."