Inspirational cases

Winner 2023
Bolivia
Share
#Fairsociety

With artificial intelligence, students translate sign language in real-time

Programming and mechanics were the path to this creation, which aims to reduce communication barriers for people with hearing deficiencies.

Teacher

Foto de Alfredo Callisaya Huanca
Alfredo Callisaya Huanca

Schools

Unidad Educativa Japón B

Project name

Detección de lenguaje de señas en tiempo real

STEM areas

Math, Sciences, Technology

Other areas of knowledge

Mecânica

Sign language can be a powerful tool for people with hearing disabilities. However, there are still barriers between this language and the spoken ones. Intending to facilitate this interaction, a young group from Bolivia developed a computer program that translates Spanish words into signs in real-time. The creation was the winner of Solve for Tomorrow, in its first edition in the country (2023).

The students were between 16 and 17 years old and in the last year of secondary and compulsory schooling. In Bolivia, this stage of teaching has a humanistic and technical focus. At the end, students obtain a bachillerato degree, that is a diploma from a Technical School. “In the morning, basic subjects such as mathematics and social sciences are taken. In the afternoon, specialty areas are taken. These project students are from Computer Systems,” says the group´s mediator teacher, Alfredo Huanca.

When Solve for Tomorrow put out the call for registrations, several groups from the Educational Unit began to form ideas, as it was the case with “Visionarios” (“Visionaries”, in English), the real-time sign language detection program team. For them, the idea began in a class by teacher Huanca about artificial vision, the technology that uses computer intelligence combined with information obtained from images. In other words, based on what the machine sees, it can generate actions or make decisions.

The young people thought they might be able to turn signs into words. “They came with ideas, they already had had difficulties communicating with people with hearing disabilities and they wanted to help with their technical knowledge,” highlights the educator.

First, they thought about creating a physical prototype, with a screen, sensors, and microphone. But then they realized that everything could be done from the computer, with an internal microphone and camera. “We use a computer vision model that already has a database. So when you move your hand to make a signal that is in the system, you can already interpret what that represents in words or numbers,” he explains.

But artificial intelligence needs to be “taught” to do its job. The team then had to video record the words they would like the system to translate at that stage. The coordinator of the Humanistic Technical Baccalaureate of the Educational Unit was an important ally because she already knew sign language in depth and suggested the vocabulary that would be essential to include. “She also taught us about the precision of movements, pointing out that sometimes people do not show their hands well, so we needed to record the movements more slowly to improve the system’s ability to detect them,” the teacher explained.

Detecto de lenguaje de señas

Other support was from the Autonomous Municipal Government of El Alto, which has a Care Unit for People with Disabilities. “My coordinator and I decided to visit the Unit and they welcomed us with open arms. They told us that this technology is what they needed and they helped us do the word choice and the tests,” says teacher Alfredo Huanca. The Unit’s specialists were the recipients of the tests and gave their opinions. Since it was a prototype, they jointly decided to focus only on basic words and expressions, which the students learned and reproduced in the software, such as “Hello” and “What time is it?”


He states that the Solve for Tomorrow mentorships were very important to implement the STEM methodology and understand how class projects can contribute to sustainable development goals. “The program coordinators were always with me and, as an educator, I learned a lot”, evaluates.

We learn that we need not only to have patience and a lot of attention but also to always try to see things from another perspective to find a solution  in programming, the teacher says.

Now, the software is ready to be implemented on any computer with an updated operating system and can be useful for educational institutions, hospitals, public management, and banks, among others. In 2024, the educator even intends to create a website for free downloading of  the system, perhaps involving other students, becoming a school-wide project. In addition, they already received an invitation to ally with a local private university, after the results of Solve for Tomorrow came out. The objective is to still test the program with the final audience: people with hearing disabilities.

“The impact has been quite large not only in the group but also in the Educational Unit. The students of the lower grades now have this vision that it is possible to achieve such an achievement and colleagues from different areas evaluated the project positively and said that we have to continue doing this type of work,” reinforces the teacher.

Now, the young people from “Visionarios” are in university and one of them, José Pablo, is studying for a degree in Systems Engineering. “He has already seen that a lot can be done with technology and he wants to continue further,” believes Huanca.

Focus on the practice!

Take a look at the teacher's guide on how to create a sign language screening program.

Empathize

The idea was born while the students were learning about artificial intelligence. When they learned about the potential uses of the computer’s ability to process images, they thought that this machine could be a good translator between people with and without hearing disabilities.

Define

As a group, the students did bibliographic research and deepened on programming management. Furthermore, from the beginning, the teacher organized the specific roles of students according to their affinities. They downloaded and installed the programming libraries, which are a set of codes that already have processing functionality.

Ideate

First, they thought about creating a physical prototype, with a screen, sensors, and microphone. But then they realized that everything could be done from the computer, with the internal microphone and camera. The solution turned out to be simpler and cheaper with the use of Python language (open-source programming language) and other free programming resources.

Prototype

To get to the prototype, they created the main file by importing the necessary libraries and general configurations. Then, they developed the algorithm for real-time hand detection. Thus, they configured the software’s work environment to enable more computer vision inferences from sensory data, such as video or audio. After a few more stages and persisting to make the program work, they managed to conclude the project. More details here (in Spanish).

Test

The final product was tested with the Care Unit for People with Disabilities of the Autonomous Municipal Government from El Alto and the Humanistic Technical Baccalaureate coordinator of the Educational Unit. “They have given us suggestions on what we can improve in the next phase, to be able to bring it to the final public,” adds mediator teacher Alfredo Huanca. The idea is to continue the project in 2024, even with other students from the school and more allies.

#Schedule

Solve for Tomorrow is present in several countries in the region.

In each place, the program follows a specific schedule and rules, adapting to the local reality. Visit your country's website.

Check the date of the 2024 finalists announcement.

Subscribe to the newsletter of Solve for Tomorrow Latam platform and receive monthly the latest updates on the program in the region.

Subscribe