Page 183 - AI for Good-Innovate for Impact Final Report 2024

AI for Good-Innovate for Impact

Use case – 42: AI 4 Health – Live Primary Health Care African

National Sign Languages Translation tool 42 - PSG

Country: Zimbabwe
Organization: Purple Signs Global

Contact person: Dominic Tinashe Tapfuma,

tapfumadominic@ gmail .com

+263 773 744 246, +263 292 200 040

42�1� Use case summary table

Domain Artificial Intelligence in Healthcare
The problem to be 1. Limited access to critical health information and services.
addressed 2. Difficulty in effective communication between deaf individuals and
service providers.
3. Marginalization of deaf and hard of hearing individuals, hindering
their ability to participate fully in society and the economy.

Key aspects of the AI-powered Live Sign Language Translation
solution Multimodal Communication Capabilities

Technology AI-powered Translation, Text to Speech, Speech to Sign Language,
keywords Facial Animation, AI Dubbing, Healthcare Accessibility, RMNCAH
(Reproductive, Maternal, Newborn, Child, and Adolescent Health &
Nutrition).

Data availability Private data available
Metadata (type of 1. Creating visuals with video
data) 2. Can convert any text to visual.
3. Computer vision to computer generated visuals
Model Training and Using AutoML, which automatically prepares a dataset for model
fine-tuning training, performs a set of trials using open-source libraries such as
scikit-learn and XGBoost, and creates a Python notebook with the
source code for each trial run so you can review, reproduce, and modify
the code.
Using hyperparameter tuning for fine tuning the model . Making use of
HyperOpt, scikit-learn, MLflow libraries.

167