Mobile Deep Learning System That Calculates UVI Using Illuminance Value of User's Location

Area of Science:

• Mobile computing
• Artificial intelligence
• Environmental science

Background:

• Accurate Ultraviolet (UV) information is crucial for human health, yet current UV Index (UVI) services are often localized and require expert interpretation.
• Existing limitations in user-based UV measurement hinder access to personalized UV exposure data.
• Research on leveraging mobile devices like smartphones for UV measurement and information dissemination is underdeveloped.

Purpose of the Study:

• To propose and validate a mobile deep learning system for calculating the Ultraviolet (UV) Index (UVI) using smartphone illuminance sensors.
• To overcome the limitations of traditional UV measurement methods by enabling user-based, location-specific UVI estimation.
• To demonstrate the feasibility of providing accurate UV information via mobile devices even in areas lacking dedicated UVI monitoring equipment.

Main Methods:

• Developed a deep learning model using TensorFlow, analyzing the correlation between illuminance and UVI based on a collected natural light database.
• Selected optimal input variables for accurate UVI calculation and designed a deep learning model with optimized layers and nodes.
• Converted the deep learning model for mobile deployment and loaded it onto mobile devices equipped with illuminance sensors.

Main Results:

• The proposed mobile deep learning system successfully calculated UVI using smartphone illuminance sensor data.
• Experimental results showed the system provides UV information with 90-95% accuracy compared to a spectrometer in both summer and winter.
• The method enables real-time, location-specific UVI estimation through readily available mobile devices.

Conclusions:

• The developed mobile deep learning system effectively provides accurate, user-based UV information, enhancing public health awareness and safety.
• This approach democratizes access to UV data, removing the need for specialized equipment and expert knowledge.
• The system's high accuracy across seasons highlights its potential for widespread adoption in mobile health applications.