Groundbreaking AI Technology Developed to Identify Asbestos in Building Roofs: A Study by UOC Scientists
May 3, 2024
A group of scientists from the Universitat Oberta de Catalunya (UOC) have developed a groundbreaking method for identifying asbestos that remains on building roofs despite legal mandates. Their innovative approach, utilizing artificial intelligence, deep learning, and computer vision techniques on aerial RGB images, has been published in the journal Remote Sensing.
Traditionally, identifying asbestos on roofs required complex and costly multiband images. However, the new software developed in collaboration with DetectA offers a cost-effective and scalable solution using widely available RGB images. This advancement marks a significant competitive edge in monitoring the removal of hazardous asbestos material.
Lead researcher Javier Borge Holthoefer from the Complex Systems Group at the Internet Interdisciplinary Institute highlighted the versatility and adaptability of the methodology, emphasizing its potential for systematic and efficient monitoring of asbestos removal.
Researchers, including Àgata Lapedriza from the UOC’s Faculty of Computer Science, Multimedia, and Telecommunications, collaborated with DetectA founders Carles Scotto and César Sánchez, along with UOC doctoral students, to train a deep learning system using thousands of images from the Cartographic and Geological Institute of Catalonia.
The success of the project, with a success rate of over 80%, demonstrates the AI system’s ability to identify asbestos in aerial photos by analyzing various patterns in the roofs and surroundings of buildings. This technology will benefit rural, coastal, industrial, and urban settings, aiding in the removal of asbestos from public buildings.
Despite asbestos being banned for use in buildings over two decades ago, it continues to pose a serious threat to public health, with millions of tons still in use in Catalonia alone. The development of this technological solution addresses the challenge of locating asbestos in buildings, facilitating its safe removal by certified professionals.
With legal deadlines for asbestos removal approaching, the UOC researchers’ system offers a cost-effective and efficient solution for authorities to identify and remove asbestos from buildings. The team is also exploring ways to enhance the system’s effectiveness in rural settings, where asbestos wear and conservation may be concealed by vegetation.
The study, titled “Explainable Automatic Detection of Fiber-Cement Roofs in Aerial RGB Images,” published in Remote Sensing, provides a comprehensive overview of the innovative methodology. The research offers a promising solution to the ongoing challenge of asbestos detection and removal.
For more information, visit UOC’s official website.
