A researcher from De La Salle University is developing an artificial intelligence (AI) system that uses unmanned aerial vehicles (UAVs) to detect structural damage, as he advanced to the finals of the National Academy of Science and Technology Philippines Talent Search for Young Scientists (NTSYS) 2026.
Timothy Scott C. Chu, an assistant professor at DLSU, presented his study on UAV-based infrastructure monitoring during the April 22 finals held at the Admiral Hotel Manila.
Chu’s research focuses on deploying AI models on UAV platforms to detect and quantify cracks in infrastructure exposed to earthquakes, typhoons, and environmental degradation.
His study, titled “A UAV-Based Concrete Crack Detection and Segmentation Using 2-Stage Convolutional Network with Transfer Learning,” proposes a two-stage inspection process to improve efficiency and accuracy.
The system first uses an AlexNet-based model to determine whether cracks are present in captured images.
Only images identified with potential damage are then processed using YOLOv4 for detailed localization and segmentation. Chu described this as performing the “inspection task in two steps instead of trying to do everything at once.”
The approach avoids unnecessary analysis of undamaged images, contributing to faster processing while maintaining accuracy. Experimental validation showed the system achieved over 90% accuracy and success rate.
“The proposed system contributes to practical inspection workflows aligned with Industry 5.0 principles by supporting human-centered, intelligent infrastructure monitoring, strengthening national infrastructure resilience, and advancing Sustainable Development Goals (SDG) 9 (Industry, Innovation, and Infrastructure) and 11 (Sustainable Cities and Communities),” Chu said.
Chu said the system has passed the proof-of-concept stage but is not yet ready for deployment in national infrastructure programs.
“Scaling toward operational deployment will require integration with larger UAV platforms, which is currently in development,” he added.
Chu is among five finalists in this year’s NAST PHL search, which selected candidates from 19 entries nationwide.
