Dr. Iqbal, a newly graduated doctorate from the Physics Study Program, Faculty of Mathematics and Natural Sciences, University of Indonesia (FMIPA UI), has successfully created an innovative deep learning-based artificial intelligence (AI) model that can transform CCTV cameras into real-time and accurate rainfall intensity prediction tools. This discovery offers an effective and low-cost rainfall measurement solution, addressing the challenges of rainfall observation in previously unreachable areas.
Dr. Iqbal completed his doctoral studies with cum laude honors after defending his dissertation entitled “Rain Intensity Estimation Based on Surveillance Camera Imagery Using Hybrid Deep Learning Model” in an open doctoral promotion session held on Monday, July 7, 2025 at the Prof. Dr. G.A. Siwabessy Hall, FMIPA UI, Depok.
The developed AI model combines advanced image processing technologies, VGG19 and ResNet34, modified to improve the system’s focus on recognizing key features of CCTV images. This hybrid system adapts to changes in lighting and camera position, consistently predicting rain intensity, both during bright daylight and at night with low light.
“The goal of this research is to provide a rainfall estimation system that is equitable and easily accessible. By leveraging existing infrastructure such as CCTV, we can expand the rainfall observation network without requiring significant investment in new hardware,” said Dr. Iqbal.
This model has been tested on over 25,000 training images and demonstrates superior performance with a very low error rate and high accuracy, surpassing similar models. Another advantage is its ability to perform optimally in low-light conditions and with less than ideal camera angles.
This innovation has significant potential to support the transportation, agriculture, and healthcare sectors, as well as disaster early warning systems, especially in areas lacking adequate rainfall gauges. The widespread use of CCTV as an AI-based weather sensor offers a practical and innovative solution.
“This hybrid model proves that deep learning can be applied to solve social problems, from flood mitigation to supporting intelligent transportation systems,” added Dr. Iqbal.
Doctoral promotion session Dr. Iqbal was led by the Dean of FMIPA UI, Prof. Dede Djuhana, Ph.D., under the guidance of promoter Prof. Alhadi Bustamam, Ph.D., with co-promoter I Dr. Adhi Harmoko Saputro, Ph.D. (Department of Physics FMIPA UI), as well as co-promoter II Dr. Ardhasena Sopaheluwakan, M.Sc. (BMKG).
Prof. Alhadi Bustamam stated, “This finding broadens the scope of research in digital image processing, machine learning, and applied meteorology. Practically, this innovation offers an efficient and affordable solution that can be used in early warning systems and intelligent transportation (ITS).”
