E-Paper
Sign In
UAE students develop lifesaving system to detect drowning incidents in real time
The system is designed to integrate seamlessly into existing pool surveillance networks without extensive modifications
- PUBLISHED: Tue 18 Feb 2025, 4:41 PM
A team of Zayed University students has developed an artificial intelligence-powered drowning detection system, a potential lifesaver with a remarkable 99.7% accuracy rate in identifying distress situations.
The project, led by Hamad Alzaabi, Saif Alzaabi, and Majed Alhammadi under the supervision of Dr Sara Kohail, leverages AI and computer vision to detect distress situations in real-time, ensuring rapid response and intervention.
Recommended For You
This groundbreaking achievement was made possible through a meticulously curated dataset comprising over 69,000 manually annotated frames.
Innovative approach to data collection
“Our dataset captured a wide range of aquatic behaviours, including normal swimming, struggling, and simulated drowning,” the students explained. “To enhance reliability, we curated a balanced dataset, preventing bias and improving generalization across scenarios.”
Stay up to date with the latest news. Follow KT on WhatsApp Channels.
Initially, the team explored various AI-based approaches, including standard object detection and pose estimation models. However, early experiments revealed that existing methods struggled with underwater distortions, motion artifacts, and environmental variability. The students developed an annotated underwater dataset that captures diverse drowning and swimming scenarios to address this.
“We then built our own AI model based on this data and integrated it into a web-based platform, allowing for live monitoring, automated alerts, and scalable deployment,” they said. “Since public datasets were insufficient for underwater drowning detection, we took a proactive approach by creating our own.”
The students recorded high-resolution underwater videos in a controlled facility to ensure high-quality training data, capturing different swimming behaviours under varied conditions. Each frame was manually labelled using annotation tools. The dataset will soon be publicly available to support further research in underwater computer vision.
Inspiration behind the innovation
Before developing their system, the students extensively researched drowning statistics and existing safety measures. “Reports from the World Health Organization (WHO) highlight that over 236,000 people die from drowning each year, with children being at the highest risk,” they noted.
Many current surveillance systems are either reactive, relying on post-incident analysis, or require wearable sensors, which may not be practical in all settings. “We saw an opportunity to leverage AI and computer vision to create a proactive, non-intrusive system capable of detecting drowning incidents in real-time,” they stated.
Overcoming challenges
The students encountered several technical and logistical challenges during development. One major hurdle was overcoming underwater distortions, such as refraction and motion blur, which impact object detection accuracy. Another challenge was achieving real-time performance while processing high-resolution video streams.
The system is designed to integrate seamlessly into existing pool surveillance networks without extensive modifications. “It features a web-based dashboard that provides real-time alerts, live camera feeds, and incident tracking. Users can configure alert settings, manage notifications, and review past detections through an intuitive interface, making it accessible for lifeguards, pool managers, and safety personnel,” they added.
The team ensured that facilities of different sizes could adopt the technology by using cost-effective underwater cameras and offering both cloud-based and on-premise deployment options. The AI model is optimized for efficiency, allowing it to run on commercially available GPUs without requiring high-end computing resources.
Future prospects
Looking ahead, the students plan to refine the system through further real-world testing and explore commercialization opportunities. “Securing funding would allow us to scale the project and expand its reach,” they said.
Reflecting on their journey, the students emphasised the broader societal impact of their work. “This project has reinforced the importance of using technology to address societal challenges. AI is not just about automation or efficiency. It has the potential to save lives. The experience has motivated us to continue exploring AI-driven solutions for public safety.”
They also hope their success will inspire others. “We want to show that Emirati students can compete globally in AI and innovation. By addressing real-world challenges, we hope to inspire future generations to drive technological advancements and contribute to the UAE’s vision of becoming a global leader in AI and smart solutions.”
Azza Al Ali is a reporter from Ras Al Khaimah, dedicated to sharing stories that bring the community’s voices to life. She’s passionate about connecting with people and highlighting what truly matters to them.