Quadrotor UAV-based smoke detection system using YOLOv8 improves wildfire prevention

Researchers from Tecnológico de Monterrey have developed a cutting-edge aerial smoke detection system combining a quadrotor UAV and an optimized YOLOv8 nano model, capable of real-time detection under challenging conditions, the system achieves 95% precision and 88.5% recall in detecting early-stage wildfires. Published in Robot Learn, this innovation addresses wildfire detection in remote, high-risk areas while operating efficiently on low computational resources.

Wildfires pose an ever-growing threat to global ecosystems and communities, with climate change exacerbating their frequency and severity. Early detection is critical to preventing these disasters from escalating. A team led by Prof. Herman Castañeda from Tecnológico de Monterrey has developed an innovative solution: a quadrotor UAV-based smoke detection system using the YOLOv8 nano object detection model.

“Our solution leverages the cost-efficiency, speed, and accessibility of drones, combined with state-of-the-art object detection, to identify smoke in hard-to-reach and high-risk areas,” says Castañeda.

The system features a Quadrotor UAV equipped with a standard camera and a YOLOv8-based detection algorithm deployed in a Robotic Operating System (ROS) framework. The system is optimized for real-time applications, demonstrating exceptional performance even on computationally limited devices like Intel Core i5 CPUs.

Key Achievements:

Real-Time Detection: Smoke is quickly detected, enabling early action to prevent wildfires from spreading.

High Accuracy: With 95% precision and 88.5% recall, the system reliably identifies smoke, even under conditions such as wind, mist, or low visibility.

Optimized Deployment: Through hyperparameter tuning and OpenVINO optimization, the model achieves robust detection without sacrificing speed.

Versatile Testing: The system was validated in laboratory conditions, using controlled smoke images, and in field experiments with real smoke, dynamic lighting, and environmental interference.

Why This Matters:

Most existing wildfire detection systems focus on identifying flames, often missing the critical early stages where smoke is the primary indicator. By detecting smoke before fires escalate, this UAV-based solution offers a timely, scalable, and cost-effective method to combat wildfires.

Challenges Addressed:

The system avoids reliance on high-cost computing, making it deployable in remote areas.

It overcomes traditional detection biases through extensive testing and dataset augmentation, which included over 2,000 labeled images covering various smoke conditions.

Real-World Impact:

This innovative system can be deployed in forests, mountains, and other inaccessible regions, offering a rapid response to emerging fire threats.

Next Steps:

The team plans to integrate the UAV system with automated fire suppression mechanisms, enabling comprehensive wildfire management. Future implementations could include fleets of drones working collaboratively to monitor larger areas and assist in real-time firefighting operations.

This breakthrough marks a significant step forward in leveraging AI and UAV technologies to protect ecosystems and communities from the devastating impacts of wildfires.

This research, titled “Quadrotor UAV-Based Smoke Detection System Using YOLOv8 Towards Wildfire Prevention,” was published in Robot Learn. DOI: 10.55092/rl20250001.