KT to Combat Wildfires with 'AI Drone Platform'... Achieves 90% Detection Rate

Enhancing Accuracy with Drone Stations and AI Video Analysis
Expanding Drone Presence in Safety Sector

KT officials are operating wildfire monitoring drones. Photo by KT

[Asia Economy Reporter Oh Su-yeon] KT secured a 90% detection rate in a pilot project using drones to detect wildfire smoke in Wonju City. This is expected to serve as an opportunity to prevent wildfires, which have been difficult to pinpoint at their source and inadequately addressed initially due to reliance on visual observation.

On the 18th, KT held the 'KT AI Drone-Wonju City Pilot Project Performance Study' to introduce the results of the pilot project, including autumn wildfire monitoring drones. The wildfire smoke detection rate exceeded 90%. KT plans to enhance its technological capabilities in video analysis and control to foster a drone ecosystem. KT and Wonju City were selected in March for the Ministry of Land, Infrastructure and Transport's 'Drone Pilot City Establishment Project.' The wildfire monitoring drone is part of this project. The core is a drone that automatically replaces its battery and flies to detect and notify fire signs in mountainous areas using artificial intelligence (AI), preventing small fires from spreading into large wildfires.

Until now, wildfire monitoring relied on officials visiting the area or CCTV surveillance, which had limitations due to dependence on visual observation. KT utilized an AI drone platform for wildfire monitoring. Drones were flown in mountainous areas, and the real-time footage was analyzed by AI. When AI detects smoke in the footage, it can pinpoint the location more accurately than CCTV using GPS.

Wildfire Monitoring with AI Drones... Establishment of Dedicated Airspace Network and Base Stations

The key elements of the automated wildfire monitoring pilot are ▲ battery-swappable drone stations ▲ AI video analysis ▲ dedicated drone airspace network. The battery-swappable drone station is a device where drones return after flight. Equipped with a robotic arm, it automatically replaces the battery, enabling continuous operation.

The small station demonstrated in this trial consumes little power and is highly mobile. It assists wildfire monitoring in rugged mountainous areas that are difficult for humans to observe directly. The small station is mounted on a truck, moved near suspected wildfire areas, and then drones are launched. When drones capture footage, AI automatically analyzes it for fire signs such as smoke.

AI video analysis, which detects early-stage wildfires, improves accuracy. Previous wildfire monitoring using drones had limitations, relying on thermal cameras to find temperature differences or training on general wildfire footage. AI video analysis introduced the latest detection model, object segmentation, which clearly distinguishes various types of smoke instead of recognizing them as a single object. Since June, training has been conducted to achieve a wildfire smoke detection rate exceeding 90%, and it is currently undergoing accreditation procedures by the Korea Laboratory Accreditation Scheme (KOLAS). It can also track object size and range, allowing future expansion to recognizing arsonist behavior patterns and more.

Expansion of Drone Utilization in Safety Fields

KT officials are operating wildfire monitoring drones. Photo by KT

Since signing a drone-related business agreement with Wonju City, KT has established dedicated airspace network base stations at three locations: near Ganhyeon Observatory, Buron Industrial Complex, and Wonju Archery Range. These are used respectively for drone tourism, wildfire monitoring, and drone testing purposes. Security is enhanced by using VPN and cryptographic verification modules (KCMVP) on wireless communication terminals. Collected information is shared with related government and local government agencies for wildfire response and post-incident investigation.

At the mid-term review of the Drone Pilot City project held last September, Wonju City and KT were selected as one of the outstanding municipalities among nine participating local governments. Although the first phase of the Drone Pilot City project ends this December, the presence of drones in the safety sector is expected to grow further. Drones can be deployed not only at construction sites or for bridge and facility inspections but also in mountainous and coastal areas where citizen safety is needed.

This content was produced with the assistance of AI translation services.