It is a technology that detects fires early by detecting smoke or sparks within the camera's shooting range.
For a quick initial response AI forest fire detection technology
99%
Recognition accuracy
High recognition accuracy of 99%
Of the 74,146 images of forest fires collected from actual cases, 73,442 were successfully classified as forest fires
0.0011
FPR
Low false positive rate
Minimize the rate at which events other than forest fires are detected as forest fires, and prevent dispatch and response resources when they are not forest fires
No. 1 in Korea, abnormal situation detection Core technologies
Wildfire Detection
Forest fires are detected in outdoor images or images input through an RGB/IR camera. Smoke is found in daytime RGB images, and flame ignition is the main characteristic in nighttime IR images.
Continual learning system
It continuously learns error data that occurs in the CCTV environment. When an AI model is applied to a real environment, new data is continuously flowing in, and this data also includes error data such as undetected/false detection. In Alchera's Anomaly Analysis (Anomaly Analysis) Lab, active learning (Active Learning) that automatically detects errors and We are studying continuous learning (Continual Learning), which effectively learns errors.
Use cases of abnormal situation detection technology
.svg)
.avif)
.svg)
.svg)
.svg)
.svg)
