Flame segmentation: Our AI model precisely segments all overlapping flames of the different burners in the infrared images. Our long-term R&D partner, the Karlsruhe Institute of Technology, provides the data source ("Dataset for the Segmentation of Industrial Burner Flames").
Artificial intelligence for object segmentation
Our AI model for object segmentation based on inspect infrared image sequences contributes to the optimization of industrial combustion processes in process plants, energy efficiency, and the reduction of CO₂ emissions. Process-relevant parameters can be extracted in real-time with the precise, pixel-wise segmentation of the flames (see image on the left). The detected flame bodies and the parameters extracted from these flames are available and visualized in your control room. You can use this information, for example, to optimize the burner settings and, thus, the process conditions in the plant. At the same time, the AI-based flame segmentation shows potentially damaging situations for the refractory lining in live operation.
If you are interested in object segmentation (e.g., of flames) for your plant, please contact us. Our deep-learning models are capable of flame segmentation with pixel accuracy and are designed to cater to a wide range of objects, ensuring that your specific needs are met.