In the project, novel methods are developed for the sawmill process optimization by utilizing state-of-the-art deep learning techniques. Especially virtual sawing and timber tracing are considered by developing methods to automatically learn the translation between multimodal log and board measurements. This would allow material traceability through production and predicting the output of the sawing providing a basis for more efficient and sustainable sawmill processes. Moreover, the developed multimodal domain translation techniques have numerous applications also beyond the sawmill industry. The project will be implemented as a collaboration between Computer Vision and Pattern Recognition Laboratory of LUT University and Finnos Oy, a developer of cutting-edge log measurements systems.

2022