Automatic consideration measurement of timber with AI/vision technology. Preliminary study
Causes of downgrading that are currently assessed manually should be detectable using image recognition, which can complement the systems used today to get closer to a fully automated quality assessment of sawlogs.
The goal is to create a database of annotated video material of defects that can be used to train and verify machine learning models for the downgrading defect types that are currently assessed manually, as well as trained models for some defect types that perform at least as well as the manual assessment. We will focus the machine learning in this project on the defect types that are detectable in visible light, well aware and coordinated with the ongoing project investigating the use of NIR to detect defects on log ends.
Project time: 2021-06-01 - 2022-03-31
Budget TCN: 740000 SEK
Area: Materials & processes
Budget: 740 000 SEK
Schedule: June 2021 - March 2022
Project leader: Peter Bomark, RISE
Funding: This subproject is funded by TCN's industrial partners.
Updated: