 The tree extraction project aimed to evaluate automatic tree extraction methods based on laser scanner data and found high variation in quality under boreal forest conditions and varying laser point densities. Omissions slash commission statistics and neighborhood relations were taken into account and four additional techniques were added to the test. Several methods outperformed manual processing in dominant, co-dominant and suppressed tree storesies. Taller trees generally had better location accuracy and the best methods achieved an accuracy of tree height greater than 0.5 m in all classes. For forest inventory, minimum curvature based tree detection with point cloud based cluster detection for suppressed trees is a promising solution for future research. This article was authored by Bernd Michael Wolf, Ji Cheng Wu, Xia Wei Yu, and others. We are article.tv, links in the description below.