Hopp til hovedinnholdet

Publications

NIBIOs employees contribute to several hundred scientific articles and research reports every year. You can browse or search in our collection which contains references and links to these publications as well as other research and dissemination activities. The collection is continously updated with new and historical material.

2019

Abstract

Short-day (SD) treatment is used by forest nurseries to induce growth cessation in Picea abies seedlings. SD treatment may however increase the risk of reflushing in autumn and earlier bud break the following spring. When the start of the SD treatment is early in order to control seedling height, the duration of the SD treatment should be longer in order to prevent reflushing in autumn. However, due to the amount of manual work involved in the short-day treatment, increasing the number of days is undesirable from a practical point of view. Splitting the SD treatment could be a way to achieve both early height control and at the same time avoid autumn bud break with less workload. We tested how different starting dates and durations of SD treatment influenced on morphological and phenological traits. Regardless of timing and duration of the SD treatment, height growth was reduced compared to the untreated controls. Seedlings given split SD (7+7 days interrupted with two weeks in long days) had less height growth than all other treatments. Root collar diameter growth was significantly less in control seedlings than in seedlings exposed to early (7 or 14 days) or split (7+7 days) SD treatment. There were also differences in the frequency of reflushing and bud break timing among the SD treated seedlings, dependent on duration and starting date. If the SD treatment started early, a continuous 14-day SD treatment was not sufficient to avoid high frequencies of reflushing. However, by splitting the SD treatment into two periods of 7+7 days these negative effects were largely avoided, although spring bud break occurred earlier than in the controls.

To document

Abstract

Advances in techniques for automated classification of point cloud data introduce great opportunities for many new and existing applications. However, with a limited number of labelled points, automated classification by a machine learning model is prone to overfitting and poor generalization. The present paper addresses this problem by inducing controlled noise (on a trained model) generated by invoking conditional random field similarity penalties using nearby features. The method is called Atrous XCRF and works by forcing a trained model to respect the similarity penalties provided by unlabeled data. In a benchmark study carried out using the ISPRS 3D labeling dataset, our technique achieves 85.0% in term of overall accuracy, and 71.1% in term of F1 score. The result is on par with the current best model for the benchmark dataset and has the highest value in term of F1 score. Additionally, transfer learning using the Bergen 2018 dataset, without model retraining, was also performed. Even though our proposal provides a consistent 3% improvement in term of accuracy, more work still needs to be done to alleviate the generalization problem on the domain adaptation and the transfer learning field.