CNN-based transfer learning for forest aboveground biomass prediction from ALS point cloud tomography

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F86652079%3A_____%2F24%3A00600527" target="_blank" >RIV/86652079:_____/24:00600527 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.tandfonline.com/doi/full/10.1080/22797254.2024.2396932" target="_blank" >https://www.tandfonline.com/doi/full/10.1080/22797254.2024.2396932</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1080/22797254.2024.2396932" target="_blank" >10.1080/22797254.2024.2396932</a>

Jazyk výsledku

angličtina

Název v původním jazyce

CNN-based transfer learning for forest aboveground biomass prediction from ALS point cloud tomography

Popis výsledku v původním jazyce

This study presents a new approach for predicting forest aboveground biomass (AGB) from airborne laser scanning (ALS) data: AGB is predicted from sequences of images depicting vertical cross-sections through the ALS point clouds. A 3D version of the VGG16 convolutional neural network (CNN) with initial weights transferred from pre-training on the ImageNet dataset was used. The approach was tested on datasets from Canada, Poland, and the Czech Republic. To analyse the effect of training sample size on model performance, different-sized samples ranging from 10 to 375 ground plots were used. The CNNs were compared with random forest models (RFs) trained on point cloud metrics. At the maximum number of training samples, the difference in RMSE between observed and predicted AGB of CNNs and RFs ranged from2 t/ha to 5 t/ha, and the difference in squared Pearson correlation coefficient ranged from0.05 to 0.06. Additional pre-training on synthetic data derived from virtual laser scanning of simulated forest stands could only improve the prediction performance of the CNNs when only a few real training samples (10-40) were available. While 3D CNNs trained on cross-section images derived from real data showed promising results, RFs remain a competitive alternative.

Název v anglickém jazyce

CNN-based transfer learning for forest aboveground biomass prediction from ALS point cloud tomography

Popis výsledku anglicky

This study presents a new approach for predicting forest aboveground biomass (AGB) from airborne laser scanning (ALS) data: AGB is predicted from sequences of images depicting vertical cross-sections through the ALS point clouds. A 3D version of the VGG16 convolutional neural network (CNN) with initial weights transferred from pre-training on the ImageNet dataset was used. The approach was tested on datasets from Canada, Poland, and the Czech Republic. To analyse the effect of training sample size on model performance, different-sized samples ranging from 10 to 375 ground plots were used. The CNNs were compared with random forest models (RFs) trained on point cloud metrics. At the maximum number of training samples, the difference in RMSE between observed and predicted AGB of CNNs and RFs ranged from2 t/ha to 5 t/ha, and the difference in squared Pearson correlation coefficient ranged from0.05 to 0.06. Additional pre-training on synthetic data derived from virtual laser scanning of simulated forest stands could only improve the prediction performance of the CNNs when only a few real training samples (10-40) were available. While 3D CNNs trained on cross-section images derived from real data showed promising results, RFs remain a competitive alternative.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20705 - Remote sensing

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

European Journal of Remote Sensing

ISSN

2279-7254

e-ISSN

2279-7254

Svazek periodika

57

Číslo periodika v rámci svazku

1

Stát vydavatele periodika

IT - Italská republika

Počet stran výsledku

19

Strana od-do

2396932

Kód UT WoS článku

001308185200001

EID výsledku v databázi Scopus

2-s2.0-85203310416