Zaznavanje olesenele vegetacije s konvolucijsko nevronsko mrežo U-Net

Authors

  • Adam Gabrič Znanstvenoraziskovalni center Slovenske akademije znanosti in umetnosti, Inštitut za antropološke in prostorske študije, Novi trg 2, SI – 1000 Ljubljana in Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo, Jamova cesta 2, SI – 1000 Ljubljana https://orcid.org/0009-0008-4816-2881
  • Krištof Oštir Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo, Jamova cesta 2, SI – 1000 Ljubljana https://orcid.org/0000-0002-4887-7798
  • Žiga Kokalj Znanstvenoraziskovalni center Slovenske akademije znanosti in umetnosti, Inštitut za antropološke in prostorske študije, Novi trg 2, SI – 1000 Ljubljana https://orcid.org/0000-0003-1712-0351

DOI:

https://doi.org/10.3986/GV95205

Keywords:

nevronska mreža, olesenela vegetacija, ortofoto, model višin krošenj, klasifikacija, //, neural network, woody vegetation, orthophoto, canopy height model, classification

Abstract

S konvolucijsko nevronsko mrežo U-Net smo zaznavali posamezna drevesa, skupine dreves in grmovnic, drevorede, gozd, mejice, obvodno vegetacijo in drevesa v sadovnjakih, ki jih gradijo različne vrste olesenele vegetacije. Osnovni vhodni podatek je bil ortofoto, ki smo mu z namenom boljše klasifikacije dodajali dodatne sloje. Najboljši rezultat smo dobili s kombinacijo ortofota in modela višin krošenj – skupna natančnost klasifikacije je bila 93,3%. Ta kombinacija vhodnih podatkov je zvišala natančnost klasifikacije razredov gozd in ostalo, ki predstavljata daleč največji delež učnih vzorcev, medtem ko je bil rezultat za manj zastopane razrede (zlasti drevored) tudi pri tej klasifikaciji slabši.

Woody vegetation detection with the U-Net convolutional neural network

Single trees, groups of trees and bushes, trees in rows, forest, hedges, riparian vegetation and trees in orchards, all of which are composed by different woody vegetation species, were classified using convolutional neural network U-Net. National orthophoto represented basic input data, however adding further layers was tested with a view of improving classification. Best results were obtained using orthophoto and canopy height model combination, the overall accuracy reached 93.3%. This combination increased accuracies achieved at two biggest classes (classes forest and other). The model performed much worse with other classes, that were less represented in training data, which is especially true for trees in rows.

References

Ahlswede, S., Asam, S., Röder, A. 2021: Hedgerow object detection in very high-resolution satellite images using convolutional neural networks. Journal of Applied Remote Sensing 15-1. DOI: https://doi.org/10.1117/1.JRS.15.018501

Barlow, M. C., Zhu, X., Glennie, C. L. 2022: Stream boundary detection of a hyper-arid, polar region using a U-Net architecture, Taylor Valley, Antarctica. Remote Sensing 14-1. DOI: https://doi.org/10.3390/rs1401023410.3390/rs14010234

Barros, T., Conde, P., Gonçalves, G., Premebida, C., Monteiro, M., Ferreira, C. S. S., Nunes, U. J. 2022: Multispectral vineyard segmentation: A deep learning approach. Computers and Electronics in Agriculture 195. DOI: https://doi.org/10.1016/j.compag.2022.106782

Castillejo-González, I. 2018: Mapping of olive trees using pansharpened quickbird images: An evaluation of pixel- and object-based analyses. Agronomy 8-12. DOI: https://doi.org/10.3390/agronomy8120288

Chen, W., Xiang, H., Moriya, K. 2020. Individual tree position extraction and structural parameter retrieval based on airborne LiDAR Data: Performance evaluation and comparison of four algorithms. Remote Sensing 12-3. DOI: https://doi.org/10.3390/rs12030571

Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., Fei-Fei, L. 2009: ImageNet: A large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition. Miami. DOI: https://doi.org/10.1109/CVPR.2009.5206848

Dong, T., Shen, Y., Zhang, J., Ye, Y., Fan, J. 2019: Progressive cascaded convolutional neural networks for single tree detection with Google Earth Imagery. Remote Sensing 11-15. DOI: https://doi.org/10.3390/rs11151786

Duelli, P. 1997: Biodiversity evaluation in agricultural landscapes: An approach at two different scales. Agriculture, Ecosystems and Environment 62, 2-3. DOI: https://doi.org/10.1016/S0167-8809(96)01143-7

Estornell, J., Hadas, E., Martí, J., López-Cortés, I. 2021: Tree extraction and estimation of walnut structure parameters using airborne LiDAR data. International Journal of Applied Earth Observation and Geoinformation 96. DOI: https://doi.org/10.1016/j.jag.2020.102273

Feng, W., Sui, H., Huang, W., Xu, C., An, K. 2019: Water body extraction from very high-resolution remote sensing imagery using deep U-Net and a superpixel-based conditional random field model. IEEE Geoscience and Remote Sensing Letters 16-4. DOI: https://doi.org/10.1109/LGRS.2018.2879492

Freudenberg, M., Nölke, N., Agostini, A., Urban, K., Wörgötter, F., Kleinn, C. 2019: Large scale palm tree detection in high resolution satellite images using U-Net. Remote Sensing 11-3. DOI: https://doi.org/10.3390/rs11030312

Garcia-Pedrero, A., Lillo-Saavedra, M., Rodriguez-Esparragon, D., Gonzalo-Martin, C. 2019: Deep learning for automatic outlining agricultural parcels: Exploiting the land parcel identification system. IEEE Access 7. DOI: https://doi.org/10.1109/ACCESS.2019.2950371

Glušič, A., Ciglič, R., Čehovin Zajc, L. 2021: Zaznavanje terasiranih pokrajin kot semantična segmentacija digitalnega modela višin. Zbornik tridesete mednarodne Elektrotehniške in računalniške konference ERK 2021. Ljubljana.

Golobič, M., Penko Seidl, N., Lestan, K. A., Žerdin, M., Pačnik, L., Libnik, N., Vrbajnščak, M., Vrščaj, B., Kralj, T., Turk, B., Bergant, J., Šinkovec, M. 2015: Opredelitev krajinske pestrosti in krajinskih značilnosti, pomembnih za ohranjanje biotske raznovrstnosti. Ciljni raziskovalni program (CRP) »Zagotovimo si hrano za jutri 2011–2020«: končno poročilo. Oddelek za krajinsko arhitekturo Biotehniške fakultete Univerze v Ljubljani. Ljubljana.

Guirado, E., Tabik, S., Alcaraz-Segura, D., Cabello, J., Herrera, F. 2017: Deep-learning versus OBIA for scattered shrub detection with Google Earth Imagery, Ziziphus lotus as case study. Remote Sensing 9-12. DOI: https://doi.org/10.3390/rs9121220

Guo, H., Shi, Q., Marinoni, A., Du, B., Zhang, L. 2021: Deep building footprint update network: A semi-supervised method for updating existing building footprint from bi-temporal remote sensing images. Remote Sensing of Environment 264. DOI: https://doi.org/10.1016/j.rse.2021.112589

He, K., Zhang, X., Ren, S., Sun, J. 2015: Deep residual learning for image recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas. DOI: https://doi.org/10.1109/CVPR.2016.90

Hoeser, T., Bachofer, F., Kuenzer, C. 2020: Object detection and image segmentation with deep learning on Earth observation data: A review–Part II: Applications. Remote Sensing 12-18. DOI: https://doi.org/10.3390/rs12183053

Hoeser, T., Kuenzer, C. 2020: Object detection and image segmentation with deep learning on earth observation data: A review–Part I: Evolution and recent trends. Remote Sensing 12-10. DOI: https://doi.org/10.3390/rs12101667

Hyyppä, J., Inkinen, M. 1999: Detecting and estimating attributes for single trees using laser scanner. The Photogrammetric Journal of Finland 16-2.

Jones, E. G., Wong, S., Milton, A., Sclauzero, J., Whittenbury, H., McDonnell, M. D. 2020: The impact of pan-sharpening and spectral resolution on vineyard segmentation through machine learning. Remote Sensing 12-6. DOI: https://doi.org/10.3390/rs12060934

Khosravipour, A., Skidmore, A. K., Isenburg, M. 2016: Generating spike-free digital surface models using LiDAR raw point clouds: A new approach for forestry applications. International Journal of Applied Earth Observation and Geoinformation 52. DOI: https://doi.org/10.1016/j.jag.2016.06.005

Kokalj, Ž., Stančič, L., Kobler, A., Noumonvi, K. D. 2020: Testiranje možnosti in izvedba kartiranja krajinskih struktur, pomembnih za biotsko raznovrstnost in blaženje podnebnih sprememb z daljinskim zaznavanjem: končno poročilo. Elaborat, Znanstvenoraziskovalni center Slovenske akademije znanosti in umetnosti, Gozdarski inštitut Slovenije. Ljubljana.

Lampič, B., Kastelic, A. 2021: Prepoznavanje in evidentiranje mejic: Preverjanje različnih metod na pilotnem območju Ljubljanskega barja. Dela 56. DOI: https://doi.org/10.4312/dela.56.5-51

Lešnik, A. 2018: Življenje v mejicah. Miklavž na Dravskem polju.

Lisiewicz, M., Kamińska, A., Stereńczak, K. 2022: Recognition of specified errors of Individual Tree Detection methods based on Canopy Height Model. Remote Sensing Applications: Society and Environment 25. DOI: https://doi.org/10.1016/j.rsase.2021.100690

Marsetič, A., Kanjir, U. Klasifikacija pokrovnosti z uporabo globokega učenja na časovnih vrstah podatkov PlanetScope. Preteklost in prihodnost, GIS v Sloveniji 16. Ljubljana. DOI: https://doi.org/10.3986/9789610506683_25

Mazza, A., Sica, F., Rizzoli, P., Scarpa, G. 2019: TanDEM-X forest mapping using convolutional neural networks. Remote Sensing 11-24. DOI: https://doi.org/10.3390/rs11242980

Mihevc, A., Mihevc, R. 2021: Morphological characteristics and distribution of dolines in Slovenia: A study of a lidar-based doline map of Slovenia. Acta Carsologica 50-1. DOI: https://doi.org/10.3986/ac.v50i1.9462

Muhammad Ali, P., Faraj, R. 2014: Data Normalization and Standardization: A Technical Report. Machine Learning Technical Reports 2014-1. DOI: https://doi.org/10.13140/RG.2.2.28948.04489

Penko Seidl, N., Golobič, M. 2020: Quantitative assessment of agricultural landscape heterogeneity. Ecological Indicators 112. DOI: https://doi.org/10.1016/j.ecolind.2020.106115

Ronneberger, O., Fischer, P., Brox, T. 2015: U-Net: Convolutional networks for biomedical image segmentation. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. Cham. DOI: https://doi.org/10.1007/978-3-319-24574-4_28

Sadiq, R., Akhtar, Z., Imran, M., Ofli, F. 2022: Integrating remote sensing and social sensing for flood mapping. Remote Sensing Applications: Society and Environment 25. DOI: https://doi.org/10.1016/j.rsase.2022.100697

Safonova, A., Tabik, S., Alcaraz-Segura, D., Rubtsov, A., Maglinets, Y., Herrera, F. 2019: Detection of fir trees (Abies sibirica) damaged by the bark beetle in unmanned aerial vehicle images with deep learning. Remote Sensing 11-6. DOI: https://doi.org/10.3390/rs11060643

Sylvain, J.-D., Drolet, G., Brown, N. 2019: Mapping dead forest cover using a deep convolutional neural network and digital aerial photography. ISPRS Journal of Photogrammetry and Remote Sensing 156. DOI: https://doi.org/10.1016/j.isprsjprs.2019.07.010

Šumrada, T., Rac, I., Juvančič, L., Erjavec, E. 2020: Ohranjanje krajinskih značilnosti in njihovo vključevanje v ukrepe slovenske kmetijske politike. Geografski vestnik 92-1. DOI: https://doi.org/10.3986/GV92103

Weinstein, B. G., Marconi, S., Bohlman, S., Zare, A., White, E. 2019: Individual tree-crown detection in RGB imagery using semi-supervised deep learning neural networks. Remote Sensing 11-11. DOI: https://doi.org/10.3390/rs11111309

Yu, M., Rui, X., Xie, W., Xu, X., Wei, W. 2022: Research on automatic identification method of terraces on the Loess Plateau based on deep transfer learning. Remote Sensing 14-10. DOI: https://doi.org/10.3390/rs14102446

Zheng, J., Li, W., Xia, M., Dong, R., Fu, H., Yuan, S. 2019: Large-scale oil palm tree detection from high-resolution remote sensing images using faster-RCNN. 2019 IEEE International Geoscience and Remote Sensing Symposium. Yokohama. DOI: https://doi.org/10.1109/IGARSS.2019.8898360

Downloads

Published

2023-12-31

How to Cite

Gabrič, A., Oštir, K., & Kokalj, Žiga. (2023). Zaznavanje olesenele vegetacije s konvolucijsko nevronsko mrežo U-Net. Geografski Vestnik, 95(2), 113–134. https://doi.org/10.3986/GV95205

Issue

Section

Articles/članki