{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,4]],"date-time":"2026-04-04T18:15:03Z","timestamp":1775326503771,"version":"3.50.1"},"reference-count":39,"publisher":"MDPI AG","issue":"11","license":[{"start":{"date-parts":[[2019,6,1]],"date-time":"2019-06-01T00:00:00Z","timestamp":1559347200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000936","name":"Gordon and Betty Moore Foundation","doi-asserted-by":"publisher","award":["GBMF4563"],"award-info":[{"award-number":["GBMF4563"]}],"id":[{"id":"10.13039\/100000936","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Remote Sensing"],"abstract":"<jats:p>Remote sensing can transform the speed, scale, and cost of biodiversity and forestry surveys. Data acquisition currently outpaces the ability to identify individual organisms in high resolution imagery. We outline an approach for identifying tree-crowns in RGB imagery while using a semi-supervised deep learning detection network. Individual crown delineation has been a long-standing challenge in remote sensing and available algorithms produce mixed results. We show that deep learning models can leverage existing Light Detection and Ranging (LIDAR)-based unsupervised delineation to generate trees that are used for training an initial RGB crown detection model. Despite limitations in the original unsupervised detection approach, this noisy training data may contain information from which the neural network can learn initial tree features. We then refine the initial model using a small number of higher-quality hand-annotated RGB images. We validate our proposed approach while using an open-canopy site in the National Ecological Observation Network. Our results show that a model using 434,551 self-generated trees with the addition of 2848 hand-annotated trees yields accurate predictions in natural landscapes. Using an intersection-over-union threshold of 0.5, the full model had an average tree crown recall of 0.69, with a precision of 0.61 for the visually-annotated data. The model had an average tree detection rate of 0.82 for the field collected stems. The addition of a small number of hand-annotated trees improved the performance over the initial self-supervised model. This semi-supervised deep learning approach demonstrates that remote sensing can overcome a lack of labeled training data by generating noisy data for initial training using unsupervised methods and retraining the resulting models with high quality labeled data.<\/jats:p>","DOI":"10.3390\/rs11111309","type":"journal-article","created":{"date-parts":[[2019,6,3]],"date-time":"2019-06-03T02:08:40Z","timestamp":1559527720000},"page":"1309","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":209,"title":["Individual Tree-Crown Detection in RGB Imagery Using Semi-Supervised Deep Learning Neural Networks"],"prefix":"10.3390","volume":"11","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-2176-7935","authenticated-orcid":false,"given":"Ben G.","family":"Weinstein","sequence":"first","affiliation":[{"name":"Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, FL 32603, USA"}]},{"given":"Sergio","family":"Marconi","sequence":"additional","affiliation":[{"name":"Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, FL 32603, USA"}]},{"given":"Stephanie","family":"Bohlman","sequence":"additional","affiliation":[{"name":"School of Forest Resources and Conservation, University of Florida, Gainesville, FL 32603, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4847-7604","authenticated-orcid":false,"given":"Alina","family":"Zare","sequence":"additional","affiliation":[{"name":"Department of Electrical and Computer Engineering, University of Florida, Gainesville, FL 32601, USA"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6728-7745","authenticated-orcid":false,"given":"Ethan","family":"White","sequence":"additional","affiliation":[{"name":"Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, FL 32603, USA"}]}],"member":"1968","published-online":{"date-parts":[[2019,6,1]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1572","DOI":"10.1111\/ele.13106","article-title":"Biodiversity monitoring, earth observations and the ecology of scale","volume":"21","author":"Anderson","year":"2018","journal-title":"Ecol. Lett."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"533","DOI":"10.1111\/1365-2656.12780","article-title":"A computer vision for animal ecology","volume":"87","author":"Weinstein","year":"2018","journal-title":"J. Animal Ecol."},{"key":"ref_3","first-page":"82","article-title":"Individual tree crown delineation using localized contour tree method and airborne LiDAR data in coniferous forests","volume":"52","author":"Wu","year":"2016","journal-title":"Int. J. Appl. Earth Obs. Geoinf."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"2369","DOI":"10.1002\/eap.1436","article-title":"A hyperspectral image can predict tropical tree growth rates in single-species stands","volume":"26","author":"Caughlin","year":"2016","journal-title":"Ecol. Appl."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"16","DOI":"10.1080\/07038992.2017.1252907","article-title":"Layer Stacking: A Novel Algorithm for Individual Forest Tree Segmentation from LiDAR Point Clouds","volume":"43","author":"Ayrey","year":"2017","journal-title":"Can. J. Remote Sens."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"34","DOI":"10.1016\/j.isprsjprs.2015.10.002","article-title":"A novel transferable individual tree crown delineation model based on Fishing Net Dragging and boundary classification","volume":"110","author":"Liu","year":"2015","journal-title":"ISPRS J. Photogramm. Sens."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"1287","DOI":"10.14358\/PERS.72.11.1287","article-title":"The Individual Tree Crown Approach Applied to Ikonos Images of a Coniferous Plantation Area","volume":"72","author":"Gougeon","year":"2006","journal-title":"Photogramm. Eng. Sens."},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Weinmann, M., Weinmann, M., Mallet, C., and Br\u00e9dif, M. (2017). A Classification-Segmentation Framework for the Detection of Individual Trees in Dense MMS Point Cloud Data Acquired in Urban Areas. Remote. Sens., 9.","DOI":"10.3390\/rs9030277"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"184","DOI":"10.1016\/j.rse.2018.04.002","article-title":"Individual tree crown detection in sub-meter satellite imagery using Marked Point Processes and a geometrical-optical model","volume":"211","author":"Gomes","year":"2018","journal-title":"Remote. Sens. Environ."},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Li, W., Fu, H., Yu, L., and Cracknell, A. (2016). Deep Learning Based Oil Palm Tree Detection and Counting for High-Resolution Remote Sensing Images. Remote. Sens., 9.","DOI":"10.3390\/rs9010022"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Guirado, E., Tabik, S., Alcaraz-Segura, D., Cabello, J., and Herrera, F. (2017). Deep-learning Versus OBIA for Scattered Shrub Detection with Google Earth Imagery: Ziziphus lotus as Case Study. Remote. Sens., 9.","DOI":"10.3390\/rs9121220"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Ayrey, E., and Hayes, D.J. (2018). The Use of Three-Dimensional Convolutional Neural Networks to Interpret LiDAR for Forest Inventory. Remote. Sens., 10.","DOI":"10.3390\/rs10040649"},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"8","DOI":"10.1109\/MGRS.2017.2762307","article-title":"Deep Learning in Remote Sensing: A Comprehensive Review and List of Resources","volume":"5","author":"Zhu","year":"2017","journal-title":"IEEE Geosci. Remote Sens. Mag."},{"key":"ref_14","unstructured":"Dahlkamp, H., Kaehler, A., Stavens, D., Thrun, S., and Bradski, G. Self-supervised Monocular Road Detection in Desert Terrain. Robotics: Science and Systems II, Available online: https:\/\/tinyurl.com\/y6xtjqfa."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"1259","DOI":"10.1109\/TIP.2017.2772836","article-title":"Semi-Supervised Deep Learning Using Pseudo Labels for Hyperspectral Image Classification","volume":"27","author":"Wu","year":"2018","journal-title":"IEEE Trans. Image Process."},{"key":"ref_16","unstructured":"Romero, A., Ballas, N., Kahou, S.E., Chassang, A., Gatta, C., and Bengio, Y. (2014). FitNets: Hints for Thin Deep Nets. arXiv preprint, 1\u201313."},{"key":"ref_17","unstructured":"Erhan, D., Manzagol, P.-A., Bengio, Y., Bengio, S., and Vincent, P. (2009, January 16\u201318). The Difficulty of Training Deep Architectures and the Effect of Unsupervised Pre-Training. Proceedings of the Twelth International Conference on Artificial Intelligence and Statistics (AISTATS), Clearwater Beach, FL, USA."},{"key":"ref_18","unstructured":"Weinstein, B., and White, E. (2019, June 01). Weecology\/DeepLidar: Resubmission II, Version 3.0. Available online: http:\/\/doi.org\/10.5281\/zenodo.3066235."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"1236","DOI":"10.1111\/2041-210X.12575","article-title":"Tree-centric mapping of forest carbon density from airborne laser scanning and hyperspectral data","volume":"7","author":"Dalponte","year":"2016","journal-title":"Methods Ecol. Evol."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"75","DOI":"10.14358\/PERS.78.1.75","article-title":"A New Method for Segmenting Individual Trees from the Lidar Point Cloud","volume":"78","author":"Li","year":"2012","journal-title":"Photogramm. Eng. Sens."},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"554","DOI":"10.1080\/07038992.2016.1196582","article-title":"Imputation of Individual Longleaf Pine (Pinus palustris Mill.) Tree Attributes from Field and LiDAR Data","volume":"42","author":"Silva","year":"2016","journal-title":"Can. J. Remote Sens."},{"key":"ref_22","unstructured":"Roussel, J.-R., Auty, D., De Boissieu, F., and Meador, A.S. (2019, June 01). lidR: Airborne LiDAR Data Manipulation and Visualization for Forestry Applications. Available online: https:\/\/rdrr.io\/cran\/lidR\/."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"154","DOI":"10.1007\/s11263-013-0620-5","article-title":"Selective Search for Object Recognition","volume":"104","author":"Uijlings","year":"2013","journal-title":"Int. J. Comput. Vis."},{"key":"ref_24","first-page":"91","article-title":"Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks","volume":"1","author":"Ren","year":"2015","journal-title":"Nips"},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Lin, T.Y., Goyal, P., Girshick, R., He, K., and Dollar, P. (2017, January 22\u201329). Focal Loss for Dense Object Detection. Proceedings of the IEEE International Conference Computer Vision, Venice, Italy.","DOI":"10.1109\/ICCV.2017.324"},{"key":"ref_26","unstructured":"Gaiser, H., de Vries, M., Lacatusu, V., Williamson, A., and Liscio, E.D.D. (2019, June 01). fizy-r\/Keras-retinanet 2018. Available online: https:\/\/github.com\/fizyr\/keras-retinanet."},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27\u201330). Deep Residual Learning for Image Recognition. Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA.","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"73","DOI":"10.1016\/j.isprsjprs.2011.10.006","article-title":"Combination of individual tree detection and area-based approach in imputation of forest variables using airborne laser data","volume":"67","author":"Vastaranta","year":"2012","journal-title":"ISPRS J. Photogramm. Sens."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"378","DOI":"10.1016\/j.rse.2013.07.044","article-title":"An efficient, multi-layered crown delineation algorithm for mapping individual tree structure across multiple ecosystems","volume":"154","author":"Duncanson","year":"2014","journal-title":"Remote. Sens. Environ."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1016\/j.rse.2017.03.017","article-title":"Area-based vs tree-centric approaches to mapping forest carbon in Southeast Asian forests from airborne laser scanning data","volume":"194","author":"Coomes","year":"2017","journal-title":"Remote. Sens. Environ."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"34","DOI":"10.1016\/j.rse.2018.12.034","article-title":"Individual mangrove tree measurement using UAV-based LiDAR data: Possibilities and challenges","volume":"223","author":"Yin","year":"2019","journal-title":"Remote. Sens. Environ."},{"key":"ref_32","first-page":"336","article-title":"Applying LiDAR Individual Tree Detection to Management of Structurally Diverse Forest Landscapes","volume":"116","author":"Kane","year":"2018","journal-title":"J. For."},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Zhou, Y., and Tuzel, O. (2017). VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection.","DOI":"10.1109\/CVPR.2018.00472"},{"key":"ref_34","unstructured":"Qi, C.R., Su, H., Mo, K., and Guibas, L.J. (2017, January 21\u201326). PointNet: Deep learning on point sets for 3D classification and segmentation. Proceedings of the 30th IEEE Conference Computer Vision Pattern Recognition, CVPR 2017, Honolulu, HI, USA."},{"key":"ref_35","doi-asserted-by":"crossref","unstructured":"He, K., Gkioxari, G., Dollar, P., and Girshick, R. (2017, January 22\u201329). Mask R-CNN. Proceedings of the IEEE International Conference Computer Vision 2017, Venice, Italy.","DOI":"10.1109\/ICCV.2017.322"},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"7619","DOI":"10.1109\/TGRS.2014.2315649","article-title":"Evaluating tree detection and segmentation routines on very high resolution UAV LiDAR ata","volume":"52","author":"Wallace","year":"2014","journal-title":"IEEE Trans. Geosci. Remote Sens."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"27","DOI":"10.1016\/j.foreco.2005.09.021","article-title":"Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities","volume":"221","author":"Wulder","year":"2006","journal-title":"Ecol. Manag."},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"294","DOI":"10.1016\/j.rse.2015.08.011","article-title":"Mapping post-fire habitat characteristics through the fusion of remote sensing tools","volume":"173","author":"Vogeler","year":"2016","journal-title":"Remote. Sens. Environ."},{"key":"ref_39","doi-asserted-by":"crossref","unstructured":"Deng, S., Katoh, M., Yu, X., Hyypp\u00e4, J., and Gao, T. (2016). Comparison of Tree Species Classifications at the Individual Tree Level by Combining ALS Data and RGB Images Using Different Algorithms. Remote. Sens., 8.","DOI":"10.3390\/rs8121034"}],"container-title":["Remote Sensing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2072-4292\/11\/11\/1309\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,11]],"date-time":"2025-10-11T12:55:20Z","timestamp":1760187320000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2072-4292\/11\/11\/1309"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,6,1]]},"references-count":39,"journal-issue":{"issue":"11","published-online":{"date-parts":[[2019,6]]}},"alternative-id":["rs11111309"],"URL":"https:\/\/doi.org\/10.3390\/rs11111309","relation":{"has-preprint":[{"id-type":"doi","id":"10.1101\/532952","asserted-by":"object"}]},"ISSN":["2072-4292"],"issn-type":[{"value":"2072-4292","type":"electronic"}],"subject":[],"published":{"date-parts":[[2019,6,1]]}}}