Automatic segmentation of overlapped poplar seedling leaves combining Mask R-CNN and DBSCAN

•Combination of Mask R-CNN and DBSCAN segments single leaves accurately.•A new approach to fuse color and depth features in the FPN structure is proposed.•Manifold distance is better than Euclidean distance for segmenting curled leaves.•Pixel-wise IoU evaluates the results more accurately than tradi...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	Computers and electronics in agriculture Jg. 178; S. 105753
Hauptverfasser:	Liu, Xuan, Hu, Chunhua, Li, Pingping
Format:	Journal Article
Sprache:	Englisch
Veröffentlicht:	Amsterdam Elsevier B.V 01.11.2020 Elsevier BV
Schlagworte:	agriculture Algorithms cameras Clustering Contact stresses electronics Feature extraction Feature fusion Genotype & phenotype Heavy metals Leaf segmentation Leaves Manifold distance Manifolds Mask R-CNN Performance evaluation phenotype Poplar Populus RGB-D camera seedlings Segmentation Segments Feature fusion Leaf segmentation RGB-D camera Manifold distance Mask R-CNN
ISSN:	0168-1699, 1872-7107
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Schreiben Sie den ersten Kommentar!