A comparative study on point cloud down-sampling strategies for deep learning-based crop organ segmentation

Table 10 Quantitative comparison of both semantic segmentation and instance segmentation performances across five down-sampling strategies on the PlantNet network for the Soybean-MVS dataset

PlantNet	Semantic Segmentation (%)				Instance Segmentation (%)				Score	AveDiff (%)
PlantNet	Precision	Recall	F1	IoU	Cov	WCov	mPrec	mRec	Score	AveDiff (%)
FPS	86.79	87.30	87.03	78.04	45.06	51.52	51.27	27.08	19	1.37
RS	88.95	88.89	88.92	80.27	45.54	45.75	46.32	24.70	19	1.97
UVS	86.23	86.87	86.54	77.31	43.56	50.93	49.82	25.03	8	2.35
VFPS	86.80	86.55	86.69	77.61	38.95	46.35	45.41	19.89	3	4.60
3DEPS	89.61	89.36	89.48	81.26	45.37	46.21	44.36	24.70	23	1.84

ISSN: 1746-4811