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

Table 4 Quantitative comparison of both semantic segmentation and instance segmentation performances across five down-sampling strategies on the ASIS network

ASIS	Semantic Segmentation (%)				Instance Segmentation (%)				Score	AveDiff (%)
ASIS	Precision	Recall	F1	IoU	Cov	WCov	mPrec	mRec	Score	AveDiff (%)
FPS	79.77	81.44	79.62	67.24	63.46	70.37	69.58	58.41	14	1.96
RS	82.67	82.09	82.04	70.28	62.85	70.34	67.73	53.56	13	1.88
UVS	83.97	81.09	81.93	70.18	61.29	69.10	68.73	52.15	3	2.14
VFPS	84.33	81.03	82.08	70.66	62.20	69.88	67.90	51.73	13	1.92
3DEPS	83.33	82.48	82.34	70.24	64.01	71.92	71.45	54.62	29	0.64

ISSN: 1746-4811