AOMC: An Adaptive Point Cloud Clustering Approach for Feature Extraction

Table 4

Comparison of the map on the classification task. The best and the second-best methods are highlighted in red and green colors, respectively.

Method	Mean intersection over union (mIoU)
	Avg (%)	Airplane (%)	Bag (%)	Chair (%)	Laptop (%)	Knife (%)	Mug (%)	Rocket (%)	Cap (%)	Table (%)	Earphone (%)
Farthest selection	81.60	82.40	81.10	90.70	90.60	86.10	95.10	54.50	89.90	82	68.50
Random selection	78.80	80.30	80	91	89	83.10	90.10	51.40	87.70	73	62
DBSCAN	79.70	79.20	78	89	90	83	95	56	82	80	65
k-means (k = 10)	80.70	83.10	74	88	91	84	87	57	90.10	81	72
Agglomerative clustering	79.30	81.40	73	83	84	83.40	91	59	88.40	80.40	69.40
Ours	83	84	79	91.70	89.40	89.90	94.30	61	83	84	74