3D Semantic VSLAM of Dynamic Environment Based on YOLACT

Algorithm 2

Update Algorithm for the construction of semantic maps.

	Input: RGB-D data stream from the camera.
	Output: Semantic map.
(1)	Pre-selection .
(2)	Use Algorithm 1 for Keyframe Processing
(3)	Use optimizing to obtain
	.
(4)	Use YOLACT to process to obtain 2D semantic information
	.
(5)	Mapping 2D semantic information to 3D data
	.
(6)	if then:
(7)	(1) Update database .
(8)	(2) Building a semantic map .
(9)	else:
(10)	(1) Calculate the shortest distance between the nearest object
	.
(11)	(2) ifthen:
(12)	Add the object to the semantic map
	.
(13)	(3) else:
(14)	Optimize the semantic map .