MISD-SLAM: Multimodal Semantic SLAM for Dynamic Environments

Table 4

Time evaluation of MISD-SLAM in dynamic sequences of TUM RGB-D datasets (ms). The results of each row are the running time of corresponding modules in each sequence. The last row is the average time of above sequences in corresponding modules.

Sequence	ORB feature extraction (ms)	Instance segmentation (ms)	Multiview geometry (ms)	Semantic map construction (ms)
Sitting_static	12.981	36.465	144.085	121.388
Walking_static	14.575	36.109	205.484	296.667
Walking_xyz	14.146	36.150	180.533	338.903
Walking_rpy	12.345	36.640	200.591	274.928
Walking_halfsphere	12.484	36.536	161.401	244.241
Average time	13.306	36.380	178.419	255.225