• Laser & Optoelectronics Progress
  • Vol. 57, Issue 20, 201012 (2020)
Bin Zou1、2, Siyang Lin1、*, and Zhishuai Yin1、2
Author Affiliations
  • 1Hubei Key Laboratory of Advanced Technology for Automotive Components, Wuhan University of Technology, Wuhan, Hubei 430070, China
  • 2Hubei Collaborative Innovation Center for Automotive Components Technology, Wuhan, Hubei 430070, China
  • show less
    DOI: 10.3788/LOP57.201012 Cite this Article Set citation alerts
    Bin Zou, Siyang Lin, Zhishuai Yin. Semantic Mapping Based on YOLOv3 and Visual SLAM[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201012 Copy Citation Text show less
    Flow chart of YOLOv3 algorithm
    Fig. 1. Flow chart of YOLOv3 algorithm
    Structure of YOLOv3
    Fig. 2. Structure of YOLOv3
    Supervoxel clustering
    Fig. 3. Supervoxel clustering
    Graph segmentation of surface patches
    Fig. 4. Graph segmentation of surface patches
    Object information label. (a) Original point cloud; (b) object detection; (c) semantic label
    Fig. 5. Object information label. (a) Original point cloud; (b) object detection; (c) semantic label
    Optimization of ORB characteristic point extraction. (a) Original ORB characteristic point extraction; (b) optical flow tracking ORB characteristic point
    Fig. 6. Optimization of ORB characteristic point extraction. (a) Original ORB characteristic point extraction; (b) optical flow tracking ORB characteristic point
    Comparison of point cloud segmentation results. (a) Algorithm 1; (b) algorithm 2; (c) proposed algorithm
    Fig. 7. Comparison of point cloud segmentation results. (a) Algorithm 1; (b) algorithm 2; (c) proposed algorithm
    Verifying semantic map. (a) System operating in real-time; (b) original map; (c) semantic map
    Fig. 8. Verifying semantic map. (a) System operating in real-time; (b) original map; (c) semantic map
    AlgorithmRunning time /ms
    Algorithm 118073
    Algorithm 29825
    Proposed algorithm10531
    Table 1. Running time of different algorithms
    Algorithmfreiburg3_sittingfreiburg3_long_office
    Number of pointsMap size /MbitTime /sNumber of pointsMap size /MbitTime /s
    Proposed algorithm210654573.50.08715467152151.30.1318
    ORB with YOLOv3450307191.40.129598408731900.1841
    ORB with Mask R-CNN[19]438223286.30.17349165934177.80.2506
    Table 2. Comparison of experimental results of map construction
    Bin Zou, Siyang Lin, Zhishuai Yin. Semantic Mapping Based on YOLOv3 and Visual SLAM[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201012
    Download Citation