Skip to main content
. 2020 Sep 12;20(18):5202. doi: 10.3390/s20185202

Figure 4.

Figure 4

Software architecture. The capture module is implemented in C++ and provides aligned depth and color images. The color image is used by the semantic segmentation module, coded in PyTorch, to generate pixel-wise object class labels. The post-processing module combines the depth field and the label information to provide alerts only for close persons. Communication is handled by the Robotic Operating System.