Building Rich Interior Hazard Maps for Public Safety
Abstract
An accurate model of building interiors with detailed annotations is critical to protecting first responders and building occupants during emergencies. First responders and building occupants can use these 3D building models to navigate indoor environments or vacate the building safely. In collaboration with the City of Memphis, we have collected extensive LiDAR and video data from seven buildings in Memphis. We apply machine learning techniques to the video frames to detect and classify objects of interest to first responders. We then utilize data fusion methods on the LiDAR and image data to create a comprehensive colored 3D indoor point cloud dataset with labeled safety-related objects. This paper documents the challenges we encountered in data collection and processing, and it presents a complete 3D mapping and labeling system for the environments inside and adjacent to buildings. Moreover, we used two of the scanned buildings as a case study to illustrate our process and show detailed evaluation results. Our results show that the deep neural network Mask R-CNN with transfer learning and hard-negative mining performs well in labeling public-safety objects in our image dataset, especially for large objects.
Publication Title
Communications in Computer and Information Science
Recommended Citation
Hossain, M., Ma, T., Watson, T., Simmers, B., Khan, J., Jacobs, E., & Wang, L. (2022). Building Rich Interior Hazard Maps for Public Safety. Communications in Computer and Information Science, 1612 CCIS, 173-196. https://doi.org/10.1007/978-3-031-17098-0_9
