Multi-modal Sensor Data Resource of Canadian Critical Electrical Infrastructure

We collect a unique data resource from multiple sensor modalities for the purpose of training and evaluating algorithms for monitoring electrical critical infrastructure within the Canadian environment. Two different sensor modes were considered for inclusion in the data resource: colour images and 3D Light Detection And Ranging (LiDAR) sensors. This research examined these different sensors for their potential use in monitoring electrical critical infrastructure, such as poles, high-tension wires, and transformers. A multimodal data acquisition system was assembled using commercially available sensors. The acquisition system was deployed on a ground vehicle in the National Capital Region to collect multimodal data of power critical infrastructure along Canadian road corridors. Data was collected and registered in both time and space in order to enable multi-sensor fusion. Data Description: A total of 1.77 GB of data captured from total of 440 scenes, for each scenes we included images, Segmentation label and LiDAR data.The image data was annotated for semantic segmentation with five different classes. We have 440 scenes and for each of scene we have following data: |___Image |___Labels | |__NPY | |__PNG |____LiDAR