ESRI: Case Study with GIS

About the Project

This project consists of detecting HVAC systems within the Portland satellite images. Using ArcGIS Pro, we utilized the deep learning tool to train a model to detect HVAC systems from the images provided to us. Like any deep learning project, we initially started off by collecting samples. We manually identified and used the training samples to train the model. After roughly 2000 samples were collected, we exported the data as a deep learning package and used it to identify the HVACs in the remaining data provided to us. The accuracy of this model was 13.68%.