Integrated Approach for Heat Envelope Identification and Energy Efficiency Analysis in Buildings Using Drone Thermal Imagery and Deep Learning Techniques

In this paper, we present an automated approach that leverages deep learning techniques and drone thermal imagery to quantify heat envelopes and estimate the total heat loss over a designated period. The proposed approach aims to significantly accelerate the assessment of multiple buildings to a short timeframe compared to manual energy auditing. An infrared (IR) camera-equipped drone is deployed to capture high-resolution thermal and visible band images of buildings. A model based on darknet deep learning (DL) framework, you only look once (YOLO), was developed. This model processes the high-resolution thermal images, extracts feature maps, and identifies the heat envelopes in the buildings. Additionally, a user-friendly application to extract temperature values at points of interest was created. By utilizing the extracted temperature data, we compute an estimate of the total heat loss. This automated approach provides valuable insights and a deeper understanding of energy consumption, enabling more informed decision-making in energy management.