Abstract:
As worries about global warming, rising energy costs, and worsening air quality grow, this study investigates how deep learning models, specifically YOLOV5 and YOLOV8, can help with finding rooftops, estimating their sizes, and planning environmentally friendly infrastructure. The first part of the study talks about how important it is to find long-term answers to environmental problems, like installing solar panels on roofs and planting trees. Using drone footage and cutting-edge deep learning frameworks, the study carefully checks how well YOLOV5 and YOLOV8 work at different steps, such as collecting data, preprocessing it, adding to it, and choosing the best model. The results show that both models are good at finding rooftops. YOLOV5 has slightly better accuracy, recall, and mean average accuracy at the 50% Intersection over Union (IoU) threshold, while YOLOV8 is more stable across a wider range of IoU thresholds, as shown by its performance on the mAP50-95 metric. The results show that there are complex trade-offs between being accurate and being consistent. Overall YOLOV5 provides better accuracy in detecting and segmenting objects with 93% mAP(Mean Average Precision). They can help people who work with green energy, urban planning, and protecting the environment. This study helps us learn more about how deep learning can be used to solve important sustainability problems by showing us the pros and cons of YOLOV5 and YOLOV8 in the areas of rooftop recognition and planning eco-friendly infrastructure.
