In the present study, we address the urgent need for an accurate assessment of building damage in the aftermath of the recent earthquake in Marrakech, Morocco. Assessing the damage is necessary to determine the impact on buildings since they bear the brunt of most natural disasters. The aim of this study is to enhance the efficiency of detecting house damage by utilizing UAV images and advanced deep-learning algorithms. The visual surface characteristics were used to categorize the impact of the post-earthquake UAV imagery into multiple classes using the state-of-the-art SOTA model, YOLOv9. The methodology employed involved preparing an annotated dataset of 337 images and dividing it into training (70%), validation (20%), and test sets (10%). The model was set up with pre-trained weights, fine-tuned with the SGD optimizer, and augmented for dataset diversity. This approach effectively detected the damage within buildings, with evaluation metrics showing a precision of 0.913, recall rate of 0.7604, mean average precision at 50%(mAP50) of 0.834, and a mean average precision from 50% to 95%(mAP50-95) of 0.543. The experiment outcomes have demonstrated the ability of YOLOv9 in detecting structural damage in buildings.