Construction Safety Inspection Based on Target Detection and Adaptive Algorithm A* with Variable Weight

This paper proposes a safety inspection technology for drones and target detection to solve the difficulty of construction safety inspection and the ineffectiveness of safety identification methods in offshore photovoltaic construction projects. The researchers made lightweight improvements to the YOLOv5s recognition algorithm, especially optimizing it to preserve fine-grained information. We use adaptive algorithm A* with variable weight for path planning and B-spline for path smoothing to optimize the inspection path. The data involving construction safety was collected extensively in the study, and the accuracy of small target detection reached 95.9% using the method proposed in this paper. In addition, it implements efficient path planning based on a small target detection algorithm to achieve smarter obstacle avoidance. Empirical studies have shown that the proposed technical scheme can significantly improve the efficiency and accuracy of construction safety inspection. In summary, it will ensure the safe operation of marine solar power generation projects and aid in safety management and monitoring in similar scenarios.