SHWD-YOLOv8: an enhanced YOLOv8 detector for resource efficient safety helmet wear identification in substation

To address the high deployment costs caused by the existing safety helmet detection algorithm for platform resources, a helmet-wearing detector based on improved YOLOv8 for substation environments, named SHWD-YOLOv8, further improves the detection performance. By combining MobileNetv3 with YOLOv8 and incorporating an efficient enhanced feature pyramid network along with a parameter-free attention mechanism (SimAM), the algorithm significantly reduces the number of parameters and computational load of the backbone network. The improved feature pyramid network enhances the network’s sensitivity to important features, while SimAM reduces the model’s parameter count and improves the detection accuracy for small safety helmets. The SHWD-YOLOv8 algorithm is characterized by its low computational requirements and minimal parameter count, posing shallow demands on the execution platform’s resources. The optimized model size is reduced to 4.65MB, greatly lowering the costs of network deployment and application, and expanding its application scope in real scenarios. Future research will continue to improve the model’s precision and explore the possibilities of deployment on hardware devices, making the safety helmet detection system more convenient and efficient in practical applications.