Abstract: As a crucial component of the power system, transmission lines play a vital role in ensuring the safety and stability of electricity supply. However, existing deep learning-based transmission line detection methods face significant challenges such as excessive network parameters, high computational complexity, and slow inference speed. To address these issues, this paper proposes a knowledge-driven efficient and lightweight transmission line detection network. Specifically, ResNet-50 is adopted as the teacher network and ResNet-18 as the student network to construct a lightweight detection framework. Spatial and channel attention masks are introduced to guide the student network in learning key features from the teacher, thereby achieving efficient knowledge transfer. Experimental results on the PTL-AI Furnas dataset demonstrate that the proposed method significantly improves detection accuracy compared with the baseline student model, while maintaining fewer parameters and faster inference. The proposed approach achieves a superior balance between performance and efficiency, offering a promising solution for intelligent inspection of power transmission lines.