As global climate continues to change, the frequent incidents of ice accretion on transmission lines pose an increasingly significant threat to the power transmission systems. Precise monitoring of such ice formations is essential for ensuring the safety and stability of the electrical grid. Conventionally, monitoring is performed by measuring tension and calculating the equivalent ice thickness with established formulas. However, these tension sensors can become unreliable or even malfunction over time due to creep and other factors, rendering traditional methods ineffective. This study introduces an innovative monitoring method for transmission lines, utilizing the cutting-edge image segmentation technology YoloV8-Seg. This approach combines high-resolution imagery with a sophisticated deep learning model to perform pixel-level segmentation of transmission lines, enabling precise ice thickness measurements through AI analysis. This method facilitates swift and accurate detection and measurement of ice accretion, significantly enhancing monitoring efficiency and accuracy. The experimental results demonstrate that this new technique substantially outperforms traditional methods, providing a novel technical solution for the mitigation and management of ice-related hazards in power systems, and offering considerable potential for widespread application and future development.