Accurate perception of transmission lines by unmanned aerial vehicles (UAVs) is a core foundation for power system operation and maintenance as well as UAV autonomous obstacle avoidance. However, in complex environments, factors such as light changes, background interference, and the inherent characteristics of transmission lines themselves pose significant challenges to accurate identification. Existing single-modal detection methods rely on visual or infrared data. Restricted by their poor adaptability to complex backgrounds, these methods perform poorly under harsh conditions and have obvious limitations. Although multi-modal detection, which fuses visual and infrared data, has improved robustness compared with single-modal methods, existing multi-modal approaches still have shortcomings in adapting to complex environments and mining task-specific characteristics. To address these issues, a transmission line detection scheme based on cross-modal interaction fusion and global feature calibration for visual and infrared data is proposed. Specifically, thecross-modal interaction guided fusion (CIGF) module enables deep interaction and complementary advantages of dual-modal features; the global feature significance modulator (GFSM) accurately calibrates pivotal features and enhances key information; and the multi-receptive enhanced decoder (MRED) efficiently reconstructs the fine spatial structure of transmission line targets and achieves pixel-level positioning. These three core modules work together to form a complete technical chain covering feature extraction, interaction fusion, global calibration, and fine decoding. Experiments on VITLD, the authoritative dataset for UAV transmission line detection, show that the proposed algorithm meets the dual requirements of detection accuracy and real-time performance. Notably, it maintains high accuracy even in complex and extreme environments such as low light at night, foggy blur, and snowy occlusion, breaking through the application bottleneck of traditional methods. This scheme provides an effective approach to solving the problem of UAV transmission line detection and holds important theoretical significance and practical application value.