FFA-GAN: A Generative Adversarial Network Based on Feature Fusion Attention for Intelligent Safety Monitoring

With the rapid development of national power grid, there is an increasing and strict demand for accurate intelligent management. However, current detection algorithms have limited abilities under adverse conditions, especially in regions like Yunnan Province with complex terrain. To address this issue, we propose a method that utilizes infrared visible images to make more informative, thereby improving accuracy algorithm electric construction site safety. First, design channel attention (CA) module pixel (PA) focus on important channels resist thick haze pixels information. Furthermore, two-stage discriminator which imposes two restrictions fused results. Finally, conduct large number comparison experiments state-of-the-art methods, results show our proposed fusion achieves excellent performance image fusion. This has good prospects application safety supervision sites provides line defense workers.