Researchers have introduced RDNet, a dynamic adaptive salient object detection network designed to overcome the challenges of detecting objects in optical remote sensing images. The network addresses the limitations of traditional convolutional neural networks (CNNs) by incorporating a region proportion-aware mechanism, allowing it to capture global context and long-range dependencies more effectively. This approach enables RDNet to adapt to diverse object scales and reduce computational costs associated with self-attention mechanisms. By improving the accuracy and efficiency of salient object detection, RDNet has significant implications for applications such as geographical mapping and environmental monitoring1. The development of RDNet highlights the importance of advancing computer vision techniques to address the unique challenges of remote sensing image analysis, which is critical for various fields, including geography, ecology, and national security. The ability to accurately detect and analyze objects in remote sensing images can have far-reaching consequences, making RDNet a notable contribution to the field.