Enhancing Object Detection in Remote Sensing Images with GFA and ADS Modules Based on InternAware Model

To improve object detection accuracy and resolve large variations in object size in remote sensing images, this paper proposes the InternAware model based on InternImage-T. The GFA (Global Feature Aware) module is used to eliminate significant background information, enhancing the model's global perception of object regions and reducing complex background interference. Additionally, the ADS (Attention Down Sampling) module refines the feature map and transmits effective semantic information, boosting the model's encoding ability within the deep network. The model's detection performance is evaluated on the public UCAS AOD dataset. Experimental results show that the proposed model outperforms the baseline model, with an APL increase of 4.1% on the UCAS AOD dataset, confirming the effectiveness of the proposed approach.