This paper proposes a novel method for remote sensing micro-object detection, named GALDET, which integrates a Global and Local Attention Mechanism (GAL) to enhance feature extraction and detection accuracy. The method addresses the challenges of detecting small, densely packed targets in high-resolution remote sensing images by combining local detail analysis with global contextual information. The GAL module enables the model to focus on both fine-grained details and overall patterns, improving the detection of small objects. Additionally, a multi-head prediction module is introduced to leverage multi-scale features, and a novel Ziou loss function is developed to enhance small target localization precision. Experimental results on the Visdrone2019 and DOTA datasets show that GALDET outperforms traditional methods in detecting small targets, achieving significant improvements in Average Precision (AP) values. The method also demonstrates strong performance in real-time processing and handles complex scenarios such as occlusions and dense object distributions. The study highlights the effectiveness of the proposed GAL mechanism and the Ziou loss function in improving remote sensing object detection. The research contributes to the development of more accurate and efficient detection models for remote sensing applications.This paper proposes a novel method for remote sensing micro-object detection, named GALDET, which integrates a Global and Local Attention Mechanism (GAL) to enhance feature extraction and detection accuracy. The method addresses the challenges of detecting small, densely packed targets in high-resolution remote sensing images by combining local detail analysis with global contextual information. The GAL module enables the model to focus on both fine-grained details and overall patterns, improving the detection of small objects. Additionally, a multi-head prediction module is introduced to leverage multi-scale features, and a novel Ziou loss function is developed to enhance small target localization precision. Experimental results on the Visdrone2019 and DOTA datasets show that GALDET outperforms traditional methods in detecting small targets, achieving significant improvements in Average Precision (AP) values. The method also demonstrates strong performance in real-time processing and handles complex scenarios such as occlusions and dense object distributions. The study highlights the effectiveness of the proposed GAL mechanism and the Ziou loss function in improving remote sensing object detection. The research contributes to the development of more accurate and efficient detection models for remote sensing applications.