AG-YOLO is a rapid citrus fruit detection algorithm that integrates global context fusion to enhance detection accuracy in occluded scenarios. The algorithm is based on the NextViT architecture, which excels in capturing global contextual information. It introduces a Global Context Fusion Module (GCFM) that uses self-attention mechanisms to fuse local and global features, improving the model's ability to detect occluded targets. A dataset of over 8000 outdoor images was collected, including scenarios of occlusion, severe occlusion, overlap, and severe overlap. AG-YOLO achieved a precision of 90.6%, mAP@50 of 83.2%, and mAP@50:95 of 60.3%, outperforming existing object detection methods. The model runs at 34.22 FPS, demonstrating high detection accuracy and speed. AG-YOLO effectively addresses severe occlusion challenges, offering improved performance in complex environments. The study highlights the effectiveness of AG-YOLO in citrus detection, with contributions including the GCFM and NextViT backbone. The model's performance was validated through experiments, showing superior results in precision, recall, and mAP metrics. AG-YOLO's design enhances robustness and accuracy in detecting occluded citrus fruits, making it a reliable solution for agricultural applications.AG-YOLO is a rapid citrus fruit detection algorithm that integrates global context fusion to enhance detection accuracy in occluded scenarios. The algorithm is based on the NextViT architecture, which excels in capturing global contextual information. It introduces a Global Context Fusion Module (GCFM) that uses self-attention mechanisms to fuse local and global features, improving the model's ability to detect occluded targets. A dataset of over 8000 outdoor images was collected, including scenarios of occlusion, severe occlusion, overlap, and severe overlap. AG-YOLO achieved a precision of 90.6%, mAP@50 of 83.2%, and mAP@50:95 of 60.3%, outperforming existing object detection methods. The model runs at 34.22 FPS, demonstrating high detection accuracy and speed. AG-YOLO effectively addresses severe occlusion challenges, offering improved performance in complex environments. The study highlights the effectiveness of AG-YOLO in citrus detection, with contributions including the GCFM and NextViT backbone. The model's performance was validated through experiments, showing superior results in precision, recall, and mAP metrics. AG-YOLO's design enhances robustness and accuracy in detecting occluded citrus fruits, making it a reliable solution for agricultural applications.