The paper "DQ-DETR: DETR with Dynamic Query for Tiny Object Detection" addresses the challenge of detecting tiny objects in aerial images, where previous DETR-like methods struggle due to their fixed query number and lack of adaptability to varying instance counts. The authors propose DQ-DETR, a novel DETR-like model that includes three main components: a categorical counting module, a counting-guided feature enhancement module, and a dynamic query selection module. These components dynamically adjust the number and positional information of object queries based on the density of objects in an image, improving detection accuracy, especially in imbalanced datasets. The model outperforms state-of-the-art methods, achieving a mAP of 30.2% on the AI-TOD-V2 dataset, which is primarily composed of tiny objects. The paper also provides a detailed analysis of the model's performance on different datasets and conducts ablation studies to validate the effectiveness of each component.The paper "DQ-DETR: DETR with Dynamic Query for Tiny Object Detection" addresses the challenge of detecting tiny objects in aerial images, where previous DETR-like methods struggle due to their fixed query number and lack of adaptability to varying instance counts. The authors propose DQ-DETR, a novel DETR-like model that includes three main components: a categorical counting module, a counting-guided feature enhancement module, and a dynamic query selection module. These components dynamically adjust the number and positional information of object queries based on the density of objects in an image, improving detection accuracy, especially in imbalanced datasets. The model outperforms state-of-the-art methods, achieving a mAP of 30.2% on the AI-TOD-V2 dataset, which is primarily composed of tiny objects. The paper also provides a detailed analysis of the model's performance on different datasets and conducts ablation studies to validate the effectiveness of each component.