The paper addresses the limitations of DETR-based semi-supervised object detection (SSOD) frameworks, particularly focusing on the challenges posed by the quality of object queries. It introduces Sparse Semi-DETR, a novel transformer-based, end-to-end SSOD solution that incorporates a Query Refinement Module and a Reliable Pseudo-Label Filtering Module to enhance the detection of small or obscured objects and improve the reliability of predictions in complex scenarios. The Query Refinement Module enhances the quality of object queries by refining low-level and high-level features, while the Reliable Pseudo-Label Filtering Module selectively filters high-quality pseudo-labels to reduce the impact of noisy labels. On the MS-COCO and Pascal VOC object detection benchmarks, Sparse Semi-DETR achieves significant improvements over current state-of-the-art methods, particularly in challenging scenarios involving small or partially obscured objects. The key contributions of the work include the introduction of query refinement and low-quality proposal filtering for the one-to-many query assignment strategy, enhancing the performance and efficiency of object detection in semi-supervised settings.The paper addresses the limitations of DETR-based semi-supervised object detection (SSOD) frameworks, particularly focusing on the challenges posed by the quality of object queries. It introduces Sparse Semi-DETR, a novel transformer-based, end-to-end SSOD solution that incorporates a Query Refinement Module and a Reliable Pseudo-Label Filtering Module to enhance the detection of small or obscured objects and improve the reliability of predictions in complex scenarios. The Query Refinement Module enhances the quality of object queries by refining low-level and high-level features, while the Reliable Pseudo-Label Filtering Module selectively filters high-quality pseudo-labels to reduce the impact of noisy labels. On the MS-COCO and Pascal VOC object detection benchmarks, Sparse Semi-DETR achieves significant improvements over current state-of-the-art methods, particularly in challenging scenarios involving small or partially obscured objects. The key contributions of the work include the introduction of query refinement and low-quality proposal filtering for the one-to-many query assignment strategy, enhancing the performance and efficiency of object detection in semi-supervised settings.