The paper introduces RefineDet, a novel single-shot object detection framework that combines the strengths of both two-stage and one-stage approaches. RefineDet consists of two interconnected modules: the anchor refinement module (ARM) and the object detection module (ODM). The ARM filters out negative anchors and coarse adjusts the locations and sizes of anchors, while the ODM uses these refined anchors to improve regression and predict multi-class labels. A transfer connection block (TCB) is designed to transfer features from the ARM to the ODM, enabling end-to-end training. Extensive experiments on PASCAL VOC 2007, PASCAL VOC 2012, and MS COCO datasets demonstrate that RefineDet achieves state-of-the-art detection accuracy with high efficiency. The main contributions of the work include the introduction of RefineDet, the design of the TCB, and the superior performance on various datasets.The paper introduces RefineDet, a novel single-shot object detection framework that combines the strengths of both two-stage and one-stage approaches. RefineDet consists of two interconnected modules: the anchor refinement module (ARM) and the object detection module (ODM). The ARM filters out negative anchors and coarse adjusts the locations and sizes of anchors, while the ODM uses these refined anchors to improve regression and predict multi-class labels. A transfer connection block (TCB) is designed to transfer features from the ARM to the ODM, enabling end-to-end training. Extensive experiments on PASCAL VOC 2007, PASCAL VOC 2012, and MS COCO datasets demonstrate that RefineDet achieves state-of-the-art detection accuracy with high efficiency. The main contributions of the work include the introduction of RefineDet, the design of the TCB, and the superior performance on various datasets.