This paper proposes Focal-EIOU loss for accurate bounding box regression (BBR) in object detection. Existing BBR loss functions suffer from inefficiency and inaccurate regression due to their inability to capture the true objective of BBR. The authors propose an Efficient Intersection over Union (EIOU) loss that explicitly measures discrepancies in three geometric factors: overlap area, central point, and side length. They also introduce a regression version of focal loss to focus on high-quality anchor boxes. Combining these two components, they propose Focal-EIOU loss, which achieves faster convergence and better localization accuracy. Extensive experiments on both synthetic and real datasets show that Focal-EIOU outperforms existing BBR losses in convergence speed and localization accuracy. The proposed loss function is effective in handling the imbalance between high and low-quality anchor boxes, and it is validated through extensive evaluations on various state-of-the-art object detection models. The results demonstrate that Focal-EIOU significantly improves detection accuracy on the COCO 2017 dataset. The contributions of this paper include: (1) proposing an efficient IOU loss to address the limitations of existing losses; (2) designing a regression version of focal loss to enhance the contributions of high-quality anchor boxes; and (3) extensive experiments showing the superiority of the proposed methods.This paper proposes Focal-EIOU loss for accurate bounding box regression (BBR) in object detection. Existing BBR loss functions suffer from inefficiency and inaccurate regression due to their inability to capture the true objective of BBR. The authors propose an Efficient Intersection over Union (EIOU) loss that explicitly measures discrepancies in three geometric factors: overlap area, central point, and side length. They also introduce a regression version of focal loss to focus on high-quality anchor boxes. Combining these two components, they propose Focal-EIOU loss, which achieves faster convergence and better localization accuracy. Extensive experiments on both synthetic and real datasets show that Focal-EIOU outperforms existing BBR losses in convergence speed and localization accuracy. The proposed loss function is effective in handling the imbalance between high and low-quality anchor boxes, and it is validated through extensive evaluations on various state-of-the-art object detection models. The results demonstrate that Focal-EIOU significantly improves detection accuracy on the COCO 2017 dataset. The contributions of this paper include: (1) proposing an efficient IOU loss to address the limitations of existing losses; (2) designing a regression version of focal loss to enhance the contributions of high-quality anchor boxes; and (3) extensive experiments showing the superiority of the proposed methods.