MPDIoU: A Loss for Efficient and Accurate Bounding Box Regression

TL;DR

This paper introduces MPDIoU, a novel bounding box similarity metric and loss function that improve the accuracy and efficiency of bounding box regression in object detection and segmentation tasks.

Contribution

The paper proposes MPDIoU, a new geometric-based similarity metric and a corresponding loss function that outperform existing methods in bounding box regression.

Findings

MPDIoU improves bounding box regression accuracy.

LMPDIoU loss enhances object detection and segmentation performance.

Experimental results outperform existing loss functions on multiple datasets.

Abstract

Bounding box regression (BBR) has been widely used in object detection and instance segmentation, which is an important step in object localization. However, most of the existing loss functions for bounding box regression cannot be optimized when the predicted box has the same aspect ratio as the groundtruth box, but the width and height values are exactly different. In order to tackle the issues mentioned above, we fully explore the geometric features of horizontal rectangle and propose a novel bounding box similarity comparison metric MPDIoU based on minimum point distance, which contains all of the relevant factors considered in the existing loss functions, namely overlapping or non-overlapping area, central points distance, and deviation of width and height, while simplifying the calculation process. On this basis, we propose a bounding box regression loss function based on MPDIoU, called LMPDIoU . Experimental results show that the MPDIoU loss function is applied to state-of-the-art instance segmentation (e.g., YOLACT) and object detection (e.g., YOLOv7) model trained on PASCAL VOC, MS COCO, and IIIT5k outperforms existing loss functions.