MoEIoU: Rethinking Bounding-Box Regression as a Mixture of Experts 事件

MoEIoU: Rethinking Bounding-Box Regression as a Mixture of Experts arXiv:2606.00844v1 Announce Type: new Abstract: Bounding-box regression is a fundamental component of object detection, playing a critical role in precise object localization. Existing Intersection-over-Union (IoU)-based loss functions extend the IoU objective by incorporating geometric penalties, such as center-distance and aspect-ratio mismatch, to improve bounding-box regression. However, these penalties typically remain fixe