Anchor-free YOLOv3 for mass detection in mammogram

作者：

Highlights：

• A novel anchor-free YOLOv3 network is proposed for mass detection in mammograms.

• A new feature aggregating manner by summing is designed in the up to down pathway.

• GIoU loss function is adopted to train bounding box regression.

• Focal loss is adopted as the objectness prediction loss.

摘要

•A novel anchor-free YOLOv3 network is proposed for mass detection in mammograms.•A new feature aggregating manner by summing is designed in the up to down pathway.•GIoU loss function is adopted to train bounding box regression.•Focal loss is adopted as the objectness prediction loss.

论文关键词：Mammogram,Mass detection,Anchor-free,YOLOv3

论文评审过程：Received 11 February 2020, Revised 18 October 2021, Accepted 21 November 2021, Available online 30 November 2021, Version of Record 2 December 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116273