M2YOLOF: Based on effective receptive fields and multiple-in-single-out encoder for object detection

作者：

Highlights：

• The proposed M2YOLOF detects objects on one feature map and performs well on small objects.

• The detection speed of our method is as fast as YOLOF, but more accurate.

• Designing a Multi-in-Single-out attention encoder to improve representation from backbone.

• Designing a dynamic sample selection method in combination with effective receptive fields.

摘要

•The proposed M2YOLOF detects objects on one feature map and performs well on small objects.•The detection speed of our method is as fast as YOLOF, but more accurate.•Designing a Multi-in-Single-out attention encoder to improve representation from backbone.•Designing a dynamic sample selection method in combination with effective receptive fields.

论文关键词：Object detection,Deep learning,YOLOF,Effective receptive field

论文评审过程：Received 22 July 2022, Revised 14 September 2022, Accepted 25 September 2022, Available online 29 September 2022, Version of Record 3 October 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118928