Interactive multi-scale feature representation enhancement for small object detection

作者：

Highlights：

• An auxiliary multi-scale input network for enhancing shallow features.

• Interactive multi-scale features for aggregating contextual information.

• An adaptive factor embedded in the interactive module to control the fusion ratio.

摘要

•An auxiliary multi-scale input network for enhancing shallow features.•Interactive multi-scale features for aggregating contextual information.•An adaptive factor embedded in the interactive module to control the fusion ratio.

论文关键词：Object detection,Small objects,Deep learning,Multi-scale feature fusion

论文评审过程：Received 4 January 2021, Revised 31 January 2021, Accepted 2 February 2021, Available online 6 February 2021, Version of Record 15 February 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104128