Feature fusion for object detection at one map
作者：
Highlights：
• The feature fusion increases feature interaction in the single-input and single-output framework.
• The feature maps of the same resolution and different receptive fields are fused to achieve a similar function with FPN.
• A weighted method with fast feature dropping is proposed for information-intensive and noisy problem.
• Results on the COCO dataset suggest that the proposed method can reduce feature correlation and improve model performance.
摘要
•The feature fusion increases feature interaction in the single-input and single-output framework.•The feature maps of the same resolution and different receptive fields are fused to achieve a similar function with FPN.•A weighted method with fast feature dropping is proposed for information-intensive and noisy problem.•Results on the COCO dataset suggest that the proposed method can reduce feature correlation and improve model performance.
论文关键词：Object detection,Multi-scale fusion,COCO
论文评审过程：Received 10 February 2022, Revised 11 April 2022, Accepted 18 April 2022, Available online 21 April 2022, Version of Record 11 May 2022.
论文官网地址：https://doi.org/10.1016/j.imavis.2022.104466
