SLMS-SSD: Improving the balance of semantic and spatial information in object detection

作者：

Highlights：

• Designing feature connectivity to balance semantic and spatial information.

• Self-learning feature strength to measure the importance of different features.

• Extracting shallow feature contextual information that facilitates detection.

• Improving accuracy compared to SSD series methods, especially for small objects.

摘要

•Designing feature connectivity to balance semantic and spatial information.•Self-learning feature strength to measure the importance of different features.•Extracting shallow feature contextual information that facilitates detection.•Improving accuracy compared to SSD series methods, especially for small objects.

论文关键词：Object detection,Deep learning,Multi-scale feature selection,Self-learning feature fusion

论文评审过程：Received 1 February 2022, Revised 15 May 2022, Accepted 28 May 2022, Available online 13 June 2022, Version of Record 21 June 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117682