Fast forest fire smoke detection using MVMNet

作者：

Highlights：

• The first multi-oriented detection applied in the field of forest fire detection data Set is made.

• Softpool-Spatial pyramid pooling is proposed to maintain more feature information.

• Value conversion-Attention mechanism module is proposed.

• Mixed-NMS based on the fusion of DIoU-NMS and Skew-NMS.

摘要

•The first multi-oriented detection applied in the field of forest fire detection data Set is made.•Softpool-Spatial pyramid pooling is proposed to maintain more feature information.•Value conversion-Attention mechanism module is proposed.•Mixed-NMS based on the fusion of DIoU-NMS and Skew-NMS.

论文关键词：Forest fire smoke detection,Multioriented,Value conversion-attention mechanism module,Softpool-spatial pyramid pooling,Mixed-NMS

论文评审过程：Received 14 September 2021, Revised 11 January 2022, Accepted 11 January 2022, Available online 21 January 2022, Version of Record 3 February 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.108219