LWDNet-A lightweight water-obstacles detection network for unmanned surface vehicles

Qilie Cai, Qiang Wang, Yulong Zhang, Zhibo He, Yuhong Zhang

Abstract: Highlights•We complete an novel lightweight water-obstacles detection network.•We adopt a novel backbone, which is bottleneck structure with attention block.•Dilated convolution is adopted in the depthwise separable convolution (DW convolution).•Focal loss has been improved in our network.