Research on Object Detection Method Based on Distance Range-Gated Images
Since the lack of color feature information in range-gated images leads to low detection rate and weak fea-ture extraction of distance information,a range-gated images detection method based on improved YOLOv8s network is proposed.Firstly,an adaptive range-gated feature fusion module is proposed to perform feature-level fusion of the range-gated image and its depth information to obtain high-level semantic information.Secondly,a spatial attention module based on the composition of multiscale depthwise separable convolutions modules is used to perform a weighting operation on each feature layer to obtain more key features.Finally,the Neck part of the PAN structure is improved to enhance the net-work's target detection ability.Experiments are carried out on the range-gated image dataset.
NETL
NSTL
万方数据
