Remote sensing image segmentation model based on improved DeepLabV3+
In view of problems such as low precision and a large number of parameters in remote sensing image segmentation caused by classical semantic segmentation algorithms,an improved DeepLabV3+-based semantic segmentation model of remote sensing images combining lightweight network and attention mechanism was proposed.Firstly,the MobileNetV3 lightweight model was used as the feature extraction network of DeepLabV3+,which could effectively reduce the number of parameters in the whole model.Secondly,an effective channel attention mechanism was added to the DeepLabV3+model in the decoding stage,so as to increase the model's ability to fit different channel features.The experiments show that compared with that of the original model,the number of parameters of the improved DeepLabV3+model in this paper is reduced by 3.6 times,and the average intersection-over-union is increased by 3.5%.
国家科技期刊平台
NSTL
万方数据
