Currently,the spotting of birds activity mainly relys on the pilots'observation in the general avi-ation activities.Besides,the methods at home and abroad are not suitable for widespread use.Therefore,focusing on the problem above,with the aid of image segmentation method based on the deep learning,the problem can be solved.U-Net image segmentation strategy is introduced to aviation images analysis,and deep convolutional neural network is used to process the images of birds.Based on the characteristics of complex image background,uneven target distribution and variable size,an optimized U-NET network is proposed to save and utilize the image information better.Experiments show that the image segmentation effect of the optimized network is better than existing method.
关键词
深度学习网络/注意力机制/鸟群活动/图像分割/航空安全
Key words
deep learning network/attention mechanism/birds activity/image segmentation/aviation safety
维普
万方数据