DEEP LEARNING-BASED ANTENNA DOWNTILT ANGLE MEASUREMENT METHOD

摘要

Disclosed in the present invention is a deep learning-based antenna downtilt angle measurement method, comprising the following steps: establishing an antenna database and performing quantization processing; inputting an antenna picture into a deep neural network, and entering a feature extraction network so as to obtain an antenna feature image; the antenna picture entering an SE characterization enhancement network to selectively enhance the inclusion of useful features and suppress useless features; and the antenna picture entering a target identification network to identify candidates for the antenna and obtain an antenna downtilt angle, wherein the SE characterization enhancement network is provided with a compression excitation unit to model a dependence relationship between channels, and adaptively adjusts a feature response value of each channel. The antenna downtilt angle is obtained by means of processing the antenna picture using a deep learning network, and a convenient, safe, effective and accurate antenna measurement method is established.