Building calibration method based on transfer learning

• Main Authors: MAO JIAFA, WANG NING, LANG JIAJIN, LU SHUFANG, YU LINLIN, HU YAHONG
• Format: Patent
• Language: Chinese; English
• Published: 27.12.2019
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

A building calibration method based on transfer learning comprises the following steps: 1) taking a self-made data set to construct a training set and a test set, and the data set comprising buildingdesign drawings from related design institutes and building pictures actually shot on line and is applied to serve as recommendation system test data; 2) constructing a building calibration technologyarchitecture model, wherein the model comprises image feature calibration, image feature extraction, database matching, Mahalanobis distance setting and final image authentication; 3) segmenting thedata set into small blocks, sequentially inputting the small blocks into a building recognition neural network, updating parameters by using a back propagation Adam algorithm, and training a finally changed classifier for training through the set number of training rounds; and 4) calibrating an actual building picture by using the trained model. 一种基于迁移学习的建筑标定方法，包括如下步骤：1)取自制数据集分别构造训练集和测试集，该数据集包含来自相关设计所的建筑物设计图以及网上实拍建筑物图片，应用作为