Obtaining retrospective data of the urbanized landscape condition using computer vision methods on the example of construction works

Retrospective information about the state of the urbanized environment can be obtained by analyzing images located in the Internet space. Artificial intelligence-based methods allow recognizing objects of a given class in images, while models based on neural networks demonstrate high recognition accuracy. In this paper, a neural network model based on its own dataset for recognizing objects of an urbanized environment is proposed using the example of a class of construction works an creation of geoinformation system with coordinate time reference. The pixellib library was used for image segmentation. Photos found on the Internet were used to create the dataset. During the testing of the model, 15 versions of the neural network were trained. The best results were demonstrated by the model mask_rcnn_model.160-0.460171.h5 (version 15). The main reason for increasing the recognition accuracy of this model was more accurate contouring of objects and the addition of self-captured images to the dataset. The authors plan to improve the recognition accuracy of the neural network and train it to recognize other classes of objects in an urbanized environment.