damage assessment in military operations using deep learning and image processing

Object detection plays an important role in the analysis of images of battle scenes, especially damage assessment. In this article, machine vision techniques along with deep learning and digital image processing are used to achieve a powerful method with maximum speed and accuracy to detect objects on the battlefield and also estimate the damage. To object detection, the features and parameters of images are extracted by convolutional neural networks and used in neural network learning. The structural similarity criteria, mean square error, and threshold method were used to assess the damage and to measure the similarity and changes in the images which received before and after military operations. Finally, for validating the method, samples of the battle scenes images have been investigated, and object detection and damage assessment has been executed on them.