firearms identifi cation

Introduction. The aim of the work is to increase the effi ciency of identifi cation of fi rearms by images of fi ring pin marks in the automatic mode. The relevance of the task is determined by the low effi ciency of the known methods of automatic identifi cation of fi rearm by the fi ring pin marks with individual topological types of individualizing features. This aff ects the investigation of crimes related to the use of fi rearms. Formation of clone images. A training sample was formed; it included 140 original images of fi ring pin marks from 50 classes, on the basis of which about 1000 clone images were made with slightly modifi ed individualizing features. In this case a specifi c specimen of a fi rearm is meant as a class. Neural network training. A fully connected neural network with the following architecture was used as a classifi er: an input layer of neurons; two hidden layers; an output layer. The input layer included 2500 neurons, the fi rst hidden layer was made up of 625 neurons, the second hidden layer contained 156 neurons; the output layer consisted of 50 neurons (in accordance with the number of the classes). Evaluation of the calculation results. The prediction accuracy of the trained neural network was estimated according to the Accuracy metric, which is the ratio of the number of correct predictions to the total number of predictions. The prediction accuracy for the maximum signal on one output neuron was 81%, and when the maximum signals on three output neurons were taken into account, the accuracy was about 91%. Conclusions. The research has shown the possibility of classifi cation of the images of fi ring pin marks by weapons using a fully connected neural network, as well as the eff ectiveness of using artifi cially generated clone images of fi ring pin marks for training a fully connected neural network in cases with a small number of initial objects.