Machine Learning Methods for Solving Scrap Metal Classification Task

Abstract

This paper deals with the task of scrap metal images classification. The authors proposed the method for automation of the cropping scrap metal images from convex quadrangle process and applied this method on railway carriages photographs. The brief description of convolutional neural networks (CNN) and machine learning methods used during the research is given in the paper. The paper presents the results of using various CNN and machine learning methods in the task of classifying images of scrap metal. The algorithm of improving image classification results is proposed. The results of the calculation showed high classification accuracy and allowed to choose the best classifier.