On Trend Hybrid Pattern Generation in Fabrics using Machine Learning

Abstract

In recent years, there has been an immense demand for unique designs across creative industries. This demand for designs is significantly fueled by social media trends, innovations in technology, industry and consumer psychology. Most of the designs patterns from the industry are curetted by designers. Therefore, there is an increased need for designers to create unique patterns based on user preferences in the shortest time possible. The proposed system described in this paper is intended to build a artificial neural network with three fully connected hidden layer to generate unique hybrid design patterns with the help of an input image from the user. The designs that are created through this network of neurons are called as hybrid designs because these patterns are generated by specialized algorithms having image transformation techniques like pencil sketch, watercolor and pattern merge. Thus the generated images ends up being ranked by the user making the model learn with updated weights every time the preferences are changed. The primary goal of utilizing this network is to improve deal through insightful automation in designing patterns.