Style Transfer

Project description

This work aims to develop an algorithm for style transfer, using the Python object orientation paradigm, among others. technologies. Style transfer is a process of migrating a style from one image to the content of another, generating a new image that is a mix of the previous two, this was only possible through the use of an architecture composed by convolutional neural networks (CNN), one of them is VGG16, being a trained network optimized for image recognition. Under these circumstances several libraries were used, such as Matplotlib, TensorFlow, NumPy, Python. Imaging Library (PIL), datetime, which assisted in the creation of the project. The program was developed with the implementation of classes following the concept of object orientation in Python, making the code more organized and easy to read. For the creation of development environment, Anaconda was used for package management. and libraries, and Visual Studio Code to edit the code. Results Provide New Imaging Representations Using Machine Learning by Neural Networks Convolutional, demonstrating its potential for image manipulation.