Размещенные на сайте материалы передаются на сайт в автоматическом режиме по API YOUTUBE. Материалы могут содержать информацию, предназначенную для пользователей старше 18 лет.

Neural Artistic Style Transfer

8 Просмотры
Part of the ECE 542 Virtual Symposium (Spring 2020)

With the amelioration of computational capabilities, it is possible to empower machines with visual perception for creating unique and fascinating contents. This is possible by the virtue of neural artistic style transfer. Neural artistic style transfer refers to the separation and reconstruction of the contents of arbitrary content and style images to create an artistic image. Due to the importance of art, this has been a popular field of interest both academia and industry. The objective of this project is to utilize the concepts studied in ECE 542 and implement and train a neural network that constructs images of high perceptual quality by taking features of both -the original and the artistic images- into consideration. For the implementation, we have used the baseline model as VGG -16 and have compared it with the VGG - 19 layers output.
Комментариев нет.