Go Art
- Hasan Alshikh
- Shahad Almasud
The main idea behind building this project is to generate stunning art/painting by merging images and random styles—thus, which to transform the image into Asir, Makkah, Najd culture themes, deep learning techniques are utilized and trained with thousands of epochs/weights to manipulate regular images by adopting the visual styles of art/painting to generate a new art style. Ultimately, the project was created autonomously by AI systems and works that collaborate between a human and AI system.
Transfer learning concepts are applied in our project and pre-trained convolutional networks like VGG network, VGG-19. Next, we loaded the model to adopt its learned feature maps to describe our images' content and style representation. Finally, we pre-processed/prepared our data, set up loss functions, applied different convolutional and pooling layers, all of which were used to create our model with a rectified linear unit (ReLU) activation function.
The goal is the embodiment of the Saudi heritage into digital art method was used to present graphic arts by merging randomly backgrounds with different Saudi styles, where photographic colors were also combined with backgrounds and random patterns with images. This artistic project creates a new art that transforms image pixels into colors combined in one final palette.