110_Poem_AI

• 

  AI_ARTATHON_110_final_result

• 

  AI_ARTATHON_110_final output_1

Inspiration

• Harnessing artificial intelligence ability through PCA and generative adversarial network (GAN) to translate historical poems feelings into a painting

What it does

• PCA generate an image from our own datasets that we collect and find the Eigen image between the datasets

How we built it

• By feeding PCA set of images then transfer the output to transfer style

Challenges we ran into

• Lack of resources of GAN models that well documented and satisfy our ideas
• Lack of Arabic language support in vector space models used by zero-shot from text model
• Resources consuming to do the training on our datasets

Accomplishments that we're proud of

• Proud of generating paining from historical Arabic poem

What we learned

• Applying AI technology and art methods

What's next for 110_Poem_AI

• Generating a painting from a poem by using an AI model with or without any human interaction which transfer words, feeling and scenery into objects in the painting this is achieved by implementing a vector space model for Arabic language (zero-shot) from text

PCA notebook

https://drive.google.com/file/d/1YpmqC1mfFhueMBiZVDxrk0Y2Har-hdKt/view?usp=sharing

Style transfer

https://drive.google.com/file/d/19W_-YSdsL06reWGbGKQyKdS0dySmiJ3a/view?usp=sharing

The presentation

• abstract presentation https://drive.google.com/open?id=1NYX6VuTKSIkvwkjN0vDPBR2CHh6ZFZ6o

• details presentation https://drive.google.com/open?id=11Q5lo4RXthkL_IzeEhqawNoBQd4yEj-J

Built With

• gan
• google-colab
• pca
• python
• pytorch
• style-transfer
• tensorflow