ScreenAI

Screen AI

Implementation of the ScreenAI model from the paper: "A Vision-Language Model for UI and Infographics Understanding". The flow is: img + text -> patch sizes -> vit -> embed + concat -> attn + ffn -> cross attn + ffn + self attn -> to out. PAPER LINK:

Install

pip3 install screenai

Usage

…

License

MIT

Citation

@misc{baechler2024screenai,
    title={ScreenAI: A Vision-Language Model for UI and Infographics Understanding}, 
    author={Gilles Baechler and Srinivas Sunkara and Maria Wang and Fedir Zubach and Hassan Mansoor and Vincent Etter and Victor Cărbune and Jason Lin and Jindong Chen and Abhanshu Sharma},
    year={2024},
    eprint={2402.04615},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

Todo

• Implement the nn.ModuleList([]) in the encoder and decoder