ExplainableML / ImageSelect

Code for the paper "If at First You Don't Succeed, Try, Try Again: Faithful Diffusion-based Text-to-Image Generation by Selection"

Faithful Diffusion-based Text-to-Image Generation by Selection

This is the PyTorch code our work If at First You Don't Succeed, Try, Try Again: Faithful Diffusion-based Text-to-Image Generation by Selection .

Setup

You can setup the environments required by Stable Diffusion and ImageReward

Demo

To generate an image for a given prompt, all you need to do is create the directories ./all_images and ./best_images and then run the command python3 src/imageselect_demo.py --num_seeds <num_imgs_generated> --prompt <prompt>. The best image selected by ImageReward will be saved in the directory ./best_images.

Diverse1k Dataset

The prompts for the Diverse1k dataset can be found in ./data/1k_prompts.json and the corresponding QA data (which can be used for the TIFA evaluation) collected from GPT3.5 is provided in ./data/qa.json.

References

If you find this work useful, please cite:

@article{karthik2023if,
  title={If at First You Don't Succeed, Try, Try Again: Faithful Diffusion-based Text-to-Image Generation by Selection},
  author={Karthik, Shyamgopal and Roth, Karsten and Mancini, Massimiliano and Akata, Zeynep},
  journal={arXiv preprint arXiv:2305.13308},
  year={2023}
}