Minimal Dataset for "Contour-Text Fusion Generative Adversarial Network for Image Generation"

# CTFGAN Minimal Dataset This record contains the preprocessed data and supporting materials used in the experiments reported in *Contour-Text Fusion Generative Adversarial Network for Image Generation*. ## Contents - `CUB/`: preprocessed caption metadata, DAMSM encoder resources, processed data files, text annotations, and the training and test partitions used for the CUB experiments.- `COCO/`: preprocessed caption metadata, DAMSM encoder resources, processed data files, text annotations, and the training and test partitions used for the COCO experiments.- `figure_sources/Image_Sources.xlsx`: source pages, creator information, license information, figure positions, and descriptions of modifications for the CC BY 4.0 images associated with Figs. 1,2,5-7. The test samples used by all compared methods correspond to the test partitions included in the `CUB/` and `COCO/` directories. ## Original datasets The original CUB-200-2011 dataset is available from the California Institute of Technology:https://www.vision.caltech.edu/datasets/cub_200_2011/ The original COCO dataset is available from the COCO project:https://cocodataset.org/#home ## Evaluation resources - Inception Score: https://github.com/hanzhanggit/StackGAN-inception-model- FID: https://github.com/MinfengZhu/DM-GAN/tree/master/eval/FID- LPIPS: https://github.com/richzhang/PerceptualSimilarity ## License and third-party materials The author-generated documentation and source-attribution compilation in this record are released under the Creative Commons Attribution 4.0 International license (CC BY 4.0). Third-party datasets, images, pretrained models, and code remain subject to their original licenses and terms of use. The CC BY 4.0 license assigned to this record does not replace or modify those original terms.