Unpaired iPhone And DSLR Flower Images For CycleGAN

Comprising unpaired images of flowers captured by iPhone and DSLR cameras, originally part of the CycleGAN benchmark datasets. It is used for unpaired image-to-image translation tasks, as described in the associated research paper from 2017. The dataset is hosted on Hugging Face by the 'huggan' organization.

Use Cases

• Train a CycleGAN model for unpaired image-to-image translation between iPhone and DSLR photo styles.
• Benchmark domain adaptation performance using the established CycleGAN dataset framework.
• Analyze visual feature differences between consumer-grade (iPhone) and professional-grade (DSLR) photography of flowers.

Strengths

• Part of the established CycleGAN benchmark suite, enabling direct comparison with published research.
• Dataset is associated with a seminal 2017 arXiv paper (abs/1703.10593) on unpaired image translation.
• Available in a structured format (Parquet) via the Hugging Face platform with associated metadata tags.

Limitations

• The specific number of images, image resolution, and file formats are not provided in the input.
• No sample data or column descriptions are available to assess the data structure.
• The dataset was last updated in April 2022 and may not reflect the latest camera models or styles.

Provenance

Source

Originally from the CycleGAN datasets hosted by researchers at UC Berkeley.

Collection Method

Created for research on unpaired image-to-image translation using Cycle-Consistent Adversarial Networks.

Freshness

Last updated on 2022-04-12.