Unpaired Cityscape Images for CycleGAN Translation

A collection of unpaired images from the Cityscapes collection, originally part of the CycleGAN project for image-to-image translation. It is associated with the research paper 'Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks' (arXiv:1703.10593). The dataset was last updated on the Hugging Face platform in April 2022.

Use Cases

• Train Cycle-Consistent Adversarial Networks (CycleGANs) for unpaired image-to-image translation tasks using the provided image collections.
• Apply the model architecture described in arXiv:1703.10593 to translate between different visual domains within the Cityscapes imagery.
• Benchmark image translation performance on urban scene data originally compiled for the CycleGAN project.

Strengths

• Associated with a seminal research paper (arXiv:1703.10593) that introduced CycleGANs.
• Part of a well-known collection of datasets specifically curated for unpaired image translation tasks.
• Hosted on a standardized platform (Hugging Face) with a recorded last update date of 2022-04-12.

Limitations

• Specific details on dataset size, number of images, resolution, and file formats are unavailable.
• The data's temporal coverage and specific geographic scope within the 'Cityscapes' theme are not specified.
• Lacks column or feature descriptions, making precise analytical use case planning difficult.

Provenance

Source

Originally hosted by the Berkeley AI Research (BAIR) laboratory, associated with authors Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A. Efros.

Collection Method

Curated as part of the CycleGAN project for unpaired image-to-image translation research.

Freshness

2022-04-12

Geography

Likely contains urban scene imagery (Cityscapes), but specific geographic coverage is not detailed.