HyperAI

VIDIT Domain Adaptation Dataset

Date

3 years ago

Size

19.28 GB

Organization

École Polytechnique Fédérale de Lausanne

Publish URL

github.com

License

其他

特色图像

VIDIT, short for Virtual Image Dataset for Illumination Transfer, is a reference evaluation benchmark. VIDIT contains 390 different Unreal Engine scenes, each captured using 40 lighting settings, generating 15,600 images.

The lighting settings are all combinations of 5 color temperatures (2500K, 3500K, 4500K, 5500K, and 6500K) and 8 light directions (N, NE, E, SE, S, SW, W, NW). The original image resolution is 1024×1024.

This dataset is released by EPFL.

VIDIT.torrent
Seeding 1Downloading 1Completed 602Total Downloads 455
  • VIDIT/
    • README.md
      1.23 KB
    • README.txt
      2.46 KB
      • data/
          • track1/
            • track1_test.zip
              62.31 MB
            • track1_train.zip
              582.81 MB
            • track1_validation.zip
              639.96 MB
            • track1_validation_gt.zip
              686.52 MB
          • track2/
            • track2_test.zip
              969.54 MB
            • track2_train.zip
              18.93 GB
            • track2_validation.zip
              19.19 GB
            • track2_validation_gt.zip
              19.28 GB