CADDY Underwater Gestures Dataset

10K stereo pair images collected in 8 different scenarios

CADDY Underwater Gestures Dataset

Home Page

8 different recording scenarios

ALT-TEXT-biograd-A_00162_left.jpg ALT-TEXT-biograd-B_00032_right.jpg ALT-TEXT-biograd-C_00098_right.jpg ALT-TEXT-brodarski-A_00018_right.jpg
ALT-TEXT-brodarski-B_00029_right.jpg ALT-TEXT-brodarski-C_00006_left.jpg ALT-TEXT-brodarski-D_00032_right.jpg ALT-TEXT-genova-A_00032_right.jpg

Citations

  1. Gomez Chavez, A.; Ranieri, A.; Chiarella, D.; Zereik, E.; Babić, A.; Birk, A. CADDY Underwater Stereo-Vision Dataset for Human–Robot Interaction (HRI) in the Context of Diver Activities. J. Mar. Sci. Eng. 2019, 7, 16. [online version], [PDF], [BibTeX], [RIS].

  2. Chiarella, D. Towards Multi-AUV Collaboration and Coordination: A Gesture-Based Multi-AUV Hierarchical Language and a Language Framework Comparison System. J. Mar. Sci. Eng. 2023, 11, 1208. https://doi.org/10.3390/jmse11061208 [online version],[BibTex], [RIS]

  3. Chiarella, D.; Bibuli, M.; Bruzzone, G.; Caccia, M.; Ranieri, A.; Zereik, E.; Marconi, L.; Cutugno, P. A Novel Gesture-Based Language for Underwater Human–Robot Interaction. J. Mar. Sci. Eng. 2018, 6, 91. [PDF], [BibTeX], [RIS]

Video

Toy dataset – 163MB (digests) CSV Toy dataset – 191KB (digests)
Complete dataset – 2.5GB (digests) CSV complete dataset (true positives) – 20.4MB (digests) CSV complete dataset (true negatives) – 12.8MB (digests)
Complete dataset (Fast.ai-friendly version) – 1.5GB (digests) Complete dataset (Fast.ai-friendly version - squashfs) – 1.5GB (digests) Complete dataset (crystal-clear vs. cloudy-and-dark split - zip version for Colab) – 1.5GB (digests)
Hands ROI images – 1.5GB (digests)
The CADDIAN sign language CADDIAN quick reference guide & slang

Changelog

  • 28.06.2023 Added reference to article containing new version of Caddian Language, see [PDF].
  • 12.09.2019 Added “crystal-clear vs. cloudy-and-dark split” zipfile to be used on Colab + Hands ROI images
  • 31.07.2019 Added Fast.ai-friendly versions of the dataset (only raw and “true positives” images, archive formats are zip and squashfs)
  • 29.07.2019 Added digests for each .zip and .csv file
  • 08.01.2019 Add columns for image quality metrics according to reference [1], and an index column for easier parsing.
  • 08.01.2019 Eliminate some duplicate entries from the true positives CSV files.
  • 12.07.2018 First version online, see reference [2].

For any inquiry about the dataset and technical details please contact Arturo Gomez Chavez at a.gomezchavez@jacobs-university.de

For any inquiry about the language CADDIAN please contact Davide Chiarella at davide.chiarella@cnr.it