Dataset Viewer
Auto-converted to Parquet Duplicate
No rows found.
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

Red Sea Fish Dataset

Dataset Description

The Red Sea Fish Dataset is a specialized collection of fish images captured using a stationary ViewIntoTheBlue 360° underwater camera system (https://www.viewintotheblue.com/). This dataset features various fish species native to the Red Sea, providing labeled data for marine life identification and monitoring.

Overview

  • Task: Image classification/detection of Red Sea fish species
  • Camera System: ViewIntoTheBlue 360° underwater camera (stationary deployment)
  • Region: Red Sea
  • Dataset Size: 1,054 total frames
    • Training set: 854 frames
    • Validation set: 200 frames
  • Number of Classes: 43 distinct fish species/categories

Species Included

The dataset includes the following fish species:

  1. Arabian Angelfish
  2. Arabian Bluestriped Snapper
  3. Barracudas
  4. Big Squirrelfish
  5. Black Surgeonfish
  6. Blackspotted Rubberlips
  7. Blacktail Dartfish
  8. Bluefin Trevally
  9. Bluespine Unicornfish
  10. Breams
  11. Emperor
  12. Fourline Wrasse
  13. Humpback Snapper
  14. Longnose Parrotfish
  15. Lutjanus
  16. Malabar Grouper
  17. Mangrove Red Snapper
  18. Moon Wrasse
  19. Moonfish
  20. Onespot Snapper
  21. Picasso Triggerfish
  22. Rabbitfish
  23. Racoon Butterflyfish
  24. Redbreasted Wrasse
  25. Red Sea Bannerfish
  26. Red Sea Demoiselle
  27. Red Sea Goatfish
  28. Sky Emperor
  29. Small Threeline Damselfish
  30. Smalltooth Emperor
  31. Sohl Surgeonfish
  32. Splendour Wrasse
  33. Squirrelfish
  34. Striated Fusilier
  35. Striated Surgeonfish
  36. Threeline Damselfish
  37. Titan Triggerfish
  38. Unknown
  39. Unknown Damselfish
  40. Unknown Grouper
  41. Unknown Parrotfish 1
  42. Whitefaced Butterflyfish
  43. Yellowfin Hind

Dataset Structure

The dataset is organized into two main splits:

  • Training set (train): 854 frames
  • Validation set (valid): 200 frames

Each split contains:

  • Image files in the imgs directory
  • Corresponding annotation files in JSON format in the annotations directory

Annotation Format

Each annotation file is in JSON format and contains:

  • Image filename
  • Species identification
  • [Add additional annotation details if available]

Usage

This dataset can be used for:

  • Training computer vision models for fish species identification
  • Underwater marine life monitoring
  • Ecological studies of Red Sea fish populations
  • Development of automated fish counting systems

Citation

If you use this dataset in your research, please cite:

Downloads last month
16