marko-barisic/sea-turtle-behavior-data

Sea Turtle Behavior Data

Growing collection of supplementary data for the sea turtle breathing-behavior project. This repo holds the underlying source material behind the labeled training/eval image dataset (link to be added) and the trained classifier (link to be added), so results can be reproduced from raw footage rather than only from already-processed crops.

Current contents: raw video

{camera}/{date}/{stem}.mp4 — 42 files, 1GB each (41.6GB total), covering the full 06:00–18:00 recording window for both cameras across both recording days:

Camera	Date	Videos
D01	2025-08-08	11
D01	2025-08-09	10
D03	2025-08-08	11
D03	2025-08-09	10

Each {stem} encodes the camera and the video's own recording start timestamp, e.g. D01_20250808060025.mp4 = camera D01, starts 2025-08-08 06:00:25. Videos are unmodified originals (no re-encoding, no trimming) — a day's ~12 hours is covered by several sequential videos per camera, not one continuous file. D01 and D03 each show a single turtle in a rehabilitation tank (different individuals per camera).

This is exactly the footage the processing pipeline (link to be added) consumed to produce the frame extractions, YOLO detections, and breathing classifications reported in the paper — uploaded as-is so the full pipeline can be re-run from source.

Planned additions

More data connected to the paper will be added here over time, likely including ground-truth review labels, segmentation data, and per-camera-day model outputs. Check back for updates; this section will be revised as content is added.

License

CC BY 4.0 — free to use with attribution.

Authors

• Marko Barišić — University of Zagreb, FER, Laboratory for Underwater Systems and Technologies (LABUST)
• Yaniv Levi — Sea Turtle Rescue Center, National Nature and Parks Authority, Israel
• Nikola Mišković — University of Zagreb, FER
• Roee Diamant — University of Haifa