metadata
task_categories:
- object-detection
pretty_name: Gilt Tracking Dataset
Gilt Tracking Dataset
- The videos are split into 5-minute lengths. The video filenames are assigned according to the recording start timestamp, pen, and camera types.
- Virtual IDs are assigned. RFID reading data is provided with a timestamp and ID for mapping later.
- The bounding box contains Center (Cx,Cy) and Width (w) and Height (h) of the bounding box annotations with tracking ID in coco format
🧠 Use Cases
- Animal behavior monitoring and tracking
- Multiview videos from two sides of the pen
- Precision livestock farming
- Continuous tracking animal for half an hour of video
License
The author has granted permission to download, use, and redistribute this dataset only for research purposes.
Citation
Please cite as Bhujel A. et al. (2025). A computer vision dataset for Gilt daily posture activity monitoring and tracking.
Contact
For questions or collaborations, feel free to reach out at [email protected]