YT-BB-Dog
The YT-BB-Dog is a short-term dog re-identification dataset. It is a variation of theYT-BB dataset, extracting dogs bounding boxes from video frames of multiple Youtube videos.
It is composed of:
- 2,723 dogs.
- 27,036 images.
We also make available the YT-BB-Dog random background, which is the test set with 100 random backgrounds.
Example of extracted crops:
Sibetan
The Sibetan dataset is a cross-camera dog re-identification dataset collected in Sibetan, Indonesia, in the context of the SEAdogSEAproject. It covers one week of recordings from 2019, extracting dogs bouding boxes from camera-traps videos of resolution 720x480.
It is composed of:
- 59 dogs.
- 1,755 images.
Example of extracted crops:
BIFOR weights
Trained weights of the models reported in the paper Background-invariant re-identification of dogs from camera-trap videos in non-controlled environments:
- f(0) - ConvNext-Base pre-trained on ImageNet.
- f(1) - Background-based Network.
- f(2) - BIFOR (Background-Invariant Feature extractOR).
Contact
Eugênio Dias Ribeiro Neto
PhD Student in Computer Science
LIRMM, University of Montpellier, CIRAD
Marc Chaumont
Researcher and associate professor
IRISA, University of Southern Brittany, LIRMM
License
For citing YT-BB-Dog dataset, please cite:
- Esteban Real, Jonathon Shlens, Stefano Mazzocchi, Xin Pan, and Vincent Vanhoucke. YouTube-BoundingBoxes: A Large High-Precision Human-Annotated Data Set for Object Detection in Video. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, CVPR’2017, pages 7464–7473, Honolulu, Hawaii, July 2017. https://research.google.com/youtube-bb/.
- Eugênio Dias Ribeiro Neto, Cyril Barrelet, Marc Chaumont, Gérard Subsol, Muhammad Nur Faiz Mahfudz, Muhammad Najib Arung Petana Raja Bone, Barandi Sapta Widartono, Hery Wijayanto, Dyah Ayu Widiasih, Mia Nur Farida, Wayan Tunas Artama, Thibaut Langlois, Hélène Guis, Etienne Loire and Michel de Garine-Wichatitsky. Background-invariant re-identification of dogs from camera-trap videos in non-controlled environments, Ecological Informatics, 2025, https://doi.org/10.1016/j.ecoinf.2025.103547.
For citing the Sibetan dataset, please cite:
- Eugênio Dias Ribeiro Neto, Cyril Barrelet, Marc Chaumont, Gérard Subsol, Muhammad Nur Faiz Mahfudz, Muhammad Najib Arung Petana Raja Bone, Barandi Sapta Widartono, Hery Wijayanto, Dyah Ayu Widiasih, Mia Nur Farida, Wayan Tunas Artama, Thibaut Langlois, Hélène Guis, Etienne Loire and Michel de Garine-Wichatitsky. Background-invariant re-identification of dogs from camera-trap videos in non-controlled environments, Ecological Informatics, 2025.
