Attention: not up to date. Will be updated soon
a collection of publicly available computer vision datasets for the re-identification of individual animals
Dataset | # of Individuals | Data | Content | Labels | License |
---|---|---|---|---|---|
AAU Zebrafish Re-Identification Dataset ![]() |
6 zebrafish | 2 videos resulting in 2224 images | side-view of fish in a tank | BB + identity annotations | CC BY 4.0 |
ATRW ![]() |
92 tigers | 8076 videos | tigers in unconstrained poses | BB + keypoint-based pose + identity annotations | CC BY-NC-SA 4.0 |
Cat Individual Images ![]() |
518 cats | 13536 images | front-view of cats with a focus on faces | identity annotations | CC BY 4.0 |
Cow Dataset ![]() |
13 cows | 3772 Images | side-view of cows | identity annotations | CC BY 4.0 |
Cows2021 ![]() |
186 cows | 10402 images + 301 videos | top down view of the cows body | oriented BB + identity + tracklet annotations | CC BY-NC-SA 4.0 |
Dog Face Recognition ![]() |
1174 dogs | ?? images | dog faces | identity annotations | CC0: Public Domain |
Happy Whale ![]() |
15587 whales and dolphins | 51033 images | 30 species of whales and dolphins entierely or partly above the water surface | identity annotations | Competition, Academic, and Non-Commercial Use Only |
SealID ![]() |
57 seals | 2080 images | seals in unconstrained poses on land | identity annotations | CC BY 4.0 |
iPanda-50 ![]() |
50 giant pandas | 6874 images | pandas in unconstrained poses | identity annotations | ?? |
Animal-Identification-from-Video ![]() |
26 pigs, 9 koi fish, 58 pigeons | 5 videos resulting in 2379 frames | pigs (above/side view), koi fish (top down view) and pigeons on the ground (sideview) in unconstrained poses | BB + identity annotaions | pixabay license |
Year | Publication* | Species | Identifying Feature | Corresponding Dataset |
---|---|---|---|---|
2022 | A Benchmark Database for Animal Re-Identification and Tracking | Pigeon, Koi Fish, Pig | whole body | Animal-Identification-from-Video |
2022 | SealID: Saimaa Ringed Seal Re-Identification Dataset | Saimaa Ringed Seal | pelage patterns | SealID |
2021 | Giant Panda Identification | Giant Panda | discriminative local patches - they found panda’s eyes play a critical role in panda identification | iPanda-50 |
2021 | Individual dairy cow identification based on lightweight convolutional neural network | Cows | unspecific | Cow Dataset |
2021 | Towards Self-Supervision for Video Identification of Individual Holstein-Friesian Cattle: The Cows2021 Dataset | Holstein-Friesian Cattle | coat pattern | Cows2021 |
2020 | ATRW: A Benchmark for Amur Tiger Re-identification in the Wild | Amur Tiger | coat pattern | ATRW |
2020 | Re-Identification of Zebrafish using Metric Learning | Zebrafish | stripes, body pattern | AAU Zebrafish Re-Identification Dataset |
*be sure to check for an ArXiv version of any paper that is behind a paywall
further publications on the topic where the dataset could not be found and might not be public OR is only available on request
[a] GitHub available but dataset cannot be donwloaded
[b] please refer to [email protected] for early access
[c] to obtain a copy please directly contact: Michael C. Scholl, Save Our Seas Foundation (CEO), [email protected]
[d] dataset can be downloaded here but unfotunetely the .zip file cannot be extracted
Website | Dataset | Species | Identifying Feature |
---|---|---|---|
Sarasota Dolphin Research Program | GoMDIS | bottlenose dolphins | dorsal fin |
Cascadia Research Collective | Photo-ID Project | humpback and blue whale | flukes (humpback whale), dorsal fin (blue whale) |
OBIS-SEAMAP | various datasets | various species | - |
Species 360 | various data | various species | - |
Bay Cetology | Bigg’s killer whale photo-ID data | killer whale | - |
I'd like to highlight this outstanding publication from 2021 which gives a great overview of the entire topic Perspectives on Individual Animal Identification from Biology and Computer Vision
AIDE: Annotation Interface for Data-driven Ecology is a free tool from Microsoft for manually annotating images & also for training and running machine learning models. Both coupled in an active learning loop
Microsoft has also published the MegaDetector on GitHub. It detects animals, people, and vehicles in camera trap images