Open-source datasets for animal behavioral video analysis.
Reference first author/year |
Name for short |
Species | Experiment | Size pixel2 |
Resolution pixel/mm |
Length d/h/min/s |
Rate frame/s |
Link | Note |
---|---|---|---|---|---|---|---|---|---|
Pereira et al. 2022 | flies13 | fly | freely interacting pairs of virgin male and female | 1024x1024 | 30.3 | 30min | 200 | data demo |
labeled 2,000 frames with a 13 node skeleton |
mice_hc | mouse | pairs of male and female in a home cage with bedding | 1280x1024 | 1.9 | 5min | 40 | labeled 1,474 frames with a 5 node skeleton | ||
Lauer et al. 2022 | marmosets | marmoset | pairs of male and female (one had light blue dye) | 1080x1080 | 9h | 30 | data | used 7,600 labeled frames from 40 individuals with 15 body points | |
Sturman et al. 2020 | OFT | mouse | open field arena (45×45×40 cm) | 10min | data labels |
distance, time in center, supported rears and unsupported rears were recorded | |||
EPM | arms (65.5×5.5 cm) elevated 61.5 cm | distance, velocity, time in zone (open/closed arms + center) and head dips were recorded | |||||||
FST | beaker (20 cm diameter, 25 cm deep) filled to 17 cm water | 6min | |||||||
Liu et al. 2020 | a_common_hub | mouse | 30 | labeled ~3,000 random and 5,000 continuous masks, also with EEG and EMG recorded | |||||
Salem et al. 2016 | fisheye_3d | mouse | labeled 6,500 frames with 4 key-points |
Here we summarize some datasets for animal keypoints tracking
Reference first author/year |
Name for short |
Species | pictures_num | Keypoints_num | have_box | Link | Note |
---|---|---|---|---|---|---|---|
Yu et al. 2021 | AP-10K | mammals | 10015 | 17 | True | data | 23 animal families and 54 species |
Yao et al. 2021 | OMC | Monkey | 111,529 | 17 | True | data | 23 animal families and 54 species |
Reference first author/year |
Name for short |
Species | pictures_num | Keypoints_num | have_box | Link | Note |
---|---|---|---|---|---|---|---|
Cao et al. 2019 | Animal-Pose | dog, cat, cow, horse, sheep | 4,000+ | 20 | True | data | also contains bounding box annotations for other 7 animal categories |
Mathis et al. 2021 | Horse-10 | horse | 8,000+ | 22 | False | data | 30 diverse Thoroughbred horses |
Labuguen et al. 2020 | MacaquePose | macaque monkeys | 13,083 | 17 | True | data | 30 diverse Thoroughbred horses |
Pereira et al. 2019 | Vinegar Fly | Fly | 1349 | 32 | False | data | frame:192*192 |
Graving et al. 2019 | Desert Locust | Desert Locust | 69 | 35 | True | data | 30 diverse Thoroughbred horses |
Graving et al. 2019 | Grévy’s Zebra | Grévy’s Zebra | 809 | 9 | True | data | 30 diverse Thoroughbred horses |
Li et al. 2020 | ATRW | Amur tiger | 8000+ | 17 | True | data | the largest wildlife re-ID dataset to date |
Ng et al. 2022 | Animal Kingdom | Animal | 8000+ | 23 | True | data | 140 Action Description, List of >850 animals and classification、Action Recognition、Pose Estimation、Video Grounding |
Yang et al. 2022 | APT-36k | Animal | 36,000 | 17 | True | data | 2,400 video clips covering 30 different animal species from different scenes |
Please contribute if you think a new dataset or a relevant paper is missing.