Download dataset
Free with an account
Files
About
The person keypoints of COCO 2017 val, keeping only the images whose Flickr owners chose a license that allows redistribution. Every person is a box plus the 17 COCO keypoints, drawn as a skeleton you can inspect in the annotator, filter, and export.
| Images | 605 from val2017 (413 CC BY 2.0, 189 CC BY-SA 2.0, 3 No known copyright restrictions) |
| Annotations | 1,609 people, each with a bounding box and 17 keypoints |
| Classes | 1 (person, with the 17 keypoint names and the COCO skeleton) |
| Keypoints | nose, left/right eye, ear, shoulder, elbow, wrist, hip, knee, ankle |
| Dataset | coco-person-keypoints-val2017-v1 |
Open any image to see each person's skeleton. To download everything (free account required):
pip install datatorch, then:datatorch login
datatorch pull datatorch/coco-person-keypoints-val2017
The subset itself is in COCO format at annotations/person_keypoints_val2017_open.json, with flickr_url and license on every image.
Built from the official person_keypoints_val2017.json. Kept: the 605 val2017 images with at least one keypointed person and a license of CC BY, CC BY-SA, No known copyright restrictions or United States Government Work (none of the val2017 keypoint images carry the last one). Dropped: images under NonCommercial or NoDerivs licenses, crowd regions, people with zero labeled keypoints, and the segmentation outlines. Keypoint coordinates, visibility flags (0 not labeled, 1 labeled but occluded, 2 labeled and visible) and boxes are the original values, and image filenames and annotation ids match the original release, so results stay comparable with published work.
Annotations: CC BY 4.0, COCO Consortium. Images: each keeps its Flickr owner's license, recorded per image in its file metadata (source) and in annotations/credits.csv, with the image's Flickr URL. The COCO Consortium does not own the images; use them under the Flickr Terms of Use and each image's license. If you use this data, please cite:
Lin, T.-Y., Maire, M., Belongie, S. et al. Microsoft COCO: Common Objects in Context. ECCV 2014.
Source: cocodataset.org.
Description
Person keypoint subset of COCO 2017 val with openly licensed images only: 605 images whose Flickr license is CC BY 2.0 (413), CC BY-SA 2.0 (189) or No known copyright restrictions (3), with 1,609 people annotated with the 17 COCO keypoints (nose, eyes, ears, shoulders, elbows, wrists, hips, knees, ankles), COCO visibility flags and bounding boxes. Every image is credited with its Flickr URL and license in its file metadata and in annotations/credits.csv. License: annotations CC BY 4.0 (COCO Consortium); each image keeps its Flickr owner's license. Source: https://cocodataset.org Please cite: Lin, T.-Y. et al., Microsoft COCO: Common Objects in Context. ECCV 2014.
Members
0Progress
100% labeled
605 of 607 files
0 B
Size
1
Datasets
0
Labels
0
Annotations