metadata

license: gpl-3.0
tags:
  - human-pose-estimation
  - computer-vision
  - person-detection
  - amodal-person-detection
size_categories:
  - 1K<n<10K
task_categories:
  - keypoint-detection
  - object-detection
language:
  - en

CropCOCO Dataset

CropCOCO is a validation-only dataset of COCO val 2017 images cropped such that some keypoints annotations are outside of the image. It can be used for keypoint detection, out-of-image keypoint detection and localization, person detection and amodal person detection.

📦 Dataset Details

Total images: 4,114
Annotations: COCO-style (bounding boxes, human keypoints, both in and out-of-image)
Resolution: Varies
Format: JSON annotations + JPG images

🔍 Tasks

Human Pose Estimation
Person Detection
Amodal Person Detection

📄 Citation

If you use CropCOCO in your research, please cite:

@inproceedings{probpose2025,
  title={{ProbPose: A Probabilistic Approach to 2D Human Pose Estimation}}, 
  author={Miroslav Purkrabek and Jiri Matas},
  year={2025},
  booktitle={Computer Vision and Pattern Recognition (CVPR)},
}

🧑‍💻 Authors

Miroslav Purkrabek (personal website)
Jiri Matas (personal website)