HorSy15k is a fully annotated synthetic dataset designed for training horse/animal pose recognition models. It contains 15,000 synthetic images of horses, each with 20 annotated keypoints and a bounding box (bbox). Even obscured keypoints/joints are annotated.
- Fully Annotated: Each horse image in the dataset is annotated with 20 keypoints, providing precise information about the pose of the horse.
- Obscured Keypoints: Even when keypoints are obscured they are annotated in HorSy15k.
The images are in two ziped files, only because of git's file-size restrictions.