Standard DatasetsHuman Posture Classification Dataset
The "Human Posture Classification Dataset" is designed for applications in visual entertainment and robotics, consisting of a collection of indoor-collected images with high resolutions exceeding 3024 x 4032 pixels. This dataset emphasizes bounding box annotations and tagging to identify half-body portraits and classify them into 14 distinct types of poses, such as crossed hands, hands around the head, and one hand on the cheek, among others.
Sample
Specification
Dataset ID
MD-Image-080
Dataset Name
Human Posture Classification Dataset
Data Type
ImageVolume
About 17k
Data Collection
Indoor collected images. Resolution is over 3024 x 4032.
Annotation
Bounding Box,Tags
Annotation Notes
Collecting half-body portraits and classifying the poses, including 14 kinds of poses such as crossed hands, hands around the head, and one hand on the cheek,etc.
Application Scenarios
Robotics;Visual Entertainment
Data Collections
This dataset includes approximately 17k images, each annotated with a bounding box and tagged with specific pose classifications, capturing a wide range of human postures in half-body portraits. The detailed classification of poses provides valuable data for developing and training algorithms for posture recognition and analysis in various applications, from enhancing character animation and interaction in visual entertainment to improving human-robot interaction systems in robotics, where accurate interpretation of human postures can significantly enhance user experience and system responsiveness.
Quality Assurance
Related products
Any further information, please contact us.
