559,460 Video Clips - 50 Types of Dynamic Gesture Recognition Data
- 1,000 people
- 5 angles
- multiple scenes, multiple lighting conditions
Datatang has passed the certification of ISO27001 Information Security Management System and ISO9001 Quality Management System.
559,460 Video Clips - 50 Types of Dynamic Gesture Recognition Data is collected in indoor and outdoor. The data covers a broad age distribution from children to the elderly of male and female, mainly the teenager and the youth. The data includes multiple scenes, multiple directions, gestures of moving to left or right, forehand and backhand, rotation, etc. This data set can be used for tasks such as smart homes, audio equipment, and on-board systems.
- Data size
- a total of 559,460 video clips with 220,030 clips collected by laptop, and 339,430 clips collected by mobile phone.
- Population distribution
- Yellow-race (Chinese), gender distribution is male and female, age distribution is under 18, 18-40, and older than 40.
- Collection environment
- onsite collection, including indoor and outdoor scenes
- Collection diversity
- different illumination, scenes, distances, and angles, gestures of moving to left or right, forehand and backhand, rotation, etc.
- Collection device
- front and rear cameras of mobile phone and iPad, camera of laptop.
- video is in .mp4 or .mov or .wmv format.
- Collection content
- dynamic gestures of five different angles of the left and right hand were collected respectively.
- Accuracy rate
- basing on the motion accuracy in videos, the accuracy rate is over 97%. the video naming accuracy is over 97%