- CMU mocap (Carnegie Mellon University – CMU Graphics Lab – motion capture library)
This database includes 2605 trials in 6 categories and 23 subcategories. The categories include: human interaction, interaction with environment, locomotion, physical activities and sports, situations and scenarios and test sequences. People in the videos are marked in 41 points and the videos are captured using 12 infrared cameras. This database is mostly used for human activity recognition but it can also be used for estimating of human pose.
- Human Eva (Brown University) This database is somehow the best existing database for human pose estimation. The main purpose of this database is human pose estimation and human tracking and is published for public use from 2006. The HumanEva-I dataset contains 7 calibrated video sequences (4 grayscale and 3 color) that are synchronized with 3D body poses obtained from a motion capture system. The database contains 4 subjects performing a 6 common actions (e.g. walking, jogging, gesturing, etc.). The error metrics for computing error in 2D and 3D pose are provided to participants. The dataset contains training, validation and testing (with withheld ground truth) sets. Recently the HumanEva-II dataset is also published.
- Hedvig Kjellström(KTH University)
The strait walking video of this database might be the oldest and most commonly used video for human pose estimation and human tracking. The database includes four video’s of about 50 to 175 frames and multiple picture collections. The videos are not labeled but the main limbs (including head, trunk,feed and hands) are labeled in the pictures.