0%

NTU-RGBD骨架数据分析

Posted on In Valine:

NTU-RGBD Dataset 是目前非常出名的关于人体动作采集的数据集，来自于南洋理工，最近在研究人体动作分类的相关问题，以下是对数据集中骨架数据的分析

骨架文件按以下格式命名：
S001C001P001R001A001
其中：
A001表示第一种动作，共60种
P001表示一号动作执行人，但并非每个人都执行了所有动作
C001表示一号相机视角，共三个视角

每个skeleton文件第一行的数字为该骨架序列的总帧数，从第二行开始分别为每一帧的信息
对于其中每一帧，第一个数字为当前帧body数量（如1或2）
body数量的下一行十个数，形如72057594037931101 0 1 1 1 1 0 0.02764709 0.05745083 为body_info，分别包含了以下信息：

1
2
3
4
5
body_info_key = [
                    'bodyID', 'clipedEdges', 'handLeftConfidence',
                    'handLeftState', 'handRightConfidence', 'handRightState',
                    'isResticted', 'leanX', 'leanY', 'trackingState'
                ]

再往下一行，数字25，代表骨架joint数量
下面25行12列数据就是这25个joints的信息joint_info，如下：

1
2
3
4
5
joint_info_key = [
                     'x', 'y', 'z', 'depthX', 'depthY', 'colorX', 'colorY',
                     'orientationW', 'orientationX', 'orientationY',
                     'orientationZ', 'trackingState'
                 ]

附上读取处理骨架数据的代码
感谢代码来源：

1
2
3
4
5
6
@inproceedings{stgcn2018aaai,
  title     = {Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition},
  author    = {Sijie Yan and Yuanjun Xiong and Dahua Lin},
  booktitle = {AAAI},
  year      = {2018},
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
def read_skeleton(file):
    with open(file, 'r') as f:
        skeleton_sequence = {}
        skeleton_sequence['numFrame'] = int(f.readline())
        skeleton_sequence['frameInfo'] = []
        for t in range(skeleton_sequence['numFrame']):
            frame_info = {}
            frame_info['numBody'] = int(f.readline())
            frame_info['bodyInfo'] = []
            for m in range(frame_info['numBody']):
                body_info = {}
                body_info_key = [
                    'bodyID', 'clipedEdges', 'handLeftConfidence',
                    'handLeftState', 'handRightConfidence', 'handRightState',
                    'isResticted', 'leanX', 'leanY', 'trackingState'
                ]
                body_info = {
                    k: float(v)
                    for k, v in zip(body_info_key, f.readline().split())
                }
                body_info['numJoint'] = int(f.readline())
                body_info['jointInfo'] = []
                for v in range(body_info['numJoint']):
                    joint_info_key = [
                        'x', 'y', 'z', 'depthX', 'depthY', 'colorX', 'colorY',
                        'orientationW', 'orientationX', 'orientationY',
                        'orientationZ', 'trackingState'
                    ]
                    joint_info = {
                        k: float(v)
                        for k, v in zip(joint_info_key, f.readline().split())
                    }
                    body_info['jointInfo'].append(joint_info)
                frame_info['bodyInfo'].append(body_info)
            skeleton_sequence['frameInfo'].append(frame_info)
    return skeleton_sequence


def read_xyz(file, max_body=2, num_joint=25):
    seq_info = read_skeleton(file)
    data = np.zeros((3, seq_info['numFrame'], num_joint, max_body)) # (3,frame_nums,25 2)
    for n, f in enumerate(seq_info['frameInfo']):
        for m, b in enumerate(f['bodyInfo']):
            for j, v in enumerate(b['jointInfo']):
                if m < max_body and j < num_joint:
                    data[:, n, j, m] = [v['x'], v['y'], v['z']]
                else:
                    pass
    return data