File size: 7,051 Bytes
a5a292f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2215715
a5a292f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
{
  "rai:dataLimitations": "The dataset focuses on controlled multi-person human-object interaction scenarios and may not generalize to unconstrained real-world environments. It primarily captures interactions involving a limited set of object categories and scripted behaviors. It is not recommended for safety-critical applications or deployment in uncontrolled real-world systems.",

  "rai:dataBiases": "The dataset may exhibit biases in interaction types, object categories, and participant roles. Certain interaction patterns (e.g., handover or cooperative tasks) may be overrepresented, while spontaneous or adversarial interactions are underrepresented. Demographic diversity is limited, as human subjects are represented through SMPL-X models rather than real-world population distributions.",

  "rai:personalSensitiveInformation": "The dataset does not contain real personal data. Human representations are synthetic SMPL-X models. Metadata includes limited attributes such as gender labels, which should not be interpreted as reflecting real individuals. No sensitive attributes such as health, political views, or identity are included.",

  "rai:dataUseCases": "The dataset is designed for research in human-object interaction modeling, motion generation, and embodied AI. It supports tasks such as text-to-motion generation, multi-person interaction modeling, and role-aware behavior synthesis. It is not intended for surveillance, identity inference, or real-world decision-making involving humans.",

  "rai:dataSocialImpact": "The dataset may contribute positively to research in robotics, human-computer interaction, and embodied intelligence. However, misuse could arise if models trained on this dataset are applied in real-world scenarios without accounting for distributional gaps or human diversity. Care should be taken to avoid overgeneralization and unfair deployment.",

  "rai:hasSyntheticData": true,

  "prov:wasDerivedFrom": [],

  "prov:wasGeneratedBy": [
    {
      "@type": "Process",
      "name": "Data Collection and Generation",
      "description": "Human motion data is represented using SMPL-X parameters. Interaction sequences are constructed through controlled capture and/or simulation pipelines involving predefined scenarios and object interactions."
    },
    {
      "@type": "Process",
      "name": "Data Annotation",
      "description": "Contact labels are generated using geometric proximity metrics (e.g., distance thresholds) and manually verified. Text annotations are written and augmented to describe interaction semantics."
    },
    {
      "@type": "Process",
      "name": "Preprocessing",
      "description": "Data is normalized, aligned, and converted into unified formats including SMPL-X motion, object transformation matrices, and contact labels. Additional prompt augmentation is applied to enrich textual diversity."
    }
  ],
  "@context": {
    "@language": "en",
    "cr": "https://mlcommons.org/croissant/",
    "rai": "https://mlcommons.org/croissant/rai/",
    "prov": "http://www.w3.org/ns/prov#",
    "sc": "https://schema.org/"
  },

  "@type": "sc:Dataset",
  "@id": "https://huggingface.co/datasets/RayZuo/CoRoleHOI",

  "sc:name": "CoRoleHOI",
  "sc:description": "CoRoleHOI is a dataset for role-asymmetric multi-person multi-object human-object interaction, including SMPL-X motion, object trajectories, contact annotations, and structured textual prompts.",
  "sc:url": "https://huggingface.co/datasets/RayZuo/CoRoleHOI",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "sc:version": "1.0",
  "sc:datePublished": "2026-01-01",
  "sc:citation": "To be added",

  "conformsTo": "https://mlcommons.org/croissant/1.1",

  "distribution": [
    {
      "@type": "cr:FileObject",
      "@id": "data-zip",
      "contentUrl": "https://huggingface.co/datasets/RayZuo/CoRoleHOI/resolve/main/CoRoleHOI.zip",
      "encodingFormat": "application/zip"
    }
  ],

  "recordSet": [

    {
      "name": "interaction_sequences",
      "description": "Each record corresponds to a multi-person multi-object interaction sequence.",

      "field": [

        {
          "name": "sequence_id",
          "dataType": "integer",
          "description": "Unique sequence identifier."
        },

        {
          "name": "text_annotation",
          "dataType": "string",
          "description": "Natural language description of the interaction."
        },

        {
          "name": "smplx_motion",
          "dataType": "array",
          "description": "SMPL-X human motion parameters for each participant.",
          "shape": ["T", "D"]
        },

        {
          "name": "object_transformation",
          "dataType": "array",
          "description": "Object trajectories represented as 4x4 transformation matrices.",
          "shape": ["T", 4, 4]
        },

        {
          "name": "contact_labels",
          "dataType": "array",
          "description": "Binary or semantic contact labels indicating human-object interactions.",
          "shape": ["T", "num_contacts"]
        },

        {
          "name": "human_metadata",
          "dataType": "object",
          "description": "Metadata of humans, including gender and identity."
        },

        {
          "name": "object_metadata",
          "dataType": "object",
          "description": "Metadata of objects, including category and semantic labels."
        }
      ]
    },

    {
      "name": "object_assets",
      "description": "Static object assets used in interaction sequences.",

      "field": [

        {
          "name": "object_id",
          "dataType": "string",
          "description": "Unique object identifier."
        },

        {
          "name": "mesh_file",
          "dataType": "string",
          "description": "Path to the object mesh (.obj file)."
        },

        {
          "name": "sdf_volume",
          "dataType": "array",
          "description": "Signed distance field representation of the object.",
          "shape": [256, 256, 256]
        }
      ]
    },

    {
  "name": "augmented_prompts",
  "dataType": "array",
  "description": "Augmented textual prompts for each interaction, providing paraphrases and semantic variations."
},
{
  "name": "prompt_dictionary",
  "dataType": "object",
  "description": "Mapping between prompt identifiers and textual descriptions."
},
{
  "name": "split_prompts",
  "dataType": "object",
  "description": "Predefined splits of prompts for training, validation, and testing."
}


  ],

  "keywords": [
    "Human-Object Interaction",
    "Multi-person",
    "Role-asymmetry",
    "SMPL-X",
    "Motion Generation",
    "3D Interaction"
  ],

  "creator": [
    {
      "name": "Lin Zuo",
      "affiliation": "MediaLab, Shanghai Jiao Tong University, China"
    }
  ],

  "citation": "To be added"
}