{
  "_asset_summary": {
    "asset_id": "HSSD:003949689122adfc2370d64c0615edc1b257db83",
    "dataset": "HSSD",
    "category": "rug.n.01",
    "synset": "rug.n.01",
    "source_asset_id": "003949689122adfc2370d64c0615edc1b257db83",
    "name": "60X96 Rug, Taiga Ivory/Black",
    "geometry_path": "/data/share/ud4scenesmith/ud4scenesmith_3_2_final_20260513/materialized_assets/static_expansion_full/HSSD_official_20260519/rug.n.01/003949689122adfc2370d64c0615edc1b257db83/files/model.glb",
    "npz": "assets/hssd_003949689122adfc2370d64c0615edc1b257db83/affordance_masks.npz",
    "labels_json": "assets/hssd_003949689122adfc2370d64c0615edc1b257db83/labels.json",
    "num_points": 2048,
    "num_mesh_parts": 1,
    "label_count": 4,
    "mean_mask_ratio": 0.197021484375,
    "min_mask_ratio": 0.0869140625,
    "max_mask_ratio": 0.39453125
  },
  "labels": [
    {
      "label": "support",
      "prompt_label": "bear",
      "instruction": "Please locate the areas of the object with function of bear. Identify the areas of the rug that can support a person standing or walking (the top surface).",
      "mode": "official_3dadllm",
      "mask_points": 808,
      "mask_ratio": 0.39453125,
      "text": "system|> You are a helpful assistant. <|end|> <|user|> Please locate the areas of the object with function of bear. Identify the areas of the rug that can support a person standing or walking (the top surface). <|end|> <|assistant|> Sure,it is  <|placeholder4|> . ",
      "proposal": {
        "canonical_label": "support",
        "query_text": "Identify the areas of the rug that can support a person standing or walking (the top surface).",
        "source": "seqafford_style_llm",
        "raw_source_label": "A rug provides a surface that can support a person standing or walking on it.",
        "confidence": 0.72,
        "evidence": {
          "llm_model": "gpt-5.2",
          "metadata_evidence": [
            "category: rug.n.01",
            "name: 60X96 Rug, Taiga Ivory/Black"
          ],
          "basis": "SeqAfford-style metadata/render-context instruction proposal"
        }
      }
    },
    {
      "label": "place_on",
      "prompt_label": "bear",
      "instruction": "Please locate the areas of the object with function of bear. Segment the rug surface where objects could be placed on top (e.g., shoes, bags, small items).",
      "mode": "official_3dadllm",
      "mask_points": 178,
      "mask_ratio": 0.0869140625,
      "text": "system|> You are a helpful assistant. <|end|> <|user|> Please locate the areas of the object with function of bear. Segment the rug surface where objects could be placed on top (e.g., shoes, bags, small items). <|end|> <|assistant|> Sure,it is  <|placeholder4|> . ",
      "proposal": {
        "canonical_label": "place_on",
        "query_text": "Segment the rug surface where objects could be placed on top (e.g., shoes, bags, small items).",
        "source": "seqafford_style_llm",
        "raw_source_label": "Flat items can be placed on top of a rug on the floor.",
        "confidence": 0.6,
        "evidence": {
          "llm_model": "gpt-5.2",
          "metadata_evidence": [
            "category: rug.n.01",
            "synset: rug.n.01"
          ],
          "basis": "SeqAfford-style metadata/render-context instruction proposal"
        }
      }
    },
    {
      "label": "move",
      "prompt_label": "reposition",
      "instruction": "Please locate the areas of the object with function of reposition. Find the parts of the rug a person would grasp or pull to move/reposition it (edges/corners).",
      "mode": "official_3dadllm",
      "mask_points": 304,
      "mask_ratio": 0.1484375,
      "text": "system|> You are a helpful assistant. <|end|> <|user|> Please locate the areas of the object with function of reposition. Find the parts of the rug a person would grasp or pull to move/reposition it (edges/corners). <|end|> <|assistant|> Sure,it is  <|placeholder4|> . ",
      "proposal": {
        "canonical_label": "move",
        "query_text": "Find the parts of the rug a person would grasp or pull to move/reposition it (edges/corners).",
        "source": "seqafford_style_llm",
        "raw_source_label": "A rug can be moved by lifting, dragging, or repositioning it.",
        "confidence": 0.55,
        "evidence": {
          "llm_model": "gpt-5.2",
          "metadata_evidence": [
            "category: rug.n.01",
            "non-articulated soft furnishing implied by rug"
          ],
          "basis": "SeqAfford-style metadata/render-context instruction proposal"
        }
      }
    },
    {
      "label": "sit",
      "prompt_label": "take a seat",
      "instruction": "Please locate the areas of the object with function of take a seat. Mark the parts of the rug where a person could sit on it (the upper surface).",
      "mode": "official_3dadllm",
      "mask_points": 324,
      "mask_ratio": 0.158203125,
      "text": "system|> You are a helpful assistant. <|end|> <|user|> Please locate the areas of the object with function of take a seat. Mark the parts of the rug where a person could sit on it (the upper surface). <|end|> <|assistant|> Sure,it is  <|placeholder4|> . ",
      "proposal": {
        "canonical_label": "sit",
        "query_text": "Mark the parts of the rug where a person could sit on it (the upper surface).",
        "source": "seqafford_style_llm",
        "raw_source_label": "A rug can be sat on when placed on the floor.",
        "confidence": 0.48,
        "evidence": {
          "llm_model": "gpt-5.2",
          "metadata_evidence": [
            "category: rug.n.01",
            "name implies a floor rug (60X96)"
          ],
          "basis": "SeqAfford-style metadata/render-context instruction proposal"
        }
      }
    }
  ],
  "mesh_meta": [
    {
      "mesh": "/data/share/ud4scenesmith/ud4scenesmith_3_2_final_20260513/materialized_assets/static_expansion_full/HSSD_official_20260519/rug.n.01/003949689122adfc2370d64c0615edc1b257db83/files/model.glb",
      "sampled_points": 2048,
      "area": 7.526202385839932,
      "vertices": 104,
      "faces": 204
    }
  ]
}