{
  "slug": "scenebind-semantic-spatial-scene-embeddings-715265",
  "title": "SceneBind: Semantic-Spatial Scene Embeddings",
  "dek": "SceneBind combines a global semantic embedding with object-centric semantic-spatial slots for scene retrieval, object grounding, and audio-visual localization.",
  "summary": "SceneBind adds semantic-spatial scene modeling across vision, audio, and language using global embeddings, object-centric slots, and object alignment.",
  "tags": [
    "SceneBind",
    "omni-modal learning",
    "semantic-spatial representation",
    "3D spatial understanding",
    "vision audio language",
    "cross-modal scene retrieval",
    "object grounding",
    "audio-visual localization"
  ],
  "published_at": "2026-07-18T23:06:37.42+00:00",
  "grade": 8.3,
  "agent_utility": 8,
  "price_usdc": 0.247,
  "stats": {
    "claims": 15,
    "entities": 17,
    "faq": 8,
    "downloads": 0,
    "paid_downloads": 0
  },
  "preview": {
    "faq_questions": [
      "What is SceneBind?",
      "What problem does SceneBind address?",
      "How does SceneBind represent a scene?",
      "What is SceneBind Matching?",
      "Which modalities are covered?",
      "What results are claimed?",
      "Are numerical results available from the abstract?",
      "Does SceneBind work with pretrained encoders?"
    ],
    "entity_names": [
      {
        "name": "SceneBind",
        "type": "model"
      },
      {
        "name": "SceneBind Matching",
        "type": "technique"
      },
      {
        "name": "Global semantic embedding",
        "type": "representation component"
      },
      {
        "name": "Object-centric semantic-spatial slots",
        "type": "representation component"
      },
      {
        "name": "Semantic-spatial entity",
        "type": "concept"
      },
      {
        "name": "Vision",
        "type": "modality"
      },
      {
        "name": "Audio",
        "type": "modality"
      },
      {
        "name": "Language",
        "type": "modality"
      },
      {
        "name": "Real-world binaural audio-visual dataset",
        "type": "dataset"
      },
      {
        "name": "Structured semantic and spatial annotations",
        "type": "annotation resource"
      },
      {
        "name": "Training protocol for aligning semantic and spatial signals",
        "type": "training method"
      },
      {
        "name": "Large-scale pretrained semantic encoders",
        "type": "model family"
      },
      {
        "name": "Cross-modal scene retrieval",
        "type": "task"
      },
      {
        "name": "Object grounding",
        "type": "task"
      },
      {
        "name": "Scene retrieval",
        "type": "task"
      },
      {
        "name": "Spatial retrieval",
        "type": "task"
      },
      {
        "name": "Audio-visual localization",
        "type": "task"
      }
    ],
    "related_work_titles": [
      "Existing omni-modal encoders",
      "Large-scale pretrained semantic encoders",
      "Object-centric semantic-spatial representations",
      "Semantic-spatial matching for retrieval and grounding",
      "Audio-visual localization as a downstream task"
    ],
    "application_industries": [
      "Multimodal AI research",
      "Scene retrieval research",
      "Grounding research",
      "Audio-visual learning research"
    ],
    "glossary_terms": [
      "omni-modal representation",
      "semantics",
      "3D spatial understanding",
      "global semantic embedding",
      "object-centric slot",
      "semantic-spatial slot",
      "spatial attributes",
      "uncertainty",
      "semantic-spatial matching",
      "global scene similarity",
      "object alignment",
      "cross-modal scene retrieval",
      "object grounding",
      "binaural audio",
      "structured semantic and spatial annotations",
      "aligning semantic and spatial signals",
      "pretrained semantic encoder",
      "lightweight spatial modeling",
      "spatial retrieval",
      "zero-shot transfer",
      "audio-visual localization"
    ]
  },
  "purchase": {
    "protocol": "x402",
    "version": 2,
    "network": "eip155:8453",
    "asset": "USDC",
    "price": 0.247,
    "full_markdown_url": "https://x402.disruptive-concepts.com/api/public/article/scenebind-semantic-spatial-scene-embeddings-715265/full",
    "full_json_url": "https://x402.disruptive-concepts.com/api/public/article/scenebind-semantic-spatial-scene-embeddings-715265/full?format=json"
  }
}