{
  "annotation_id": "57e39fab-1330-4daf-b673-162637c80dfc",
  "date_created": "2026-02-17T05:53:12.015000Z",
  "date_modified": "2026-02-17T05:53:12.015000Z",
  "file_hash": "de7b7c535e1d31b6e16dbf14ff008581422e5d0c7211af2195fb89ecfdf6dbb7",
  "private": false,
  "record": {
    "abstract": "We present SceneNAT, a single-stage masked non-autoregressive Transformer that synthesizes complete 3D indoor scenes from natural language instructions through only a few parallel decoding passes, offering improved performance and efficiency compared to prior state-of-the-art approaches. SceneNAT is trained via masked modeling over fully discretized representations of both semantic and spatial attributes. By applying a masking strategy at both the attribute level and the instance level, the model can better capture intra-object and inter-object structure. To boost relational reasoning, SceneNAT employs a dedicated triplet predictor for modeling the scene\u0027s layout and object relationships by mapping a set of learnable relation queries to a sparse set of symbolic triplets (subject, predicate, object). Extensive experiments on the 3D-FRONT dataset demonstrate that SceneNAT achieves superior performance compared to state-of-the-art autoregressive and diffusion baselines in both semantic compliance and spatial arrangement accuracy, while operating with substantially lower computational cost.",
    "arxiv_id": "2601.07218",
    "authors": [
      "Jeongjun Choi",
      "Yeonsoo Park",
      "H. Jin Kim"
    ],
    "categories": [
      "cs.CV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "title": "SceneNAT: Masked Generative Modeling for Language-Guided Indoor Scene Synthesis",
    "url": "https://arxiv.org/abs/2601.07218",
    "version": "v1"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "9ecffe63-2e41-44dc-8c5a-e6c852832aec",
    "id": "arXiv Dataset",
    "type": "Model",
    "variant": "snapshot-2026-01-17",
    "version": "0.1.0"
  },
  "user_id": 1000002
}