Alapfogalmak
FurniScene is a novel large-scale 3D room dataset with intricate furnishing scenes, surpassing existing datasets in both data volume and level of detail, particularly in the inclusion of numerous small furnishings to enhance realism.
Kivonat
FurniScene is a novel large-scale 3D room dataset proposed by the authors, which aims to address the limitations of existing indoor scene datasets. The dataset contains 111,698 meticulously designed rooms and 39,691 high-quality furniture CAD models, covering 89 different object types.
The key highlights of FurniScene are:
Meticulous Interior Design: The rooms in FurniScene are directly curated by interior design professionals, ensuring a high level of aesthetic quality and attention to detail.
Substantial Data Volume: FurniScene far exceeds existing datasets in terms of the number of rooms (111,698) and individual CAD models (39,691).
Diverse Room Furnishings: Each room in FurniScene contains an average of 14.4 objects, with a maximum of 119 objects, significantly more than comparable datasets. The dataset includes a wide variety of small decorative items, enhancing the realism of the scenes.
Rich Details: The rooms in FurniScene are highly detailed, with each one fully furnished and containing numerous decor items. This level of detail is unmatched by existing datasets.
In addition to the dataset, the authors also propose a Two-Step Diffusion Scene Model (TSDSM) for indoor scene generation, which outperforms several baseline methods on the FurniScene benchmark.
Statisztikák
The rooms in FurniScene contain an average of 14.4 objects, with a maximum of 119 objects.
Idézetek
"FurniScene showcases a striking level of diversity and intricacy, as shown in Fig. 1."
"To achieve real-life indoor scene generation, we present FurniScene, a novel large-scale 3D room dataset consisting of intricate furnishing scenes from interior design professionals."