3D-FRONT: 3D Furnished Rooms with layOuts and semaNTics

Fu, Huan; Cai, Bowen; Gao, Lin; Zhang, Lingxiao; Wang, Jiaming; Li, Cao; Zeng, Qixun; Sun, Chengyue; Jia, Rongfei; Zhao, Binqiang; Zhang, Hao

doi:10.1109/iccv48922.2021.01075

Cited by 144 publications

(66 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our representation can obtain better reconstruction details while saving computational overhead. In this section, we illustrate the superiority of OctField on large scene dataset 3D-Front [16].…”

Section: Scene Reconstructionmentioning

confidence: 99%

“…In Figure 7, we present two camera views in a large scene from 3D-Front [16] dataset. From the visualization results, we can observe that our results is capable of capturing fine-grained geometric and structure details compared to LIG.…”

Section: Scene Reconstructionmentioning

confidence: 99%

“…We show that our method can provide more accurate reconstruction of geometric and structural details of large scenes. The experiment are performed on 3D-Front [16]. We evaluate our method in the task of shape completion.…”

Section: Scene Reconstructionmentioning

confidence: 99%

See 2 more Smart Citations

OctField: Hierarchical Implicit Functions for 3D Modeling

Tang¹,

Chen²,

Jie

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recent advances in localized implicit functions have enabled neural implicit representation to be scalable to large scenes. However, the regular subdivision of 3D space employed by these approaches fails to take into account the sparsity of the surface occupancy and the varying granularities of geometric details. As a result, its memory footprint grows cubically with the input volume, leading to a prohibitive computational cost even at a moderately dense decomposition. In this work, we present a learnable hierarchical implicit representation for 3D surfaces, coded OctField, that allows high-precision encoding of intricate surfaces with low memory and computational budget. The key to our approach is an adaptive decomposition of 3D scenes that only distributes local implicit functions around the surface of interest. We achieve this goal by introducing a hierarchical octree structure to adaptively subdivide the 3D space according to the surface occupancy and the richness of part geometry. As octree is discrete and non-differentiable, we further propose a novel hierarchical network that models the subdivision of octree cells as a probabilistic process and recursively encodes and decodes both octree structure and surface geometry in a differentiable manner. We demonstrate the value of OctField for a range of shape modeling and reconstruction tasks, showing superiority over alternative approaches.

show abstract

Section: Scene Reconstructionmentioning

confidence: 99%

Section: Scene Reconstructionmentioning

confidence: 99%

See 1 more Smart Citation

OctField: Hierarchical Implicit Functions for 3D Modeling

Tang¹,

Chen²,

Jie

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Datasets: We train our model on the 3D-FRONT dataset [21] which contains a collection of 6, 813 houses with roughly 14, 629 rooms, populated with 3D furniture objects from the 3D-FUTURE dataset [22]. In our evaluation, we focus on four room types: (i) bedrooms, (ii) living rooms, (iii) dining rooms and (iv) libraries.…”

Section: Experimental Evaluationmentioning

confidence: 99%

“…In this supplementary document, we provide a detailed overview of our network architecture and the training procedure. Subsequently, we describe the preprocessing steps that we followed to filter out problematic rooms from the 3D-FRONT dataset [21]. Next, we provide ablations on how different components of our system impact the performance of our model on the scene synthesis task and we compare ATISS with various transformer models that consider ordering.…”

Section: Supplementary Materials For Atiss: Autoregressive Transforme...mentioning

confidence: 99%

ATISS: Autoregressive Transformers for Indoor Scene Synthesis

Paschalidou,

Kar,

Shugrina

et al. 2021

Preprint

View full text Add to dashboard Cite

The ability to synthesize realistic and diverse indoor furniture layouts automatically or based on partial input, unlocks many applications, from better interactive 3D tools to data synthesis for training and simulation. In this paper, we present ATISS, a novel autoregressive transformer architecture for creating diverse and plausible synthetic indoor environments, given only the room type and its floor plan. In contrast to prior work, which poses scene synthesis as sequence generation, our model generates rooms as unordered sets of objects. We argue that this formulation is more natural, as it makes ATISS generally useful beyond fully automatic room layout synthesis. For example, the same trained model can be used in interactive applications for general scene completion, partial room rearrangement with any objects specified by the user, as well as object suggestions for any partial room. To enable this, our model leverages the permutation equivariance of the transformer when conditioning on the partial scene, and is trained to be permutation-invariant across object orderings. Our model is trained end-to-end as an autoregressive generative model using only labeled 3D bounding boxes as supervision. Evaluations on four room types in the 3D-FRONT dataset demonstrate that our model consistently generates plausible room layouts that are more realistic than existing methods. In addition, it has fewer parameters, is simpler to implement and train and runs up to 8x faster than existing methods.

show abstract