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

TL;DR

3D-FRONT addresses the need for large-scale, richly annotated synthetic indoor scenes by providing 18,968 textured rooms with professionally designed layouts and a large pool of style-consistent CAD models. The authors present a full construction pipeline—from room-suite generation using a visual-embedding recommender and GAE to layout optimization and viewpoint design—and release Trescope for rendering and annotation. Through rigorous recommender-system validation and extensive user studies, they demonstrate the dataset's value for interior scene synthesis and texture synthesis tasks, outperforming prior datasets like SUNCG in diversity and perceptual quality. Overall, 3D-FRONT enables data-driven design, robust scene understanding, and texture-aware rendering, with plans for richer content and industrial rendering tooling.

Abstract

We introduce 3D-FRONT (3D Furnished Rooms with layOuts and semaNTics), a new, large-scale, and comprehensive repository of synthetic indoor scenes highlighted by professionally designed layouts and a large number of rooms populated by high-quality textured 3D models with style compatibility. From layout semantics down to texture details of individual objects, our dataset is freely available to the academic community and beyond. Currently, 3D-FRONT contains 18,968 rooms diversely furnished by 3D objects, far surpassing all publicly available scene datasets. In addition, the 13,151 furniture objects all come with high-quality textures. While the floorplans and layout designs are directly sourced from professional creations, the interior designs in terms of furniture styles, color, and textures have been carefully curated based on a recommender system we develop to attain consistent styles as expert designs. Furthermore, we release Trescope, a light-weight rendering tool, to support benchmark rendering of 2D images and annotations from 3D-FRONT. We demonstrate two applications, interior scene synthesis and texture synthesis, that are especially tailored to the strengths of our new dataset. The project page is at: https://tianchi.aliyun.com/specials/promotion/alibaba-3d-scene-dataset.

Figures (19)

• Figure 1: 3D-FRONT is a new, large-scale, and comprehensive repository of synthetic indoor scenes with professionally and distinctively designed layouts, a large number (18,968) of rooms populated with 3D furniture objects that are stylistically compatible and endowed with high-quality textures. All freely available to the academic community and beyond.
• Figure 2: Pipeline of building 3D-FRONT. We start from an empty house with professional design ideas, create the room suites, optimize the layouts (e.g., to resolve artifacts highlighed in the red boxes), and finally verify the furnished rooms.
• Figure 3: Viewpoint Generation. Each scene is associated with several natural camera views to facilitate rendering.
• Figure 4: House Examples in 3D-FRONT. The left column shows the top-down views of three houses. The middle column presents several rooms contained in these houses, including bedrooms, living rooms, dining rooms, etc. The interior design ideas at the right column summarize the textured objects involved in the rooms and their high-quality 3D CAD models.
• Figure 5: Statistics of room numbers per house. 3D-FRONT contains 6,813 distinct houses constructed by 44,427 rooms. There are 6.5 rooms per house on average.