Back to Search
R

Replica

Dataset

Replica is a dataset of 18 highly photo-realistic 3D indoor scene reconstructions released by Facebook Reality Labs in 2019. Each scene consists of a dense mesh, high-resolution HDR textures, per-primitive semantic class and instance information, and planar mirror and glass reflectors. The scenes span room and building scale, including apartments, offices, hotels, and retail spaces. Replica was designed specifically to support machine learning research that relies on visually, geometrically, and semantically realistic generative models of the world. It has been adopted as a core dataset in the Habitat simulation platform, where it provides high-quality 3D environments for navigation, interaction, and rearrangement tasks. With 1,300+ GitHub stars, Replica is a foundational dataset for egocentric computer vision, semantic segmentation, geometric inference, and embodied AI research, particularly in the Habitat ecosystem.

Details

Updated:7/9/2026
sample count18
licenseReplica Dataset License (non-commercial research)
modality3D mesh, HDR texture, semantic/instance segmentation

Tags

photo-realistic3d-reconstructionsimulationhabitatembodied-ai

Relationships

Sources

The Replica Dataset: A Digital Replica of Indoor Spaces
paper
Visit
Replica GitHub Repository
github
Visit

Appears In

No related landscapes.