Mining 3D Indoor Scenes with Human Motion

Home

Discover

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and Machine Learning and more.Subscribe

MIME: Human-Aware 3D Scene Generation

We propose a generative model of indoor scenes that produces furniture layouts that are consistent with the human movement.

To train the model, we build a dataset by populating the 3d front scene dataset with 3d humans.

Our experiments show that the model produces more diverse and plausible 3dscenes than a recent generative scene method that does not know about humanmovement.

Authors

Hongwei Yi, Chun-Hao P. Huang, Shashank Tripathi, Lea Hering, Justus Thies, Michael J. Black

Related Topics

Auto-regressive transformer architecture

Generative model of indoor scenes

Pose2room @ cite7

Pose-conditioned generative model

3d human body models

Posa @ cite1

Atiss @ cite6

3d scene generation baselines

Transformer-based auto-regressive 3d scene generation method

Motion-conditioned generative model

Introduction

We train a new method called (mining interaction and movement to infer 3d environments) that generates plausible indoor 3d scenes based on 3d human motion.

The key intuitions are that (1) a human s motion through free space indicates the lack of objects, effectively regions of the scene that are free of furniture, and (2) when they are in contact with the scene, this constrains both the type and placement of 3d objects ; e.g, a sitting human must be sitting on something, such as a chair, a sofa, a bed, etc.

To fulfill the free-space constraints induced by the human motion.

To make these intuitions concrete, we develop, which is a transformer-based auto-regressive 3d scene generation method that, given an empty floor plan and a human motion sequence, predicts the furniture that is in contact with it.

It also predicts plausible objects that have no contact with the human but that fit with the other objects and respect both the contact and free-space constraint induced by the motion.

Result

Human motion capture data has many applications, particularly for generating synthetic training data at scale.

We introduce a method called, by populating humans into the large-scale synthetic scene dataset @cite0.0.0, that generates realistic 3d scenes that are consistent with input human motion and contacts.

Our method can generate multiple realistic scenes, where the input motion can take place, by incorporating input human motion into free space and contact boxes.

Read the Paper

Read Paper

Bookmarks

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and Machine Learning and more.Subscribe

Mining 3D Indoor Scenes with Human Motion

MIME: Human-Aware 3D Scene Generation

We propose a generative model of indoor scenes that produces furniture layouts that are consistent with the human movement.

To train the model, we build a dataset by populating the 3d front scene dataset with 3d humans.

Our experiments show that the model produces more diverse and plausible 3dscenes than a recent generative scene method that does not know about humanmovement.

Authors

Hongwei Yi, Chun-Hao P. Huang, Shashank Tripathi, Lea Hering, Justus Thies, Michael J. Black

Related Topics

Auto-regressive transformer architecture

Generative model of indoor scenes

Pose2room @ cite7

Pose-conditioned generative model

3d human body models

Posa @ cite1

Atiss @ cite6

3d scene generation baselines

Transformer-based auto-regressive 3d scene generation method

Motion-conditioned generative model

Introduction

We train a new method called (mining interaction and movement to infer 3d environments) that generates plausible indoor 3d scenes based on 3d human motion.

To fulfill the free-space constraints induced by the human motion.

It also predicts plausible objects that have no contact with the human but that fit with the other objects and respect both the contact and free-space constraint induced by the motion.

Result

Human motion capture data has many applications, particularly for generating synthetic training data at scale.

Our method can generate multiple realistic scenes, where the input motion can take place, by incorporating input human motion into free space and contact boxes.

Read the Paper

Read Paper

Home

Discover

Bookmarks

◐ Latest Activity

◐ Recommended Follows