Add our code to parse 3D_Front dataset into Autodesk Maya

See the tutorial

Luminous is a framework for testing the performance of Embodied AI (EAI) models in indoor tasks. Generally, we intergrete different kind of functionalities into this repository that are related to evaluate EAI performance for indoor tasks. The Indoor Scene Synthesis module provides different methods for synthesize randomized indoor scenes that be visualized in Unity Engine. The Luminous for Alfred offers the complete pipline to conduct data augmentment such as generating Alfred tasks in Luminous, solving the trajectories of tasks, rendering images, and generating natural language descripitions.

We highlight the features of Luminous here and readers may go directly into the modules for detailed implementations.

We present Constrainted Stochastic Scene Generation in Luminous, which generates high-quality scene from the a graph-based representation of indoor scenes. By specifying the required objects and relationships between objects, we can sample an unlimited number of indoor scenes that satisfy the constraints.

We can generate Alfred trajectories from Luminous. After taking the task descriptions in Alfred, we apply the Task Parser to get the required indoor scenes. Then, a Task Sampler sample the tasks for Alfred challenge. Finally, we can add Natural Language Descriptions along with agent actions.

We evaluation three models on Luminous for Alfred challenge.

• MOCA
• ET
• HiTUT

See EAIEvaluation folder for more details.

Luminous
│   readme.md   
│
└─── EAIEvaluation
│   │   ET
│   │   MOCA
│   │   HiTUT
└───IndoorSceneSynthesis
│   │   3DFrontToolBox
│   │   ContrainedStochsticIndoorScene
│   │   LearningBasedMethod
│   │   └───3DSLN
│   │   └───DeepSynth
└───Luminous4Alfred
│   │   NaturalLanguageTool
│   │   TaskParser
│   │   TaskSampler
│   │   TaskSolver

EAIEvaluation

CSSG

NaturalLanguageTool

3DFrontToolBox