Synthesizing Environment-Aware Activities via Activity Sketches

This is the official implementation of ResActGraph (CVPR2019).

Synthesizing Environment-Aware Activities via Activity Sketches

Yuan-Hong Liao∗, Xavier Puig∗, Marko Boben, Antonio Torralba, Sanja Fidler

If you find the code useful in your research, please consider citing:

@InProceedings{Liao_2019_CVPR,
author = {Liao, Yuan-Hong and Puig, Xavier and Boben, Marko and Torralba, Antonio and Fidler, Sanja},
title = {Synthesizing Environment-Aware Activities via Activity Sketches},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}

Contents

• Introduction
• Environment Setup
• Training

Introduction

In order to perform activities from demonstrations or descriptions, agents need to distill what the essense of the given activity is. In this work, we address the problem of environment-aware program generation. Given a visual demonstration or a description of an activity, we generate program sketches representing the essential instructions and propose a model, ResActGraph, to transform these into full programs representing the actions needed to perform the activity under the presented environmental constraints.

Demo

Here is one short clip where the agent is chilling out in his living room.

Environment Setup

Create a virtual environment (Optional)

$ virtualenv -p python3 virtualhome
$ source virtualhome/bin/activate
(virtualhome) $ git clone https://github.com/andrewliao11/env-aware-program-gen.git
(virtualhome) $ cd env-aware-program-gen
(virtualhome) $ pip3 install -r requirements.txt

Install VirtualHome

To execute or evalutate the sampled programs, VirtualHome need to be installed. Please see here for the installation.

Data

Download the program dataset here.

Here is how the dataset structure should look like:

dataset
└── VirtualHome-Env
    ├── augment_programs
    │   ├── augment_exception
    │   └── augment_location
    ├── demonstration
    │   ├── images
    │   └── images_augment
    ├── original_programs
    ├── resources
    │   ├── class_name_equivalence.json
    │   ├── knowledge_base.npz
    │   ├── object_merged.json
    │   ├── object_prefabs.json
    │   └── object_script_placing.json
    ├── sketch_annotation.json
    └── split
        ├── test_progs_paths.txt
        └── train_progs_paths.txt

Training

The training of the program/sketch generation model is documneted here