This is code that may be useful for people on the Machine Common Sense program.

Home for machinecommonsense.com Link to Homepage

The code in this directory creates placeholders / examples of ground truth and submissions. It is useful to see what a submission should look like and allows us to practice ingesting the data for scoring and visualizations. The ground truth information, like the submission data, is made up. Link to Ingest Results

The code in this directory makes the mask information for training data consistent. Link to Masks

Contains code and instructions for setting up the evaluation tools Link Evaluation Tool Setup

A labelling of the objects in the scene that are occluders. Link to Occluder Labeller

Contains the python api for interacting with our AI2Thor/Unity Testing Environment Link to Python API

Scripts for generating scenes of training data. Link to Scene Generator

Contains code for establishing the ground truth of the Intphy tests for Evaluation 1 Link to Truthing

Validating Evaluation 1 Submissions Link to Validation

Code in this repository is made available by Next Century Corporation under the Apache 2 Open Source License. You may freely download, use, and modify, in whole or in part, the source code or release packages. Any restrictions or attribution requirements are spelled out in the license file. For more information about the Apache license, please visit the The Apache Software Foundation’s License FAQ.