Download icon

Dataset

20,000 simulation scenarios
RGB, segmentation and depth modalities
Detailed annotation of (in)stability

View details »

Github icon

Source Code

Data loader for Tensorflow dataset API
Script examples for stability prediction
Pre-trained stability predictor

View details »

Generic placeholder image

Tech Report

Dataset construction & details
Intuitive physics of stability prediction
Application to object stacking

View details »



Source Code

Handling of the ShapeStacks data with Tensorflow dataset API.
Example scripts for training and evaluation of stablity prediction models.
Check out the code on GitHub »

Pre-trained stability prediction models based on InceptionV4 (1.8 GB):
shapestacks-incpv4.tar.gz · shapestacks-incpv4.md5

Code example of the stability predictor

Tech Report

Physical intuition is pivotal for intelligent agents to perform complex tasks. In this paper we investigate the passive acquisition of an intuitive understanding of physical principles as well as the active utilisation of this intuition in the context of generalised object stacking. To this end, we provide ShapeStacks: a simulation-based dataset featuring 20,000 stack configurations composed of a variety of elementary geometric primitives richly annotated regarding semantics and structural stability.
[paper] · [poster] · [video]

Teaser figure of ShapeStacks paper
Funding of the ShapeStacks project

Research funded by

ERC 677195-IDIU
AIMS-CDT