Alexander G. Schwing, Raquel Urtasun
The data is originated from the joint layout estimation and semantic segmentation task (project page
). More specifically, they are the 101 test samples in the dataset.
As the input file is rather large, we provide 3 samples as examples. The full set of input files of 101 samples will be provided upon requests.
Download the 3 examples. The full set of input file will be provided upon requests.
Please see ReadMe for specification of the above input files.
We provide the C++ implementation for our approach with factor graph as the input representation:
Download code of the inference engine.
Please see ReadMe for detailed derivation better aligned with the implementation for input in factor graph representation.
The data and code were developed as part of the following publication:
Message Passing Inference for Large Scale Graphical Models with High Order Potentials
J. Zhang, A. G. Schwing, R. Urtasun.
Neural Information Processing Systems (NIPS), Montreal, Quebec, Canada, December 2014.
For questions regarding the data and code, please contact Jian Zhang