RTNs (Recurrent Transformer Networks)

Version 1.0 (11 Mar. 2019)

Contributed by Seungryong Kim (seungryong.kim@epfl.ch).

This code is written in MATLAB, and implements the RTNs [project website].

Dependencies

• Download [VLFeat] and [MatConvNet].
• Download the datasets:
  • [Proposal Flow Benchmark];
  • [PASCAL-2011 Part Dataset].

Pre-trained Models

• RTNs with synthetic training (in data/RTNs_synthetic) [RTNs_synthetic].
• RTNs on PF-PASCAL (in data/RTNs) [RTNs].

Getting started

• get_train_pascal_synthetic.m get the synthetic training data from PASCAL-VOC 2011 datasets.
• get_train_pf_pascal.m get the training data from PF-PASCAL datasets.
• train_pascal_synthetic.m trains the RTNs with the synthetic training data
• train_pf_pascal.m trains the RTNs on the PF-PASCAL training data
• test_pf_pascal.m tests the RTNs on the PF-PASCAL training data

Notes

• The code is provided for academic use only. Use of the code in any commercial or industrial related activities is prohibited.
• If you use our code, please cite the paper.

@InProceedings{kim2018nips,
author = {Seungryong Kim and Stephen Lin and Sangryul Jeon and Dongbo Min and Kwanghoon Sohn},
title = {Recurrent Transformer Networks for Semantic Correspondence},
booktitle = {Proceedings of the Neural Information Processing Systems (NeurIPS 2018)},
year = {2018}
}