A CNN Cascade for Landmark Guided Semantic Part Segmentation

Abstract

This paper proposes a CNN cascade for semantic part segmentation guided by pose-specific information encoded in terms of a set of landmarks (or keypoints). There is large amount of prior work on each of these tasks separately, yet, to the best of our knowledge, this is the first time in literature that the interplay between pose estimation and semantic part segmentation is investigated. To address this limitation of prior work, in this paper, we propose a CNN cascade of tasks that firstly performs landmark localisation and then uses this information as input for guiding semantic part segmentation. We applied our architecture to the problem of facial part segmentation and report large performance improvement over the standard unguided network on the most challenging face datasets.

Proposed Architecture

Visual Example

Links

• Proceedings hosted by Springer.
• [pdf] Paper
• [tar.gz] approx 1GB. This archive contains the models and Python code to test the network. Some test images are included, but you are welcome to try your own. This code depends on several Python packages such as matplotlib, numpy and scipy. It is unlikely to work correctly on the master branch of caffe. Please use caffe-future.tar.gz which is a clone of longjon/caffe@future for safe keeping.

BibTeX

If you find this code or paper useful, please cite in using the reference below. If you use it for something outside of academia, I would love to hear how. Please email me to let me know

@inproceedings{jackson2016guided,
  title={A CNN Cascade for Landmark Guided Semantic Part Segmentation},
  author={Jackson, Aaron and Valstar, Michel and Tzimiropoulos, Georgios},
  booktitle={Proceedings of ECCV 2016 Workshops, Geometry meets Deep Learning},
  year={2016}
}