Welcome to the CALVIN research group homepage. We are part of the IPAB institute of the University of Edinburgh.
We research several topics related to learning in computer vision.
Our current focus is on weakly supervised learning of object classes, semantic segmentation, and large-scale auto-annotation.
- One paper on Closed-Form Training of Conditional Random Fields for Large Scale Image Segmentation submitted to arXiv.org. The paper is available online.
- One paper on ImageNet auto-annotation accepted to CVPR 2014. The camera-ready version will be online soon. Object bounding-boxes output by this method are available
- Technical report (to appear in IJCV) and its results improving our ECCV 2012 paper on segmentation propagation in ImageNet released.
- Release of Fast Video Segmentation v1.1. This version includes scripts to reproduce our results on SegTrack.
- Release of objectness V2.2. This version runs faster.
- One paper on fast video object segmentation accepter to ICCV 2013. The camera-ready version is available online. The source code is available here.
- One paper on global shape priors for segmentation accepted to BMVC 2013. The camera-ready version is available online. This work also won the Best Poster Award at BMVC 2013.
- Vittorio Ferrari's invited talk at Google about large-scale object localization and segmentation
- One paper on energy minimization accepted to CVPR 2013. The camera-ready version is available online, as well as the code.
- Best Paper Award at ECCV 2012 for our paper Segmentation Propagation in ImageNet.
- Matthieu Guillaumin received Outstanding Reviewer awards at both BMVC 2012 and ECCV 2012.
- Two CVPR 2012 papers in the top list of a Google reporter
- Vittorio Ferrari received an ERC Starting Grant.