Dense Semantic Correspondence Where Every Pixel is a Classifier
By Hilton Bristow, Jack Valmadre, Simon Lucey. 2015
57 citations recorded by the source corpus.
Determining dense semantic correspondences across objects and scenes is a difficult problem that underpins many higher-level computer vision algorithms. Unlike canonical dense correspondence problems which consider images that are spatially or temporally adjacent, semantic correspondence is characterized by images that share similar high-level structures whose exact appearance and geometry may differ. Motivated by object recognition literature and recent work on rapidly estimating linear classifiers, we treat semantic correspondence as a constrained detection problem, where an exemplar LDA classifier is learned for each pixel. LDA classifiers have two distinct benefits: (i) they exhibit higher average precision than similarity metrics typically used in correspondence problems, and (ii) unlike exemplar SVM, can output globally interpretable posterior probabilities without calibration, whilst also being significantly faster to train. We pose the correspondence problem as a graphical model, where the unary potentials are computed via convolution with the set of exemplar classifiers, and the joint potentials enforce smoothly varying correspondence assignment.
Subfields: Computer Vision and Pattern Recognition
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Andrea Vedaldi
- Luc Van Gool
- Andrew Zisserman
- Jitendra Malik
- Dacheng Tao
- Ming-Hsuan Yang
- Alan L. Yuille
- Xiaogang Wang
Citations in the corpus
- DensePose: Dense Human Pose Estimation in the Wild (2018, background): “Several other works have recently aimed at recovering dense correspondences between pairs [3] or sets of RGB images [50, 10] in an unsupervised setting.”
- Self-Supervised Learning of Geometrically Stable Features Through Probabilistic Introspection (2018, background): “Yet, without any further training, they have been shown not to improve over hand-engineered features for geometric tasks [21, 8] and most approaches still combine hand-crafted features and spatial regularization [3, 13, 17, 19, 43].”