Image Segmentation by Data-driven Markov Chain Monte Carlo

By Zhuowen Tu, Song-Chun Zhu, Harry Shum. 2002

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

On the Binding Problem in Artificial Neural Networks (2020, background): “The results that can be obtained by incorporating domain-specific knowledge are impressive (Zhao and Zhu, 2011; Del Pero et al., 2012, 2013; Tu et al., 2005; Tu and Zhu, 2002).”
Picture: A probabilistic programming language for scene perception (2015, background): “However, vision problems often lend themselves to much faster bottom-up inference based on data-driven proposals [19, 43].”
Using Multiple Segmentations to Discover Objects and their Extent in Image Collections (2006, methodology): “There are approaches in the literature for sampling likely segmentations [26] and multiscale segmentations [19].”
Approximate Bayesian Image Interpretation using Generative Probabilistic Graphics Programs (2013, background): “But like traditional bottom-up pipelines for vision, these approaches have relied on considerable problem-specific engineering, chiefly to design and/or learn custom inference strategies, such as MCMC proposals [18, 22] that incorporate bottom-up cues.”
Neural Scene De-rendering (2017, methodology): “Our framework also relates to the field of “vision as inverse graphics”, analysis-by-synthesis, or generative models with data-driven proposals [36, 37, 30, 20, 34, 16], as we are incorporating a graphics engine as a black-box synthesizer.”
Efficient Multilevel Brain Tumor Segmentation With Integrated Bayesian Model Classification (2008, unknown): “[23], [24], uses class models to explicitly represent the different heterogeneous processes.”
Patches, Planes and Probabilities: A Non-Local Prior for Volumetric 3D Reconstruction (2016, methodology): “To avoid this situation, in this work, we take advantage of a data-driven strategy [48] that generates samples from an initial 3D reconstruction of the scene.”
Empirical Minimum Bayes Risk Prediction: How to Extract an Extra Few % Performance from Vision Models with Just Three More Parameters (2014, methodology): “Two common techniques for producing multiple solutions in probabilistic models can be broadly characterized as follows: (1) M -best MAP algorithms [2,16,30] that find the topM most probable solutions and (2) sampling-based algorithms [1, 21, 25].”
Bayesian human segmentation in crowded situations (2003, background): “A random proposal probability will lead to very slow convergence rate while a proposal probability designed with domain knowledge ([18] [14] [7]) will make the Markov chain traverse the solution space more efﬁciently.”
Multi-task low-rank affinity pursuit for image segmentation (2011, background): “Generally, image segmentation is a comprehensive task which is related with several cues, e.g. regions, contours and textons [21].”
MCMC-based particle filtering for tracking a variable number of interacting targets (2005, methodology): “Intuitively, by using the image information, the detector provides a more informed way of adding targets, i.e., it is data-driven [ 38 ].”
Learning a classification model for segmentation (2003, methodology): “The DDMCMC [24] is a generative approach which builds explicit models of image regions.”
State-of-the-Art Methods for Brain Tissue Segmentation: A Review (2017, unknown): “[105], [106].”
Probabilistic models with unknown objects (2006, background): “However, it is often necessary to develop a special-purpose proposal distribution to achieve fast convergence on a particular task [Tu and Zhu, 2002; Pasula et al., 2003; Oh et al., 2004].”
Statistical Modeling and Conceptualization of Visual Patterns (2003, background): “Recent work include the data driven Markov chain Monte Carlo (DDMCMC) algorithms for image segmentation, parsing, and object recognition [82], [83], [98].”
I2T: Image Parsing to Text Description (2010, background): “In a series of work [14], [38]–[40], Zhu and his collaborators posed image parsing as a unified framework for image segmentation, object detection, and recognition.”