Document Detail


Learning Hierarchical Features for Scene Labeling.
MedLine Citation:
PMID:  23091265     Owner:  NLM     Status:  Publisher    
Abstract/OtherAbstract:
Scene labeling consists in labeling each pixel in an image with the category of the object it belongs to. We propose a method that uses a multiscale convolutional network trained from raw pixels to extract dense feature vectors that encode regions of multiple sizes centered on each pixel. The method alleviates the need for engineered features, and produces a powerful representation that captures texture, shape and contextual information. We report results using multiple post-processing methods to produce the final labeling. Among those, we propose a technique to automatically retrieve, from a pool of segmentation components, an optimal set of components that best explain the scene; these components are arbitrary, e.g. they can be taken from a segmentation tree, or from any family of over-segmentations. The system yields record accuracies on the Sift Flow Dataset (33 classes) and the Barcelona Dataset (170 classes) and near-record accuracy on Stanford Background Dataset (8 classes), while being an order of magnitude faster than competing approaches, producing a 320x240 image labeling in less than a second, including feature extraction.
Authors:
Clement Farabet; Camille Couprie; Laurent Najman; Yann Lecun
Related Documents :
16638655 - Imaging findings in mucin-producing carcinoma of the gallbladder.
25029245 - Optimal uniformity index selection and acquisition counts for daily gamma camera qualit...
23509445 - Motion detection in diffusion mri via online odf estimation.
9724445 - Pediatric pancreas: an overview.
18833485 - Size and shape variability in the skull of myotis nigricans (schinz, 1821) (chiroptera:...
1957245 - Disc edema, transient obscurations of vision, and a temporal fossa mass.
Publication Detail:
Type:  JOURNAL ARTICLE     Date:  2012-10-17
Journal Detail:
Title:  IEEE transactions on pattern analysis and machine intelligence     Volume:  -     ISSN:  1939-3539     ISO Abbreviation:  IEEE Trans Pattern Anal Mach Intell     Publication Date:  2012 Oct 
Date Detail:
Created Date:  2012-10-23     Completed Date:  -     Revised Date:  -    
Medline Journal Info:
Nlm Unique ID:  9885960     Medline TA:  IEEE Trans Pattern Anal Mach Intell     Country:  -    
Other Details:
Languages:  ENG     Pagination:  -     Citation Subset:  -    
Affiliation:
New York University, New York and Universite Paris-Est, Paris.
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine


Previous Document:  Inflammation in abdominal aortic aneurysms: cellular infiltrate and cytokine profiles.
Next Document:  Depth-Fighting Aware Methods for Multi-Fragment Rendering.