Welcome to the InfoSci Platform
Reference Hub1
Scene-Based Priors for Bayesian Semantic Image Segmentation

Scene-Based Priors for Bayesian Semantic Image Segmentation

Christopher Menart, James W. Davis, Muhammad N. Akbar, Roman Ilin
Copyright: © 2019 |Volume: 6 |Issue: 1 |Pages: 14
EISBN13: 9781522546795|ISSN: 2640-4079|EISSN: 2640-4087|DOI: 10.4018/IJSST.2019010101
Cite Article Cite Article

MLA

Menart, Christopher, et al. "Scene-Based Priors for Bayesian Semantic Image Segmentation." IJSST vol.6, no.1 2019: pp.1-14. http://doi.org/10.4018/IJSST.2019010101

APA

Menart, C., Davis, J. W., Akbar, M. N., & Ilin, R. (2019). Scene-Based Priors for Bayesian Semantic Image Segmentation. International Journal of Smart Security Technologies (IJSST), 6(1), 1-14. http://doi.org/10.4018/IJSST.2019010101

Chicago

Menart, Christopher, et al. "Scene-Based Priors for Bayesian Semantic Image Segmentation," International Journal of Smart Security Technologies (IJSST) 6, no.1: 1-14. http://doi.org/10.4018/IJSST.2019010101

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

Based on the observation that semantic segmentation errors are partially predictable, this study proposes a compact formula using the confusion statistics of a trained classifier to refine (re-estimate) the initial label hypotheses. The proposed strategy is contingent upon computing the classifier confusion probabilities for a given dataset and estimating a relevant prior on the object classes present in the image to be classified. This study provides a procedure to robustly estimate the confusion probabilities and explore multiple prior definitions. Experiments are shown comparing performances on multiple challenging datasets using different priors to improve a state-of-the-art semantic segmentation classifier. The study demonstrates the potential to significantly improve semantic labeling and motivates future work for reliable label prior estimation from images.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.