A hierarchical probabilistic model for rapid object categorization in natural scenes

Xiaofu He, Zhiyong Yang, Joseph Zhuo Tsien

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Humans can categorize objects in complex natural scenes within 100-150 ms. This amazing ability of rapid categorization has motivated many computational models. Most of these models require extensive training to obtain a decision boundary in a very high dimensional (e.g., ~6,000 in a leading model) feature space and often categorize objects in natural scenes by categorizing the context that co-occurs with objects when objects do not occupy large portions of the scenes. It is thus unclear how humans achieve rapid scene categorization. To address this issue, we developed a hierarchical probabilistic model for rapid object categorization in natural scenes. In this model, a natural object category is represented by a coarse hierarchical probability distribution (PD), which includes PDs of object geometry and spatial configuration of object parts. Object parts are encoded by PDs of a set of natural object structures, each of which is a concatenation of local object features. Rapid categorization is performed as statistical inference. Since the model uses a very small number (~100) of structures for even complex object categories such as animals and cars, it requires little training and is robust in the presence of large variations within object categories and in their occurrences in natural scenes. Remarkably, we found that the model categorized animals in natural scenes and cars in street scenes with a near human-level performance. We also found that the model located animals and cars in natural scenes, thus overcoming a flaw in many other models which is to categorize objects in natural context by categorizing contextual features. These results suggest that coarse PDs of object categories based on natural object structures and statistical operations on these PDs may underlie the human ability to rapidly categorize scenes.

Original languageEnglish (US)
Article numbere20002
JournalPLoS One
Volume6
Issue number5
DOIs
StatePublished - May 31 2011

Fingerprint

probabilistic models
Statistical Models
Animal Models
Space Simulation
Animals
Railroad cars
animal models
probability distribution
Probability distributions
Defects
Geometry
animals

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)

Cite this

A hierarchical probabilistic model for rapid object categorization in natural scenes. / He, Xiaofu; Yang, Zhiyong; Tsien, Joseph Zhuo.

In: PLoS One, Vol. 6, No. 5, e20002, 31.05.2011.

Research output: Contribution to journalArticle

@article{9ec70d80dff94cb2a5255dd02b249a6c,
title = "A hierarchical probabilistic model for rapid object categorization in natural scenes",
abstract = "Humans can categorize objects in complex natural scenes within 100-150 ms. This amazing ability of rapid categorization has motivated many computational models. Most of these models require extensive training to obtain a decision boundary in a very high dimensional (e.g., ~6,000 in a leading model) feature space and often categorize objects in natural scenes by categorizing the context that co-occurs with objects when objects do not occupy large portions of the scenes. It is thus unclear how humans achieve rapid scene categorization. To address this issue, we developed a hierarchical probabilistic model for rapid object categorization in natural scenes. In this model, a natural object category is represented by a coarse hierarchical probability distribution (PD), which includes PDs of object geometry and spatial configuration of object parts. Object parts are encoded by PDs of a set of natural object structures, each of which is a concatenation of local object features. Rapid categorization is performed as statistical inference. Since the model uses a very small number (~100) of structures for even complex object categories such as animals and cars, it requires little training and is robust in the presence of large variations within object categories and in their occurrences in natural scenes. Remarkably, we found that the model categorized animals in natural scenes and cars in street scenes with a near human-level performance. We also found that the model located animals and cars in natural scenes, thus overcoming a flaw in many other models which is to categorize objects in natural context by categorizing contextual features. These results suggest that coarse PDs of object categories based on natural object structures and statistical operations on these PDs may underlie the human ability to rapidly categorize scenes.",
author = "Xiaofu He and Zhiyong Yang and Tsien, {Joseph Zhuo}",
year = "2011",
month = "5",
day = "31",
doi = "10.1371/journal.pone.0020002",
language = "English (US)",
volume = "6",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "5",

}

TY - JOUR

T1 - A hierarchical probabilistic model for rapid object categorization in natural scenes

AU - He, Xiaofu

AU - Yang, Zhiyong

AU - Tsien, Joseph Zhuo

PY - 2011/5/31

Y1 - 2011/5/31

N2 - Humans can categorize objects in complex natural scenes within 100-150 ms. This amazing ability of rapid categorization has motivated many computational models. Most of these models require extensive training to obtain a decision boundary in a very high dimensional (e.g., ~6,000 in a leading model) feature space and often categorize objects in natural scenes by categorizing the context that co-occurs with objects when objects do not occupy large portions of the scenes. It is thus unclear how humans achieve rapid scene categorization. To address this issue, we developed a hierarchical probabilistic model for rapid object categorization in natural scenes. In this model, a natural object category is represented by a coarse hierarchical probability distribution (PD), which includes PDs of object geometry and spatial configuration of object parts. Object parts are encoded by PDs of a set of natural object structures, each of which is a concatenation of local object features. Rapid categorization is performed as statistical inference. Since the model uses a very small number (~100) of structures for even complex object categories such as animals and cars, it requires little training and is robust in the presence of large variations within object categories and in their occurrences in natural scenes. Remarkably, we found that the model categorized animals in natural scenes and cars in street scenes with a near human-level performance. We also found that the model located animals and cars in natural scenes, thus overcoming a flaw in many other models which is to categorize objects in natural context by categorizing contextual features. These results suggest that coarse PDs of object categories based on natural object structures and statistical operations on these PDs may underlie the human ability to rapidly categorize scenes.

AB - Humans can categorize objects in complex natural scenes within 100-150 ms. This amazing ability of rapid categorization has motivated many computational models. Most of these models require extensive training to obtain a decision boundary in a very high dimensional (e.g., ~6,000 in a leading model) feature space and often categorize objects in natural scenes by categorizing the context that co-occurs with objects when objects do not occupy large portions of the scenes. It is thus unclear how humans achieve rapid scene categorization. To address this issue, we developed a hierarchical probabilistic model for rapid object categorization in natural scenes. In this model, a natural object category is represented by a coarse hierarchical probability distribution (PD), which includes PDs of object geometry and spatial configuration of object parts. Object parts are encoded by PDs of a set of natural object structures, each of which is a concatenation of local object features. Rapid categorization is performed as statistical inference. Since the model uses a very small number (~100) of structures for even complex object categories such as animals and cars, it requires little training and is robust in the presence of large variations within object categories and in their occurrences in natural scenes. Remarkably, we found that the model categorized animals in natural scenes and cars in street scenes with a near human-level performance. We also found that the model located animals and cars in natural scenes, thus overcoming a flaw in many other models which is to categorize objects in natural context by categorizing contextual features. These results suggest that coarse PDs of object categories based on natural object structures and statistical operations on these PDs may underlie the human ability to rapidly categorize scenes.

UR - http://www.scopus.com/inward/record.url?scp=79957576835&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79957576835&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0020002

DO - 10.1371/journal.pone.0020002

M3 - Article

VL - 6

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 5

M1 - e20002

ER -