Object categorization using co-occurrence, location and appearance

Galleguillos, Carolina; Rabinovich, Andrew; Belongie, Serge

doi:10.1109/cvpr.2008.4587799

Cited by 419 publications

(301 citation statements)

References 25 publications

(42 reference statements)

Supporting

Mentioning

297

Contrasting

Unclassified

Order By: Relevance

“…Our best method (merging multiple segmentation with mean operator) ranks second with for global pixel-wise accuracy with 76.1 behind Gould (76.5). The third one is our method using multiple segmentation and the max operator (75.4), then comes Yang et al (75.1) and our method using edison segmentation (75.1).The main advantage of the method of Gould et al is the introduction of the relative location priors between the object classes (a full 3-D spatial relationships between objects is inferred), which clearly helps the recognition as was also highlighted by Galleguillos et al [12]. We model only the co-occurences between adjacent regions in the relaxation labelling.…”

Section: Impact Of the Relaxation Labellingmentioning

confidence: 93%

“…A mapping between the keywords and the visual blobs is performed using a method based on Expectation Maximization. The rest of the literature [15,26,32,12,2] noticeably differs from the original work by Duygulu et al in the sense that the models built try to exploit the maximum of information that can be extracted from the image: not only low level features (color, texture, etc. ), but also local contextual relationships between pixels or image segments, location and even global relevance estimates.…”

mentioning

confidence: 93%

“…They studied two spatial extensions of the aspect models, one based on a forest of minimal spanning trees and the other one on a regular 8-neighbor MRF. Galleguillos et al [12] have shown that introducing contextual information about the co-occurrences and the relative location of image regions with local appearance-based features improves the global labellin. Athanasiadis et al [2] define a framework for simultaneous image segmentation and object labellin operating at the semantic level.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Segmentation-based multi-class semantic object detection

Vieux

Benois-Pineau

Domenger

et al. 2010

Multimed Tools Appl

View full text Add to dashboard Cite

In this paper we study the problem of the detection of semantic objects from known categories in images. Unlike existing techniques which operate at the pixel or at a patch level for recognition, we propose to rely on the categorization of image segments. Recent work has highlighted that image segments provide a sound support for visual object class recognition. In this work, we use image segments as primitives to extract robust features and train detection models for a predefined set of categories. Several segmentation algorithms are benchmarked and their performances for segment recognition are compared. We then propose two methods for enhancing the segments classification, one based on the fusion of the classification results obtained with the different segmentations, the other one based on the optimization of the global labelling by correcting local ambiguities between neighbor segments. We use as a benchmark the Microsoft MSRC-21 image database and show that our method competes with the current state-of-the-art.

show abstract

Section: Impact Of the Relaxation Labellingmentioning

confidence: 93%

mentioning

confidence: 93%

mentioning

confidence: 99%

See 1 more Smart Citation

Segmentation-based multi-class semantic object detection

Vieux

Benois-Pineau

Domenger

et al. 2010

Multimed Tools Appl

View full text Add to dashboard Cite

show abstract

“…However, identifying the abnormality of a person hanging in the air becomes difficult because the "person" is still above the "road". Third, contextual models become more informative when the more context types, such as co-occurrence and relative position/scale among objects, are used [7]. Finally, the models should not restrict the interpretation of scenes to find abnormal object properly.…”

Section: Introductionmentioning

confidence: 99%

Abnormal Object Detection by Canonical Scene-Based Contextual Model

Park

Kim

Lee

2012

Computer Vision – ECCV 2012

View full text Add to dashboard Cite

Abstract. Contextual modeling is a critical issue in scene understanding. Object detection accuracy can be improved by exploiting tendencies that are common among object configurations. However, conventional contextual models only exploit the tendencies of normal objects; abnormal objects that do not follow the same tendencies are hard to detect through contextual model. This paper proposes a novel generative model that detects abnormal objects by meeting four proposed criteria of success. This model generates normal as well as abnormal objects, each following their respective tendencies. Moreover, this generation is controlled by a latent scene variable. All latent variables of the proposed model are predicted through optimization via population-based Markov Chain Monte Carlo, which has a relatively short convergence time. We present a new abnormal dataset classified into three categories to thoroughly measure the accuracy of the proposed model for each category; the results demonstrate the superiority of our proposed approach over existing methods.

show abstract

“…Computer vision approaches have demonstrated that the use of context improves recognition performance [4]- [8]. While the term context is frequently used in the literature as one of important keywords, it is difficult to give its clear definition.…”

Section: Introductionmentioning

confidence: 99%

Image Categorization Using Scene-Context Scale Based on Random Forests

Kang

Nagahashi

Sugimoto

2011

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYScene-context plays an important role in scene analysis and object recognition. Among various sources of scene-context, we focus on scene-context scale, which means the effective scale of local context to classify an image pixel in a scene. This paper presents random forests based image categorization using the scene-context scale. The proposed method uses random forests, which are ensembles of randomized decision trees. Since the random forests are extremely fast in both training and testing, it is possible to perform classification, clustering and regression in real time. We train multi-scale texton forests which efficiently provide both a hierarchical clustering into semantic textons and local classification in various scale levels. The scene-context scale can be estimated by the entropy of the leaf node in the multi-scale texton forests. For image categorization, we combine the classified category distributions in each scale and the estimated scene-context scale. We evaluate on the MSRC21 segmentation dataset and find that the use of the scene-context scale improves image categorization performance. Our results have outperformed the state-of-the-art in image categorization accuracy.

show abstract

Object categorization using co-occurrence, location and appearance

Cited by 419 publications

References 25 publications

Segmentation-based multi-class semantic object detection

Segmentation-based multi-class semantic object detection

Abnormal Object Detection by Canonical Scene-Based Contextual Model

Image Categorization Using Scene-Context Scale Based on Random Forests

Contact Info

Product

Resources

About