Video topic modelling with behavioural segmentation

Research output: Contribution to conferencePaper

8 Citations (Scopus)
56 Downloads (Pure)

Abstract

Topic models such as Latent Dirichlet Allocation (LDA) are used extensively for modelling multi-object behaviour and anomaly detection in busy scenes. However, existing topic models suffer from the sensitivity problem, where they are unable to detect anomalies that are mixed in with large numbers of co-occurring normal behaviours. Also at issue is the localisation problem, where anomalies are detected but not localised within a given video clip. To address these two problems this paper proposes a novel region LDA model, which encodes the spatial awareness that is ignored by conventional topic models. Both scene decomposition and behavioural modelling are simultaneously performed. Consequentially, abnormality is detected per-region rather than for the entire scene, resolving both the sensitivity and localisation issues. Experiments conducted on busy real world scenes demonstrate the superiority of the proposed model.
Original languageEnglish
Pages53-58
Number of pages6
Publication statusPublished - 2010
EventACM international workshop on Multimodal pervasive video analysis - Firenze, Italy
Duration: 25 Oct 201029 Oct 2010
Conference number: 1

Workshop

WorkshopACM international workshop on Multimodal pervasive video analysis
CountryItaly
CityFirenze
Period25/10/1029/10/10

Fingerprint

Decomposition
Experiments

Cite this

Fincham Haines, T., & Xiang, T. (2010). Video topic modelling with behavioural segmentation. 53-58. Paper presented at ACM international workshop on Multimodal pervasive video analysis, Firenze, Italy.

Video topic modelling with behavioural segmentation. / Fincham Haines, Tom; Xiang, Tao.

2010. 53-58 Paper presented at ACM international workshop on Multimodal pervasive video analysis, Firenze, Italy.

Research output: Contribution to conferencePaper

Fincham Haines, T & Xiang, T 2010, 'Video topic modelling with behavioural segmentation' Paper presented at ACM international workshop on Multimodal pervasive video analysis, Firenze, Italy, 25/10/10 - 29/10/10, pp. 53-58.
Fincham Haines T, Xiang T. Video topic modelling with behavioural segmentation. 2010. Paper presented at ACM international workshop on Multimodal pervasive video analysis, Firenze, Italy.
Fincham Haines, Tom ; Xiang, Tao. / Video topic modelling with behavioural segmentation. Paper presented at ACM international workshop on Multimodal pervasive video analysis, Firenze, Italy.6 p.
@conference{c4dde24b4b914d6c891dc5411f42d32c,
title = "Video topic modelling with behavioural segmentation",
abstract = "Topic models such as Latent Dirichlet Allocation (LDA) are used extensively for modelling multi-object behaviour and anomaly detection in busy scenes. However, existing topic models suffer from the sensitivity problem, where they are unable to detect anomalies that are mixed in with large numbers of co-occurring normal behaviours. Also at issue is the localisation problem, where anomalies are detected but not localised within a given video clip. To address these two problems this paper proposes a novel region LDA model, which encodes the spatial awareness that is ignored by conventional topic models. Both scene decomposition and behavioural modelling are simultaneously performed. Consequentially, abnormality is detected per-region rather than for the entire scene, resolving both the sensitivity and localisation issues. Experiments conducted on busy real world scenes demonstrate the superiority of the proposed model.",
author = "{Fincham Haines}, Tom and Tao Xiang",
year = "2010",
language = "English",
pages = "53--58",
note = "ACM international workshop on Multimodal pervasive video analysis ; Conference date: 25-10-2010 Through 29-10-2010",

}

TY - CONF

T1 - Video topic modelling with behavioural segmentation

AU - Fincham Haines, Tom

AU - Xiang, Tao

PY - 2010

Y1 - 2010

N2 - Topic models such as Latent Dirichlet Allocation (LDA) are used extensively for modelling multi-object behaviour and anomaly detection in busy scenes. However, existing topic models suffer from the sensitivity problem, where they are unable to detect anomalies that are mixed in with large numbers of co-occurring normal behaviours. Also at issue is the localisation problem, where anomalies are detected but not localised within a given video clip. To address these two problems this paper proposes a novel region LDA model, which encodes the spatial awareness that is ignored by conventional topic models. Both scene decomposition and behavioural modelling are simultaneously performed. Consequentially, abnormality is detected per-region rather than for the entire scene, resolving both the sensitivity and localisation issues. Experiments conducted on busy real world scenes demonstrate the superiority of the proposed model.

AB - Topic models such as Latent Dirichlet Allocation (LDA) are used extensively for modelling multi-object behaviour and anomaly detection in busy scenes. However, existing topic models suffer from the sensitivity problem, where they are unable to detect anomalies that are mixed in with large numbers of co-occurring normal behaviours. Also at issue is the localisation problem, where anomalies are detected but not localised within a given video clip. To address these two problems this paper proposes a novel region LDA model, which encodes the spatial awareness that is ignored by conventional topic models. Both scene decomposition and behavioural modelling are simultaneously performed. Consequentially, abnormality is detected per-region rather than for the entire scene, resolving both the sensitivity and localisation issues. Experiments conducted on busy real world scenes demonstrate the superiority of the proposed model.

M3 - Paper

SP - 53

EP - 58

ER -