Multivariate Adaptive Gaussian Mixture for Scene Level Anomaly Modeling

Kumari, Pratibha; Saini, Mukesh

doi:10.1109/bigmm50055.2020.00018

Cited by 4 publications

(17 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table 4 shows the following attributes: if the video is recorded in continuation, i.e., it is untrimmed (column 2), the dataset is recorded from how many locations (column 3), where is the data recorded (column 4), the recording location is indoor or outdoor (column 5), what is the density of people in the scene (column 6). Continuity (yes or no) and location (how many) helps to see the applicability of a dataset for the task at hand such as if the aim is to examine the context adaptation of anomaly detection framework, then the footage should be untrimmed as well as recorded at a single location [162]. The adaptive models need sufficient data to learn the context on their own and adapt over time; hence a dataset with too short clips from different spatial and temporal aspects is not useful for evaluation.…”

Section: Discussion On Video Datasetsmentioning

confidence: 99%

Multimedia Datasets for Anomaly Detection: A Review

Kumari¹,

Bedi²,

Saini³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Audio or video anomaly datasets play a crucial role in automated surveillance. The range of applications expand from outlier object/ situation detection to the detection of lifethreatening events. This research area has been active for more than 1.5 decades, and consequently, more and more datasets dedicated to anomalous actions and object detection have been created. Making use of these public anomaly datasets enable researchers to compare various anomaly detection frameworks with the same input data. This survey aims to address the lack of a comprehensive comparison and analysis of public datasets for audio, video, and audio-visual based anomaly detection. It also assists in selecting the best dataset for bench-marking frameworks. Additionally, we discuss gaps in the existing dataset and future direction insights toward developing multimodal anomaly detection datasets.

show abstract

Section: Discussion On Video Datasetsmentioning

confidence: 99%

Multimedia Datasets for Anomaly Detection: A Review

Kumari¹,

Bedi²,

Saini³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…With the above-mentioned frameworks, the samples which were regarded abnormal in training data will always remain abnormal. Thus, with one-time training, the model's performance may degrade when applied to data having concept drift [10].…”

Section: Introductionmentioning

confidence: 99%

“…To account for such drifts, the model needs to be updated online, also referred to as adaptive learning [10], [11]. Adaptive learning techniques facilitate lifelong dynamic updation of the model at the arrival of any data sample.…”

Section: Introductionmentioning

confidence: 99%

“…The development of adaptive anomaly detection frameworks has received less attention. There have been a few attempts to use the Adaptive Gaussian Mixture Model (AGMM) in an online manner for anomaly detection [10], [12], [13] or foreground detection [14], [15]. The parameters of the modes in the mixture continuously adjust themselves in order to accommodate varying data distributions.…”

Section: Introductionmentioning

confidence: 99%

“…The 'Learning' module learns about the distribution of conceptually distinct events occurring in the target scenario. For learning the scene dynamics, we use an AGMM-based framework [10]. Each time a sample (normal or abnormal) is received, the mixture parameters (mean, standard deviation, and weight) are updated to reflect changes in the environment.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

An Adaptive Framework for Anomaly Detection in Time-Series Audio-Visual Data

Kumari

Saini

2022

IEEE Access

Self Cite

View full text Add to dashboard Cite

Anomaly detection is an integral part of a number of surveillance applications. However, most of the existing anomaly detection models are statically trained on pre-recorded data from a single source, thus making multiple assumptions about the surrounding environment. As a result, their usefulness is limited to controlled scenarios. In this paper, we fuse information from live streams of audio and video data to detect anomalies in the captured environment. We train a deep learning-based teacher-student network using video, image, and audio information. The pre-trained visual network in the teacher model distills its information to the image and audio networks in the student model. Features from image and audio networks are combined and compressed using principal component analysis. Thus, the teacher-student network produces an image-audio-based light-weight joint representation of the data. The data dynamics are learned in a multivariate Adaptive Gaussian Mixture Model. Empirical results from two audio-visual datasets demonstrate the effectiveness of joint representation over single modalities in the adaptive anomaly detection framework. The proposed framework outperforms the state-of-the-art methods by an average of 15.00 % and 14.52 % for dataset 1 and dataset 2, respectively.INDEX TERMS Adaptive learning, concept drift, long term surveillance, multimodal anomaly detection, unsupervised model.

show abstract