Histograms of Optical Flow Orientation and Magnitude and Entropy to Detect Anomalous Events in Videos

Colque, Rensso Mora; Caetano, Carlos; Andrade, Matheus Toledo Lustosa de; Schwartz, William Robson

doi:10.1109/tcsvt.2016.2637778

Cited by 157 publications

(61 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…As one of the earliest representations for irregularity detection, trajectory were used [6,19], such that an event not following a learned normal trajectory pattern is considerd as anomaly. Optical-flows [4,[20][21][22], social forces (SF) [23], gradient features [3,24], mixture of dynamic textures [2], and mixture of probabilistic PCAs (MPPCA) [25] are types of low-level motion representations used to model regular concepts. Deep learned features, using auto-encoders [26,27] pre-trained networks [9], or PCAnet [28,29] have recently shown great success for anomaly detection.…”

Section: Related Workmentioning

confidence: 99%

“…Several different methods are used in the literature for learning the normal concept in visual scenes. Low-level visual features such as histogram of oriented gradients (HOG) [3] and histogram of optical flow (HOF) [4,5] were the first feature subsets explored for representing regular scenes in videos. Besides, trajectory features [6] are also used for representing and modeling the videos, although they are not robust against problems like occlusion [3,7].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AVID: Adversarial Visual Irregularity Detection

Sabokrou

Pourreza²,

Fayyaz

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Real-time detection of irregularities in visual data is very invaluable and useful in many prospective applications including surveillance, patient monitoring systems, etc. With the surge of deep learning methods in the recent years, researchers have tried a wide spectrum of methods for different applications. However, for the case of irregularity or anomaly detection in videos, training an end-to-end model is still an open challenge, since often irregularity is not well-defined and there are not enough irregular samples to use during training. In this paper, inspired by the success of generative adversarial networks (GANs) for training deep models in unsupervised or self-supervised settings, we propose an end-to-end deep network for detection and fine localization of irregularities in videos (and images). Our proposed architecture is composed of two networks, which are trained in competing with each other while collaborating to find the irregularity. One network works as a pixel-level irregularity Inpainter, and the other works as a patch-level Detector.After an adversarial self-supervised training, in which I tries to fool D into accepting its inpainted output as regular (normal), the two networks collaborate to detect and fine-segment the irregularity in any given testing video. Our results on three different datasets show that our method can outperform the state-of-the-art and fine-segment the irregularity.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

AVID: Adversarial Visual Irregularity Detection

Sabokrou

Pourreza²,

Fayyaz

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Table 1 lists both the EER and DR values of our approaches and comparison approaches. Table 1 shows that our performances are better than the approaches of SF-MPPCA [41] by Bertini et al [15] and Adam et al [42] as well as HOFME [19], HOG3D [19], MBH [19], and HOOF [19]. Figure 6 shows a performance comparison between the HORG descriptor and the direct use of regional feature descriptors.…”

Section: Ucsd Dataset 411 Resultsmentioning

confidence: 88%

“…Figure 4 shows examples of detection results; it is seen that different types of abnormalities, such as bicycles, skaters, and vehicles, can be detected and localized more accurately. To carry out a quantitative evaluation, we compare the performances of the HORG descriptor with several state-of-the-art descriptors, such as MDT [20], MOHF [17], and HOMFE [19] among others. Figure 5 shows the ROC (receiver operating characteristic) curves of the detection results from our approach as well as from other comparison approaches.…”

Section: Ucsd Dataset 411 Resultsmentioning

confidence: 99%

“…Kaltsa et al [18] proposed a histogram of oriented swarms (HOS) descriptor based on swarm theory together with the HOG to capture both the motion and appearance in a volume. Colque et al [19] developed a 3D descriptor for AED called histogram of optical flow orientation and magnitude and entropy (HOFME), which can effectively capture motion (velocity and orientation), appearance, and entropy information. Li et al [20] modeled the pixels of a volume as a mixture dynamic textures (MDT) to jointly model both the motion and appearance within the volume.…”

Section: ) Pixel-wise Descriptormentioning

confidence: 99%

See 1 more Smart Citation

Abnormal event detection in crowded scenes using histogram of oriented contextual gradient descriptor

Duan

et al. 2018

EURASIP J. Adv. Signal Process.

View full text Add to dashboard Cite

Detecting abnormal events in crowded scenes is an important but challenging task in computer vision. Contextual information is useful for discovering salient events in scenes; however, it cannot be characterized well by commonly used pixel-based descriptors, such as the HOG descriptor. In this paper, we propose contextual gradients between two local regions and then construct a histogram of oriented contextual gradient (HOCG) descriptor for abnormal event detection based on the contextual gradients. The HOCG descriptor is a distribution of contextual gradients of sub-regions in different directions, which can effectively characterize the compositional context of events. We conduct extensive experiments on several public datasets and compare the experimental results using state-of-the-art approaches. Qualitative and quantitative analysis of experimental results demonstrate the effectiveness of the proposed HOCG descriptor.

show abstract

Vision-Based Human Activity Recognition

Mahmud¹,

Hasan

2021

Contactless Human Activity Analysis

View full text Add to dashboard Cite

Histograms of Optical Flow Orientation and Magnitude and Entropy to Detect Anomalous Events in Videos

Cited by 157 publications

References 26 publications

AVID: Adversarial Visual Irregularity Detection

AVID: Adversarial Visual Irregularity Detection

Abnormal event detection in crowded scenes using histogram of oriented contextual gradient descriptor

Vision-Based Human Activity Recognition

Contact Info

Product

Resources

About