scholarly journals Vision Transformer-Based Tailing Detection in Videos

2021 ◽  
Vol 11 (24) ◽  
pp. 11591
Author(s):  
Jaewoo Lee ◽  
Sungjun Lee ◽  
Wonki Cho ◽  
Zahid Ali Siddiqui ◽  
Unsang Park
Keyword(s):  
Time Series ◽  
Anomaly Detection ◽  
Spatial Information ◽  
Detection Problem ◽  
Walking Pattern ◽  
Series Of Experiments ◽  
Time Series Images ◽  
Small Dataset ◽  
Better Than ◽  
Do So

Tailing is defined as an event where a suspicious person follows someone closely. We define the problem of tailing detection from videos as an anomaly detection problem, where the goal is to find abnormalities in the walking pattern of the pedestrians (victim and follower). We, therefore, propose a modified Time-Series Vision Transformer (TSViT), a method for anomaly detection in video, specifically for tailing detection with a small dataset. We introduce an effective way to train TSViT with a small dataset by regularizing the prediction model. To do so, we first encode the spatial information of the pedestrians into 2D patterns and then pass them as tokens to the TSViT. Through a series of experiments, we show that the tailing detection on a small dataset using TSViT outperforms popular CNN-based architectures, as the CNN architectures tend to overfit with a small dataset of time-series images. We also show that when using time-series images, the performance of CNN-based architecture gradually drops, as the network depth is increased, to increase its capacity. On the other hand, a decreasing number of heads in Vision Transformer architecture shows good performance on time-series images, and the performance is further increased as the input resolution of the images is increased. Experimental results demonstrate that the TSViT performs better than the handcrafted rule-based method and CNN-based method for tailing detection. TSViT can be used in many applications for video anomaly detection, even with a small dataset.

scholarly journals INTRUSION DETECTION FOR DISCRETE SEQUENCES

2013 ◽  
pp. 43-45
Author(s):  
MRS. M. VIJAYALAKSHMI ◽  
MR. K . JANARDHAN
Keyword(s):  
Time Series ◽  
Anomaly Detection ◽  
Problem Formulation ◽  
Detection Problem ◽  
Basic Technique ◽  
Global Understanding ◽  
Discrete Sequences ◽  
Problem Formulations ◽  
Online Anomaly Detection ◽  
Disconnected Domains

Global understanding of the sequence anomaly detection problem and how techniques proposed for different domains relate to each other. Our specific contributions are as follows: We identify three distinct formulations of the anomaly detection problem, and review techniques from many disparate and disconnected domains that address each of these formulations. Within each problem formulation, we group techniques into categories based on the nature of the underlying algorithm. For each category, we provide a basic anomaly detection technique, and show how the existing techniques are variants of the basic technique. This approach shows how different techniques within a category are related or different from each other. Our categorization reveals new variants and combinations that have not been investigated before for anomaly detection. We also provide a discussion of relative strengths and weaknesses of different techniques. We show how techniques developed for one problem formulation can be adapted to solve a different formulation; thereby providing several novel adaptations to solve the different problem formulations. We highlight the applicability of the techniques that handle discrete sequences to other related areas such as online anomaly detection and time series anomaly detection.

scholarly journals Data Anomaly Detection of Bridge Structures Using Convolutional Neural Network Based on Structural Vibration Signals

Symmetry ◽  
2021 ◽  
Vol 13 (7) ◽  
pp. 1186
Author(s):  
Yixiao Zhang ◽  
Ying Lei
Keyword(s):  
Neural Network ◽  
Time Series ◽  
Anomaly Detection ◽  
Convolutional Neural Network ◽  
Classification Problem ◽  
Structural Vibration ◽  
Time Series Classification ◽  
Detection Problem ◽  
Vibration Signals ◽  
Data Expansion

Structural monitoring provides valuable information on the state of structural health, which is helpful for structural damage detection and structural state assessment. However, when the sensors are exposed to harsh environmental conditions, various anomalies caused by sensor failure or damage lead to abnormalities of the monitoring data. It is inefficient to remove abnormal data by manual elimination because of the massive number of data obtained by monitoring systems. In this paper, a data anomaly detection method based on structural vibration signals and a convolutional neural network (CNN) is proposed, which can automatically identify and eliminate abnormal data. First, the anomaly detection problem is modeled as a time series classification problem. Data preprocessing and data augmentation, including data expansion and down-sampling to construct new samples, are employed to process the original time series. For a small number of samples in the data set, randomly increase outliers, symmetrical flipping, and noise addition methods are used for data expansion, and samples with the same label are added without increasing the original samples. The down-sampling method of symmetrically extracting the maximum value and the minimum value at the same time can effectively reduce the dimensionality of the input sample, while retaining the characteristics of the data to the greatest extent. Using hyperparameter tuning of the classification weights, CNN is more effective in dealing with unbalanced training sets. Finally, the effectiveness of the proposed method is proved by the anomaly detection of acceleration data on a long-span bridge. For the anomaly detection problem modeled as a time series classification problem, the proposed method can effectively identify various abnormal patterns.

Phonological and Spatial Processing Abilities in Language- and Reading-Impaired Children

1988 ◽  
Vol 53 (3) ◽  
pp. 316-327 ◽  
Author(s):  
Alan G. Kamhi ◽  
Hugh W. Catts ◽  
Daria Mauer ◽  
Kenn Apel ◽  
Betholyn F. Gentry
Keyword(s):  
Phonological Processing ◽  
Spatial Information ◽  
Normal Children ◽  
Language Impaired ◽  
Word Repetition ◽  
Paper Folding ◽  
Syllable Segmentation ◽  
Processing Abilities ◽  
Impaired Children ◽  
Better Than

In the present study, we further examined (see Kamhi & Catts, 1986) the phonological processing abilities of language-impaired (LI) and reading-impaired (RI) children. We also evaluated these children's ability to process spatial information. Subjects were 10 LI, 10 RI, and 10 normal children between the ages of 6:8 and 8:10 years. Each subject was administered eight tasks: four word repetition tasks (monosyllabic, monosyllabic presented in noise, three-item, and multisyllabic), rapid naming, syllable segmentation, paper folding, and form completion. The normal children performed significantly better than both the LI and RI children on all but two tasks: syllable segmentation and repeating words presented in noise. The LI and RI children performed comparably on every task with the exception of the multisyllabic word repetition task. These findings were consistent with those from our previous study (Kamhi & Catts, 1986). The similarities and differences between LI and RI children are discussed.

An Anomaly Detection Method with Exemplar Subsequence for Time Series Data

2016 ◽  
Vol 136 (3) ◽  
pp. 363-372
Author(s):  
Takaaki Nakamura ◽  
Makoto Imamura ◽  
Masashi Tatedoko ◽  
Norio Hirai
Keyword(s):  
Time Series ◽  
Anomaly Detection ◽  
Time Series Data ◽  
Detection Method ◽  
Series Data

Research of the activity and viability of spermatozoa at different concentrations and proportions of diluents after incubation

2020 ◽  
pp. 88-95
Author(s):  
Svitlana Lobchenko ◽  
Tetiana Husar ◽  
Viktor Lobchenko
Keyword(s):  
Plasma Concentration ◽  
Incubation Time ◽  
Coefficient Of Variation ◽  
Incubation Medium ◽  
Wide Range ◽  
Series Of Experiments ◽  
The Subject ◽  
Boar Spermatozoa ◽  
Native Plasma ◽  
Better Than

The results of studies of the viability of spermatozoa with different incubation time at different concentrations and using different diluents are highlighted in the article. (Un) concentrated spermatozoa were diluented: 1) with their native plasma; 2) medium 199; 3) a mixture of equal volumes of plasma and medium 199. The experiment was designed to generate experimental samples with spermatozoa concentrations prepared according to the method, namely: 0.2; 0.1; 0.05; 0.025 billion / ml. The sperm was evaluated after 2, 4, 6 and 8 hours. The perspective of such a study is significant and makes it possible to research various aspects of the subject in a wide range. In this regard, a series of experiments were conducted in this area. The data obtained are statistically processed and allow us to highlight the results that relate to each stage of the study. In particular, in this article it was found out some regularities between the viability of sperm, the type of diluent and the rate of rarefaction, as evidenced by the data presented in the tables. As a result of sperm incubation, the viability of spermatozoa remains at least the highest trend when sperm are diluted to a concentration of 0.1 billion / ml, regardless of the type of diluent used. To maintain the viability of sperm using this concentration of medium 199 is not better than its native plasma, and its mixture with an equal volume of plasma through any length of time incubation of such sperm. Most often it is at this concentration of sperm that their viability is characterized by the lowest coefficient of variation, regardless of the type of diluent used, which may indicate the greatest stability of the result under these conditions. The viability of spermatozoa with a concentration of 0.1 billion / ml is statistically significantly reduced only after 6 or even 8 hours of incubation. If the sperm are incubated for only 2 hours, regardless of the type of diluent used, the sperm concentrations tested do not affect the viability of the sperm. Key words: boar, spermatozoa, sperm plasma, concentration, incubation, medium 199, activity, viability, rarefaction.

Aristotle on How Pleasure Perfects Activity (Nicomachean Ethics x.5 1175a29-b14): The Optimising-View

2020 ◽  
Vol 0 (0) ◽  
Author(s):  
David Machek
Keyword(s):  
Nicomachean Ethics ◽  
New Interpretation ◽  
Better Than ◽  
Do So

AbstractThis article offers a new interpretation of Aristotle’s ambiguous and much-discussed claim that pleasure perfects activity (NE x.4). This interpretation provides an alternative to the two main competing readings of this claim in the scholarship: the addition-view, which envisages the perfection conferred by pleasure as an extra perfection beyond the perfection of activity itself; and the identity-view, according to which pleasure just is the perfect activity itself. The proposed interpretation departs from both these views in rejecting their assumption that pleasure cannot perfect the activity itself, and argues that pleasure makes activity perfect by optimising the exercise of one’s capacities for that activity. Those who build or play music with pleasure do so better than those who do not delight in these activities. The basis of this interpretation is Aristotle’s little-read remarks from the following chapter, i. e. NE x.5, about how pleasure “increases” the activity.

Sign in / Sign up

Export Citation Format

Share Document