Visual surveillance is of utmost importance for ensuring public safety and, detecting and preventing violent activities. The rapidly increasing number of surveillance cameras makes automated visual surveillance necessary, since monitoring a large of number of cameras by operators is not feasible, requiring a huge workload. In this paper, we propose a compact method for automated analysis of behaviours in crowds, specifically detecting the abnormal activities in crowd videos, which is one of the most critical applications of visual surveillance. The most intuitive way of abnormal activity detection is to consider common and typical activities in the scenes as normal and any unseen strange activities as anomalies that might be due to dangerous events. When tragic incidents such as accidents, disasters, shootings and violent behaviours happen, people tend to move in a very fast pace and in arbitrary directions. Thus, the proposed method consists of modelling the activities of crowds in the scenes during regular events, and analysing the spatial and temporal changes in their motion, which may be related to abnormal activities. For defining the crowd activities, first, crowd specific motion representations are computed. The computed representations utilize motion attributes such as speed, direction and acceleration of people in the crowds. Next, by employing these representations, typical activities in crowd videos, related to normal behaviours of people when no abnormal activities are present, are learned. Later, the distributions of motion representations are inspected; abrupt changes in the distributions of motion representations, occurring in several parts of the scenes, are labelled as anomalies. Experiments, conducted on a publicly available dataset, involving videos of crowds, reveal that the proposed method is effective in detecting abnormal activities. Additionally, quantitative performance of variants of the proposed method including the baseline approaches were measured for a comparison.
Current state-of-the-art technology for in-vitro diagnostics employ laboratory tests such as ELISA that consists of a multi-step test procedure and give results in analog format. Results of these tests are interpreted by the color change in a set of diluted samples in a multi-well plate. However, detection of the minute changes in the color poses challenges and can lead to false interpretations. Instead, a technique that allows individual counting of specific binding events would be useful to overcome such challenges. Digital imaging has been applied recently for diagnostics applications. SPR is one of the techniques allowing quantitative measurements. However, the limit of detection in this technique is on the order of nM. The current required detection limit, which is already achieved with the analog techniques, is around pM. Optical techniques that are simple to implement and can offer better sensitivities have great potential to be used in medical diagnostics. Interference Microscopy is one of the tools that have been investigated over years in optics field. More of the studies have been performed in confocal geometry and each individual nanoparticle was observed separately. Here, we achieve wide-field imaging of individual nanoparticles in a large field-of-view (~166 μm × 250 μm) on a micro-array based sensor chip in fraction of a second. We tested the sensitivity of our technique on dielectric nanoparticles because they exhibit optical properties similar to viruses and cells. We can detect non-resonant dielectric polystyrene nanoparticles of 100 nm. Moreover, we perform post-processing applications to further enhance visibility.
The need for capabilities of automated visual content analysis has substantially increased due to presence of large number of images captured by surveillance cameras. With a focus on development of practical methods for extracting effective visual data representations, deep neural network based representations have received great attention due to their success in visual categorization of generic images. For fine-grained image categorization, a closely related yet a more challenging research problem compared to generic image categorization due to high visual similarities within subgroups, diverse applications were developed such as classifying images of vehicles, birds, food and plants. Here, we propose the use of deep neural network based representations for categorizing and identifying marine vessels for defense and security applications. First, we gather a large number of marine vessel images via online sources grouping them into four coarse categories; naval, civil, commercial and service vessels. Next, we subgroup naval vessels into fine categories such as corvettes, frigates and submarines. For distinguishing images, we extract state-of-the-art deep visual representations and train support-vector-machines. Furthermore, we fine tune deep representations for marine vessel images. Experiments address two scenarios, classification and verification of naval marine vessels. Classification experiment aims coarse categorization, as well as learning models of fine categories. Verification experiment embroils identification of specific naval vessels by revealing if a pair of images belongs to identical marine vessels by the help of learnt deep representations. Obtaining promising performance, we believe these presented capabilities would be essential components of future coastal and on-board surveillance systems.
Fine-grained object recognition is a potential computer vision problem that has been recently addressed by utilizing deep Convolutional Neural Networks (CNNs). Nevertheless, the main disadvantage of classification methods relying on deep CNN models is the need for considerably large amount of data. In addition, there exists relatively less amount of annotated data for a real world application, such as the recognition of car models in a traffic surveillance system. To this end, we mainly concentrate on the classification of fine-grained car make and/or models for visual scenarios by the help of two different domains. First, a large-scale dataset including approximately 900K images is constructed from a website which includes fine-grained car models. According to their labels, a state-of-the-art CNN model is trained on the constructed dataset. The second domain that is dealt with is the set of images collected from a camera integrated to a traffic surveillance system. These images, which are over 260K, are gathered by a special license plate detection method on top of a motion detection algorithm. An appropriately selected size of the image is cropped from the region of interest provided by the detected license plate location. These sets of images and their provided labels for more than 30 classes are employed to fine-tune the CNN model which is already trained on the large scale dataset described above. To fine-tune the network, the last two fully-connected layers are randomly initialized and the remaining layers are fine-tuned in the second dataset. In this work, the transfer of a learned model on a large dataset to a smaller one has been successfully performed by utilizing both the limited annotated data of the traffic field and a large scale dataset with available annotations. Our experimental results both in the validation dataset and the real field show that the proposed methodology performs favorably against the training of the CNN model from scratch.
Attention Deficit Hyperactivity Disorder (ADHD) is receiving lots of attention nowadays mainly because it is
one of the common brain disorders among children and not much information is known about the cause of this
disorder. In this study, we propose to use a novel approach for automatic classification of ADHD conditioned
subjects and control subjects using functional Magnetic Resonance Imaging (fMRI) data of resting state brains.
For this purpose, we compute the correlation between every possible voxel pairs within a subject and over the
time frame of the experimental protocol. A network of voxels is constructed by representing a high correlation
value between any two voxels as an edge. A Bag-of-Words (BoW) approach is used to represent each subject
as a histogram of network features; such as the number of degrees per voxel. The classification is done using
a Support Vector Machine (SVM). We also investigate the use of raw intensity values in the time series for
each voxel. Here, every subject is represented as a combined histogram of network and raw intensity features.
Experimental results verified that the classification accuracy improves when the combined histogram is used.
We tested our approach on a highly challenging dataset released by NITRC for ADHD-200 competition
and obtained promising results. The dataset not only has a large size but also includes subjects from different
demography and edge groups. To the best of our knowledge, this is the first paper to propose BoW approach in
any functional brain disorder classification and we believe that this approach will be useful in analysis of many
brain related conditions.
Conference Committee Involvement (4)
Counterterrorism, Crime Fighting, Forensics, and Surveillance Technologies V
13 September 2021 | Madrid, Spain
Counterterrorism, Crime Fighting, Forensics, and Surveillance Technologies IV
22 September 2020 | Online Only, United Kingdom
Counterterrorism, Crime Fighting, Forensics, and Surveillance Technologies III
9 September 2019 | Strasbourg, France
Counterterrorism, Crime Fighting, Forensics, and Surveillance Technologies