Metric learning methods, for person re-identification, estimate a scaling distances in vector space that is optimized picking out observations of the same individual. This paper presents novel approach to pedestrian re-identification problem uses metric improve state-of-the-art performance on standard public datasets. Very high dimensional features are extracted from source color image. A first processing stage performs unsupervised PCA dimensionality reduction, constrained maintain...

Breast cancer is the most common cause of death for women worldwide. Thus, ability artificial intelligence systems to detect possible breast very important. In this paper, an ensemble classification mechanism proposed based on a majority voting mechanism. First, performance different state-of-the-art machine learning algorithms were evaluated Wisconsin Cancer Dataset (WBCD). The three best classifiers then selected their F3 score. score used emphasize importance false negatives (recall) in...

Human activity recognition has attracted the attention of researchers around world. This is an interesting problem that can be addressed in different ways. Many approaches have been presented during last years. These applications present solutions to recognize kinds activities such as if person walking, running, jumping, jogging, or falling, among others. Amongst all these activities, fall detection special importance because it a common dangerous event for people ages with more negative...

The understanding of crowd behaviour in semi-confined spaces is an important part the design new pedestrian facilities, for major layout modifications to existing areas and daily management sites-subject traffic. Conventional manual measurement techniques are not suitable comprehensive data collection patterns site occupation movement. Real-time monitoring tedious tiring, but safety-critical. This paper presents some image processing being developed at King's College London which, using...

An automatic monitoring system is proposed in this paper for detecting overcrowding conditions the platforms of underground train services. Whenever detected, will notify station operators to take appropriate actions prevent accidents, such as people falling off or being pushed onto tracks. The designed use existing closed circuit television (CCTV) cameras acquiring images platforms. In order focus on passengers platform, background subtraction and update techniques are used. addition, due...

This paper describes a body of multicamera human action video data with manually annotated silhouette that has been generated for the purpose evaluating silhouette-based recognition methods. It provides realistic challenge to both segmentation and communities can act as benchmark objectively compare proposed algorithms. The public multi-camera, multi-action dataset is an improvement over existing datasets (e.g. PETS, CAVIAR, soccerdataset) have not developed specifically complements other...

This paper presents a system for vehicle detection, tracking and classification from roadside CCTV. The counts vehicles separates them into four categories: car, van, bus motorcycle (including bicycles). A new background Gaussian Mixture Model (GMM) shadow removal method have been used to deal with sudden illumination changes camera vibration. Kalman filter tracks enable by majority voting over several consecutive frames, level set has refine the foreground blob. Extensive experiments real...

Object detection is a critical task that becomes difficult when dealing with onboard using aerial images and computer vision technique. The main challenges are small target sizes, low resolution, occlusion, attitude, scale variations, which affect the performance of many object detectors. accuracy efficiency inference always trade-offs. We modified architecture CenterNet used different CNN-based backbones ResNet18, ResNet34, ResNet50, ResNet101, ResNet152, Res2Net50, Res2Net101, DLA-34,...

The goal of this work is to assess the efficacy texture measures for estimating levels crowd densities in images. This estimation crucial problem monitoring and control. assessment carried out on a set nearly 300 real images captured from Liverpool Street Train Station, London, UK, using extracted through following four different methods: gray level dependence matrices, straight line segments, Fourier analysis, fractal dimensions. estimations are given terms classification input five classes...

The estimation of the number people in an area under surveillance is very important for problem crowd monitoring. When reaches occupation level greater than projected one, people's safety can be danger. This paper describes a new technique density based on Minkowski fractal dimension. dimension has been widely used to characterize data texture large physical and biological sciences. results our experiments show that also levels congestion images crowds. proposed compared with statistical...

In this study, a new multi‐view human action recognition approach is proposed by exploiting low‐dimensional motion information of actions. Before feature extraction, pre‐processing steps are performed to remove noise from silhouettes, incurred due imperfect, but realistic segmentation. Two‐dimensional templates based on history image (MHI) computed for each view/action video. Histograms oriented gradients (HOGs) used as an efficient description the MHIs which classified using nearest...

Classroom communication involves teacher's behavior and student's responses. Extensive research has been done on the analysis of facial expressions, but impact instructor's expressions is yet an unexplored area research. Facial expression recognition potential to predict emotions in a classroom environment. Intelligent assessment instructor during lecture delivery not only might improve learning environment also could save time resources utilized manual strategies. To address issue...

Breast cancer is one of the leading causes death among women, more so than all other cancers. The accurate diagnosis breast very difficult due to complexity disease, changing treatment procedures and different patient population samples. Diagnostic techniques with better performance are important for personalized care reduce control recurrence cancer. main objective this research was select feature selection using correlation analysis variance input features before passing these significant...

Recently, the scientific community has placed great emphasis on recognition of human activity, especially in area health and care for elderly. There are already practical applications activity unusual conditions that use body sensors such as wrist-worn devices or neck pendants. These relatively simple may be prone to errors, might uncomfortable wear, forgotten not worn, unable detect more subtle incorrect postures. Therefore, other proposed methods based images videos carry out recognition,...

On-line surveillance to improve safety and security is a major requirement for the management of public transport networks other places. The task complex one involving people, procedures, technology. This work describes an architecture that takes into account distributed nature detection processes need allow different types devices actuators. was part European initiative on intelligent systems. Because dominant closed circuit television in surveillance, detail computer-vision module used...