Semantic feature-based visual attention model for pedestrian detection

Objective
Pedestrian detection under video surveillance systems has always been a hot topic in computer vision research. These systems are widely used in train stations, airports, large commercial plazas, and other public places. However, pedestrian detection remains difficult because of complex backgrounds. Given its development in recent years, the visual attention mechanism has attracted increasing attention in object detection and tracking research, and previous studies have achieved substantial progress and breakthroughs. We propose a novel pedestrian detection method based on the semantic features under the visual attention mechanism.
Method
The proposed semantic feature-based visual attention model is a spatial-temporal model that consists of two parts: the static visual attention model and the motion visual attention model. The static visual attention model in the spatial domain is constructed by combining bottom-up with top-down attention guidance. Based on the characteristics of pedestrians, the bottom-up visual attention model of Itti is improved by intensifying the orientation vectors of elementary visual features to make the visual saliency map suitable for pedestrian detection. In terms of pedestrian attributes, skin color is selected as a semantic feature for pedestrian detection. The regional and Gaussian models are adopted to construct the skin color model. Skin feature-based visual attention guidance is then proposed to complete the top-down process. The bottom-up and top-down visual attentions are linearly combined using the proper weights obtained from experiments to construct the static visual attention model in the spatial domain. The spatial-temporal visual attention model is then constructed via the motion features in the temporal domain. Based on the static visual attention model in the spatial domain, the frame difference method is combined with optical flowing to detect motion vectors. Filtering is applied to process the field of motion vectors. The saliency of motion vectors can be evaluated via motion entropy to make the selected motion feature more suitable for the spatial-temporal visual attention model.
Result
Standard datasets and practical videos are selected for the experiments. The experiments are performed on a MATLAB R2012a platform. The experimental results show that our spatial-temporal visual attention model demonstrates favorable robustness under various scenes, including indoor train station surveillance videos and outdoor scenes with swaying leaves. Our proposed model outperforms the visual attention model of Itti, the graph-based visual saliency model, the phase spectrum of quaternion Fourier transform model, and the motion channel model of Liu in terms of pedestrian detection. The proposed model achieves a 93% accuracy rate on the test video.
Conclusion
This paper proposes a novel pedestrian method based on the visual attention mechanism. A spatial-temporal visual attention model that uses low-level and semantic features is proposed to calculate the saliency map. Based on this model, the pedestrian targets can be detected through focus of attention shifts. The experimental results verify the effectiveness of the proposed attention model for detecting pedestrians.

A framework for experimental determination of localised vertical pedestrian forces on full-scale structures using wireless attitude and heading reference systems

A major weakness among loading models for pedestrians walking on flexible structures proposed in recent years is the various uncorroborated assumptions made in their development. This applies to spatio-temporal characteristics of pedestrian loading and the nature of multi-object interactions. To alleviate this problem, a framework for the determination of localised pedestrian forces on full-scale structures is presented using a wireless attitude and heading reference systems (AHRS). An AHRS comprises a triad of tri-axial accelerometers, gyroscopes and magnetometers managed by a dedicated data processing unit, allowing motion in three-dimensional space to be reconstructed. A pedestrian loading model based on a single point inertial measurement from an AHRS is derived and shown to perform well against benchmark data collected on an instrumented treadmill. Unlike other models, the current model does not take any predefined form nor does it require any extrapolations as to the timing and amplitude of pedestrian loading. In order to assess correctly the influence of the moving pedestrian on behaviour of a structure, an algorithm for tracking the point of application of pedestrian force is developed based on data from a single AHRS attached to a foot. A set of controlled walking tests with a single pedestrian is conducted on a real footbridge for validation purposes. A remarkably good match between the measured and simulated bridge response is found, indeed confirming applicability of the proposed framework.

Abnormal Pedestrian Trajectory analysis based on arbitrary-length clustering

This papers examines the use of trajectory distance measures and clustering techniques to define normal
and abnormal trajectories in the context of pedestrian tracking in public spaces. In order to detect abnormal
trajectories, what is meant by a normal trajectory in a given scene is firstly defined. Then every trajectory
that deviates from this normality is classified as abnormal. By combining Dynamic Time Warping and a
modified K-Means algorithms for arbitrary-length data series, we have developed an algorithm for trajectory
clustering and abnormality detection. The final system performs with an overall accuracy of 83% and 75%
when tested in two different standard datasets.

Monitoring and model updating of an FRP pedestrian truss bridge

This paper presents the results of a real bridge field experiment, carried out on a fiber reinforced polymer (FRP) pedestrian truss bridge of which nodes are reinforced with stainless steel plates. The aim of this paper is to identify the dynamic parameters of this bridge by using both conventional techniques and a model updating algorithm. In the field experiment, the bridge was instrumented with accelerometers at a number of locations on the bridge deck, recording both vertical and transverse vibrations. It was excited via jump tests at particular locations along its span and the resulting acceleration signals are used to identify dynamic parameters, such as the bridge mode shape, natural frequency and damping constant. Pedestrianinduced vibrations are also measured and utilized to identify dynamic parameters of the bridge. For a complete analysis of the bridge, a numerical model of the FRP bridge is created whose properties are calibrated utilizing a model updating algorithm. Comparable frequencies and mode shapes to those from the experiment were obtained by the FE models considering the reinforcement by increasing elastic modulus at every node of the bridge by stainless steel plate. Moreover, considering boundary conditions at both ends as fixed in the model resulted in modal properties comparable/similar to those from the experiment. This study also demonstrated that the effect of reinforcement and boundary conditions must be properly considered in an FE model to analyze real behavior of the FRP bridge.

Element partitioning and potential mobility within surface dusts on buildings in a polluted urban environment, Budapest.

A voluminous literature exists on the analysis of water-soluble ions extracted from gypsum crusts and patinas formed on building surfaces. However, less data is available on the intermediate dust layer and the important role its complex matrix and constituents play in crust/patina formation. To address this issue, surface dust samples were collected from two buildings in the city of Budapest. Substrate properties, different pollution levels and environmental variations were considered by collecting samples from a city centre granite building exposed to intense traffic conditions and from an oolitic limestone church situated in a pedestrian area outside and high above the main pollution zone. Selective extraction examines both water-soluble ions (Ca2+, Mg2+, Na+, K+, Cl-, NO3- SO42-) and selected elements (Fe, Mn, Zn, Cu, Cr, Pb, Ni) from the water-soluble, exchangeable/carbonate, amorphous Mn, amorphous Fe/Mn, crystalline Fe/Mn, organic and residual phases, their mobility and potential to catalyse heterogeneous surface reactions. Salt weathering processes are highlighted by high concentrations of water-soluble Ca2+, Na+, Cl- and SO42-- at both sites. Manganese, Zn and Cu and to a lesser extent Pb and Ni, are very mobile in the city centre dust, where 30%, 54%, 38%, 11% and 11% of their totals are bound by the water-soluble phase, respectively. Church dust shows a sharp contrast for Mn, Zn, Cu and Pb with only 3%, 1%, 12% and 3% of their totals being bound by the water-soluble phase respectively. This may be due to (a) different environmental conditions at the church e.g. lower humidity (b) continuous replenishment of salts under intensive city centre traffic conditions (c) enrichment in oxidisable organic carbon by a factor of 4.5 and a tenfold increase in acidity in the city centre dust.

Hybrid image/ray-shooting UHF radio propagation predictor for populated indoor environments

A novel model for indoor wireless communication, based on a dual image and ray-shooting approach, is presented. The model, capable of improved site-specific indoor propagation prediction, considers multiple human bodies moving within the environment. In a modern office at 2.45GHz, the combined effect of pedestrian traffic and a moving receiver causes rapid temporal fading of up to 30dB.

The effect of construction pattern and unit interlock on the structural behaviour of block pavements

The maintenance or even replacement of cracked pavements requires considerable financial resources and puts a large burden on the budgets of local councils. In addition to these costs, local councils also face liability claims arising from uneven or cracked pedestrian pavements. These currently cost the Manchester City Council and Preston City Council around £6 million a year each. Design procedures are empirical. A better understanding of the interaction between paving blocks, bedding sand and subbase was necessary in order to determine the mode of failure of pavements under load. Increasing applied stress was found to mobilise ‘‘rotational interlock’’, providing increased pavement stiffness and thus increased load dissipation resulting in lower transmitted stress on the subgrade. The indications from the literature
review were that pavements are designed to fail by excessive deformation and that paving blocks remained uncracked at failure. This was confirmed with experimental data which was obtained from tests on segments of pavements that were laid/constructed in a purpose built test frame in the laboratory.

Architecture and Construction Management at Queens University Belfast:Regional Cultural Centre

The Regional Cultural Centre in Letterkenny is a new 2000sqm arts center containing theatre, galleries, workshops and ancillary offices. The site is set back from the street, on high ground with good views. The form and envelope of the building was derived from geometrically connecting the site with the town’s two other main public buildings, the Cathedral (1901) and new Civic Offices (2002, also designed by MacGabhann Architects). This geometrical connection or vectors informed the geometry and shape of the building. This urban matrix of geometrically connecting three corner stones of society, namely the ecclesiastical headquarters, the administrative head quarters and the art centre helps to improve the town planning and urban design of the disparate and chaotic development that Letterkenny has become.
The large cantilever, which houses a 300sqm gallery, is aligned towards the Civic Offices, marks the entrance, and signifies a change of direction of the pedestrian route past the building, like a modern day obelisk.
The circulation routes and stairs internally provide views towards the civic offices and cathedral, thus reinforcing the connection between the three buildings and helps visitors make some sense of Letterkenny as an urban center. The main stairs and vertical circulation are contained behind the large glazed foyer, which is framed to be viewed externally like a proscenium stage, with visitors to the building passively acting their routes through the building.

Risk of fatal injury in older adult drivers, passengers, and pedestrians

Objectives To compare risk of fatal injury in elderly road users (drivers, passengers, pedestrians) with that of younger age groups and to assess the contribution of elderly road users to the number of reported fatalities in the population. Design Fatality age was categorized as 21 to 29, 30 to 39, 40 to 49, 50 to 59, 60 to 69, or 70 and older, and road user was categorized as driver, passenger, or pedestrian. Estimated number of trips made by each age group was used to adjust for exposure and to measure individual risk. Setting Fatalities recorded in Britain between 1989 and 2009. Participants Population-wide fatal injury counts in Britain. Measurements Age of fatally injured drivers, passengers, and pedestrians. Estimated number of trips made per year by drivers, passengers, and pedestrians. Results Risk of fatal injury, but not fatality numbers in the population, were higher for older adult (=70) drivers than for younger age groups. Risk of fatal injury was also high for older adult passengers and pedestrians, who represented the majority of older adult fatalities. Conclusion Previous emphasis on driver impairment in older age has unduly focussed attention on elderly drivers, who represent a minority of all driver fatalities. Older adults represent a much larger proportion of passenger and pedestrian fatalities. Additional policy schemes and initiatives should be targeted at safeguarding older adult passengers and making the road environment safer for elderly pedestrians. © 2012, Copyright the Authors Journal compilation © 2012, The American Geriatrics Society.

Accessibility and Health: Towards Walkability Tools for Planning Practice

There is increasing research interest in how we can most effectively intervene in the built environment to change behaviours such as physical activity and improve health. Much of this work has focussed around the concept of walkability and the identification of those attributes of our cities that encourage pedestrian activity, including density, connectivity and the aesthetic of the urban realm (Saelens et al 2003, Frank et al 2010). Much of the existing research has clarified the strength of the relationships between various environmental attributes and the differential impact on different demographic groups (e.g. Panter et al 2011). This has not yet been effectively translated into tools to help integrate the concepts of walkability into decision-making by statutory authorities that can help shape the spatial development and delivery of public services which can support more active lifestyles. A key reason for this has been that standard models for transport planning and accessibility are based on networks of road infrastructure, which provides a weak basis for modelling pedestrian accessibility (Chin et al 2008).

This paper reports the findings of Knowledge Exchange project funded by UK’s Economic and Social Research Council (ES/J010588/1) and partners including Belfast and Derry City Councils and Northern Ireland’s Public Health Agency, the Department of Regional Development and Belfast Healthy Cities, that has attempted to address this problem. This project has mapped city-wide footpath networks and used these to assist partner organisations in developing the evidence base for making decisions on public services based on health impacts and pedestrian access. The paper describes the tool developed, uses a number of examples to highlight its impact on areas of decision-making and evaluates the benefits of further integrating walkability into planning and development practice.

Small–scale fading characteristics of diversity combining schemes used for body–to–body communications within an urban environment at 2.45 GHz

In this paper, an analysis of spatial diversity and small-scale fading characteristics for body-to-bodycommunications is presented. The measurements were made at 2.45 GHz in an urban environment with uncontrolled pedestrian and vehicular traffic. The virtual array of four distributed receive antennas where situated on the centralchest, central waist, left waist and left wrist of the user’s body. Combining of the received signal measured at each ofthe antennas in the virtual array has shown that an average diversity gain of up to 11.8 dB can be achieved when usingfour distributed antennas and a maximal ratio combining scheme. To model the small-scale fading characteristics obtained at the output of the virtual combiners, we use diversity specific, theoretical probability density functions for multi-branch receivers operating in Nakagami-m fading channels. It is shown that these equations provide an excellent fit to the measured channel data.

Propagation modelling and measurements in a populated indoor environment at 5.2 GHz

Human occupants within indoor environments are not always stationary and their movement will lead to temporal channel variations that strongly affect the quality of indoor wireless communication systems. This paper describes a statistical channel characterization, based on experimental measurements, of human body effects on line-of-sight indoor narrowband propagation at 5.2 GHz. The analysis shows that, as the number of pedestrians within the measurement location increases, the Ricean K-factor that best fits the empirical data tends to decrease proportionally, ranging from K=7 with 1 pedestrian to K=0 with 4 pedestrians. Level crossing rate results were Rice distributed, while average fade duration results were significantly higher than theoretically computed Rice and Rayleigh, due to the fades caused by pedestrians. A novel CDF that accurately characterizes the 5.2 GHz channel in the considered indoor environment is proposed. For the first time, the received envelope CDF is explicitly described in terms of a quantitative measurement of pedestrian traffic within the indoor environment.

Connectivity and Physical Activity: Using Footpath Networks to Measure the Walkability of Built Environments

There is now a strong body of research that suggests that the form of the built environment can influence levels of physical activity, leading to an increasing interest in incorporating health objectives into spatial planning and regeneration policies and projects. There have been a number of strands to this research, one of which has sought to develop “objective” measurements of the built environment using Geographic Information Science (GIS) involving measures of connectivity and proximity to compare the relative “walkability” of different neighbourhoods. The development of the “walkability index” (e.g. Leslie et al 2007, Frank et al 2010) has become a popular indicator of spatial distribution of those features of the built environment that are considered to have the greatest positive influence on levels of physical activity. The success of this measure is built on its ability to succinctly capture built environment correlates of physical activity using routinely available spatial data, which includes using road centre lines as a basis of a proxy for connectivity.

This paper discusses two key aspects of the walkability index. First, it follows the suggestion of Chin et al (2008) that the use of a footpath network (where available), rather than road centre lines, may be far more effective in evaluating walkability. This may be particularly important for assessing changes in walkability arising from pedestrian-focused infrastructure projects, such as greenways. Second, the paper explores the implication of this for how connectivity can be measured. The paper takes six different measures of connectivity and first analyses the relationships between them and then tests their correlation with actual levels of physical activity of local residents in Belfast, Northern Ireland. The analysis finds that the best measurements appear to be intersection density and metric reach and uses this finding to discuss the implications of this for developing tools that may better support decision-making in spatial planning.

Knowledge Exchange, Spatial Analysis & Healthy Urban Environments (KESUE): Project Report

The Knowledge Exchange, Spatial Analysis and Healthy Urban Environments (KESUE) project has extended work previously undertaken by a QUB team of inter-disciplinary researchers engaged with the Physical Activity in the Regeneration of Connswater (PARC) project (Tully et al, 2013). The PARC project focussed on parts of East Belfast to assess the health impact of the Connswater Community Greenway. The KESUE project has aimed to extend some of the tools used initially in East Belfast so that they have data coverage of all of Belfast and Derry-Londonderry. The purpose of this has been to enable the development of evidence and policy tools that link features of the built environment with physical activity in these two cities. The project has used this data to help shape policy decisions in areas such as physical activity, park management, public transport and planning.

Working with a range of local partners who part-funded the project (City Councils in Belfast and Derry-Londonderry, Public Health Agency, Belfast Healthy Cities and Department of Regional Development), this project has mapped all the footpaths in the two cities (covering 37% of the NI population) and employed this to develop evidence used in strategies related to healthy urban planning. Using Geographic Information Systems (GIS), the footpath network has been used as a basis for a wide range of policy-relevant analyses including pedestrian accessibility to public facilities, site options for new infrastructure and assessing how vulnerable groups can access services such as pharmacies. Key outputs have been Accessibility Atlases and maps showing how walkability of the built environment varies across the two cities.

In addition to generating this useful data, the project included intense engagement with potential users of the research, which has led to its continued uptake in a number of policies and strategies, creating a virtuous circle of research, implementation and feedback. The project has proved so valuable to Belfast City Council that they have now taken on one of the researchers to continue the work in-house.

Street Space—North Street, Belfast

The Belfast city center is fractured, divided by motorways, parking lots, empty buildings, and big box stores. Its 19th-century heyday put it on the international map of textile production, which transformed and enriched its built structure. This tight architectural fabric was slowly destroyed in the 1940s by the Blitz, in the 1970s by road plans and “the troubles” and in the 1990s by large retail buildings. Few pedestrian streets traverse Belfast, and among them, most are recently-developed conduits for the passage of shoppers from one chain store to the next.Within this seemingly bleak urban landscape, there remain a few areas that offer a richer, more architecturally and socially diverse, more memory-laden conception of public space. Current redevelopment plans, however, threaten the mere existence of these few remaining historic streets in Belfast.This reality inspired the current project of one of the Masters in Architecture design units at Queen’s University Belfast. Our team (led by urban designer Michael Corr and myself) has been exploring North Street, one of the main arteries in Belfast City Center. Although North Street has a reputation for being run-down, derelict, and in need of redevelopment, it is one of the few intact 19th-century streets left in the area, and as such is worthy of study as an example of public space that is not strictly synonymous with commercial space.