Context-aware Features Research Articles

Learning Pixel-Level and Instance-Level Context-Aware Features for Pedestrian Detection in Crowds

Pedestrian detection in crowded scenes is an intractable problem in computer vision, in which occlusion often presents a great challenge. In this paper, we propose a novel context-aware feature learning method for detecting pedestrians in crowds, with the purpose of making better use of context information for dealing with occlusion. Unlike most current pedestrian detectors that only extract context information from a single and fixed region, a new pixel-level context embedding module is developed to integrate multi-cue context into a deep CNN feature hierarchy, which enables access to the context of various regions by multi-branch convolution layers with different receptive fields. In addition, to utilize the distinctive visual characteristics formed by pedestrians that appear in groups and occlude each other, we propose a novel instance-level context prediction module which is actually implemented by a two-person detector, to improve the one-person detection performance. Applying with these strategies, we achieve an efficient and lightweight detector that can be trained in an end-to-end fashion. We evaluate the proposed approach on two popular pedestrian detection datasets, i.e., Caltech and CityPersons. The extensive experimental results demonstrate the effectiveness of the proposed method, especially under heavy occlusion cases.

Learning to group and label fine-grained shape components

A majority of stock 3D models in modern shape repositories are assembled with many fine-grained components. The main cause of such data form is the component-wise modeling process widely practiced by human modelers. These modeling components thus inherently reflect some function-based shape decomposition the artist had in mind during modeling. On the other hand, modeling components represent an over-segmentation since a functional part is usually modeled as a multi-component assembly. Based on these observations, we advocate that labeled segmentation of stock 3D models should not overlook the modeling components and propose a learning solution to grouping and labeling of the fine-grained components. However, directly characterizing the shape of individual components for the purpose of labeling is unreliable, since they can be arbitrarily tiny and semantically meaningless. We propose to generate part hypotheses from the components based on a hierarchical grouping strategy, and perform labeling on those part groups instead of directly on the components. Part hypotheses are mid-level elements which are more probable to carry semantic information. A multi-scale 3D convolutional neural network is trained to extract context-aware features for the hypotheses. To accomplish a labeled segmentation of the whole shape, we formulate higher-order conditional random fields (CRFs) to infer an optimal label assignment for all components. Extensive experiments demonstrate that our method achieves significantly robust labeling results on raw 3D models from public shape repositories. Our work also contributes the first benchmark for component-wise labeling.

Personalized travel time estimation for urban road networks: A tensor-based context-aware approach

Urban travel time estimation is of significant importance at many levels of traffic operation and transportation management. This paper develops a tensor-based context-aware approach to dynamically provide personalized travel time estimation from a citywide perspective, using sparse and large-scale GPS trajectories. This novel model is comprised of four major components: map matching, travel time tensor construction, context-aware feature extraction, and travel time tensor factorization. First, GPS trajectories are map-matched onto the road network. Then, travel times of different drivers on different road segments in different time slots are modeled with a 3-order tensor. Following these, three categories of context features, i.e., historical, geographical and spatial-temporal features, are extracted to capture the contextual information of travel time and traffic condition in the road network. Finally, an objective function is devised to factorize travel time tensors with context features collaboratively. In addition, a gradient-based algorithm is developed to find an optimal solution for the context-aware estimation model. The novel model incorporates both the spatial correlation between different road segments and the deviation between different drivers, as well as the fine-grain temporal correlation between different time slots and the coarse-grain temporal correlation between recent and historical traffic conditions. The proposed model is applied in a real case on the urban road network of Beijing, China, based on the sparse and large-scale GPS trajectories collected from over 32,000 drivers in a period of 2 months. Empirical results on extensive experiments demonstrate that the proposed model provides an effective and robust approach for citywide personalized travel time estimation, and outperforms the competing methods.

A proposed scheme for sentiment analysis

PurposeThis paper aims to propose a statistical and context-aware feature reduction algorithm that improves sentiment classification accuracy. Classification of reviews with different granularities in two classes of reviews with negative and positive polarities is among the objectives of sentiment analysis. One of the major issues in sentiment analysis is feature engineering while it severely affects time complexity and accuracy of sentiment classification.Design/methodology/approachIn this paper, a feature reduction method is proposed that uses context-based knowledge as well as synset statistical knowledge. To do so, one-dimensional presentation proposed for SentiWordNet calculates statistical knowledge that involves polarity concentration and variation tendency for each synset. Feature reduction involves two phases. In the first phase, features that combine semantic and statistical similarity conditions are put in the same cluster. In the second phase, features are ranked and then the features which are given lower ranks are eliminated. The experiments are conducted by support vector machine (SVM), naive Bayes (NB), decision tree (DT) and k-nearest neighbors (KNN) algorithms to classify the vectors of the unigram and bigram features in two classes of positive or negative sentiments.FindingsThe results showed that the applied clustering algorithm reduces SentiWordNet synset to less than half which reduced the size of the feature vector by less than half. In addition, the accuracy of sentiment classification is improved by at least 1.5 per cent.Originality/valueThe presented feature reduction method is the first use of the synset clustering for feature reduction. In this paper features reduction algorithm, first aggregates the similar features into clusters then eliminates unsatisfactory cluster.

Towards Adaptive user Interfaces for Mobile-Phone in Smart World

All applications are developed for context adaptation and provide communication with users through their interfaces. These applications offer new opportunities for developers as well as users by collecting context data and adapting systems behavior accordingly. Particularly, in mobile devices, these mechanisms provide usability increment tremendously. Rigid and non-adaptive interface blocks the features of context awareness. In this paper, we study methods, technologies and criteria which have been proposed specifically for adaptive interfaces. Based on these guidelines, we elaborate the intelligence of adaptivity and usage of context according to user mental model. Further, we have proposed a model to develop user context ontology (UCO) and adaptive interface ontology (AIO) to optimize the use of adaptive mobile interfaces in the context of user preferences. These ontologies organize the perceptions and thoughts of user. The philosophy of User Centered Design (UCD) is proposed to analyze the usability and validity of mobile device interfaces according to user contexts.

Dynamically Adaptable Software Is All about Modeling Contextual Variability and Avoiding Failures

Contextual-variability modeling is integral to dealing with dynamically adaptable software. Several techniques support this task using dynamic-software-product-line engineering, which enables runtime management of software variants. In particular, context-aware feature modeling (CFM) has been considered suitable for dealing with contextual variability. However, it has limited expressiveness to specify real-world constraints related to context information. To tackle this limitation, this article proposes eCFM, an extended form of CFM. In addition, it presents an approach using eCFM and model checking to identify common design faults that could lead to runtime failures. A comparison of CFM and eCFM showed that eCFM was more expressive and easier to use.

Let’s FOCUS

With the increasingly frequent appearance of mobile phones in college classrooms, there have been growing concerns regarding their negative aspects including distractive off-task multitasking. In this work, we design and evaluate Let’s FOCUS, a software-based intervention service that assists college students in self-regulating their mobile phone use in classrooms. Our preliminary survey study (with 47 professors and 283 students) reveals that it is critical to encourage voluntary participation by framing intervention as a learning tool and to raise awareness regarding appropriate mobile phone usage by establishing social norms in colleges. Let’s FOCUS introduces a virtual limiting space for each class (or a virtual classroom) where the students can explicitly restrict their mobile phone use voluntarily. Furthermore, it promotes students’ willing participation by leveraging social facilitation and context-aware reminders associated with virtual classrooms. We conducted a campus-wide campaign for approximately six weeks to evaluate the feasibility of the proposed approach. The results confirm that 379 students used the app to limit 9,335 hours of mobile phone usage over 233 classrooms. Let’s FOCUS was used in diverse learning contexts and for different purposes and its social learning and context-awareness features significantly motivated prolonged participation. We present the design considerations of software-based intervention.

CAFÉ-Map: Context Aware Feature Mapping for mining high dimensional biomedical data

Feature selection and ranking is of great importance in the analysis of biomedical data. In addition to reducing the number of features used in classification or other machine learning tasks, it allows us to extract meaningful biological and medical information from a machine learning model. Most existing approaches in this domain do not directly model the fact that the relative importance of features can be different in different regions of the feature space. In this work, we present a context aware feature ranking algorithm called CAFÉ-Map. CAFÉ-Map is a locally linear feature ranking framework that allows recognition of important features in any given region of the feature space or for any individual example. This allows for simultaneous classification and feature ranking in an interpretable manner. We have benchmarked CAFÉ-Map on a number of toy and real world biomedical data sets. Our comparative study with a number of published methods shows that CAFÉ-Map achieves better accuracies on these data sets. The top ranking features obtained through CAFÉ-Map in a gene profiling study correlate very well with the importance of different genes reported in the literature. Furthermore, CAFÉ-Map provides a more in-depth analysis of feature ranking at the level of individual examples.Availability: CAFÉ-Map Python code is available at:http://faculty.pieas.edu.pk/fayyaz/software.html#cafemap .The CAFÉ-Map package supports parallelization and sparse data and provides example scripts for classification. This code can be used to reconstruct the results given in this paper.

Context-Based Energy Disaggregation in Smart Homes

In this paper, we address the problem of energy conservation and optimization in residential environments by providing users with useful information to solicit a change in consumption behavior. Taking care to highly limit the costs of installation and management, our work proposes a Non-Intrusive Load Monitoring (NILM) approach, which consists of disaggregating the whole-house power consumption into the individual portions associated to each device. State of the art NILM algorithms need monitoring data sampled at high frequency, thus requiring high costs for data collection and management. In this paper, we propose an NILM approach that relaxes the requirements on monitoring data since it uses total active power measurements gathered at low frequency (about 1 Hz). The proposed approach is based on the use of Factorial Hidden Markov Models (FHMM) in conjunction with context information related to the user presence in the house and the hourly utilization of appliances. Through a set of tests, we investigated how the use of these additional context-awareness features could improve disaggregation results with respect to the basic FHMM algorithm. The tests have been performed by using Tracebase, an open dataset made of data gathered from real home environments.

Adaptive multimodal web user interfaces for smart work environments

This paper presents a solution for supporting adaptive user interfaces in work environments that require operators to move about in dynamic contexts for manipulating various physical objects. The solution architecture is built upon the use of logical languages for interactive applications integrated with context aware and adaptive features. The proposed architecture is able to adapt to specific aspects of the context during run-time by communicating with a context server and applying the specified adaptation rules. In order to show the possibilities of the proposed solution, we report on its application in the development of an adaptive user interface for a warehouse picking system, and discuss the results of the associated tests.

A Vision based Fall Detection System for Elderly People

Among the elderly people, falls have become a common health problem. With a growing population of elderly people, health systems are needed to meet the necessities of elderly people. A Microsoft Kinect sensor monitors the usual activities of people and the acquired image frames are processed in Raspberry pi. The context aware feature extraction technique identify the shape of a person and a mean based classification distinguish the fall from usual activities, if it encounters the unusual activity then the alert is sent to the particular person's caregivers through SIM800 GSM modem. Using the low cost Kinect sensor with Raspberry pi, action dataset is built that consist of three types of actions such as sitting, standing and falling for three different persons. The proposed computer vision approach accomplished the static background model which is resistant to the variations in illumination and provides better results.

Coupled Collaborative Filtering for Context-aware Recommendation

Context-aware features have been widely recognized as important factors in recommender systems. However, as a major technique in recommender systems, traditional Collaborative Filtering (CF) does not provide a straight-forward way of integrating the context-aware information into personal recommendation. We propose a Coupled Collaborative Filtering (CCF) model to measure the contextual information and use it to improve recommendations. In the proposed approach, coupled similarity computation is designed to be calculated by interitem, intra-context and inter-context interactions among item, user and context-ware factors. Experiments based on different types of CF models demonstrate the effectiveness of our design.

Discriminative vessel segmentation in retinal images by fusing context-aware hybrid features

Vessel segmentation is an important problem in medical image analysis and is often challenging due to large variations in vessel appearance and profiles, as well as image noises. To address these challenges, we propose a solution by combining heterogeneous context-aware features with a discriminative learning framework. Our solution is characterized by three key ingredients: First, we design a hybrid feature pool containing recently invented descriptors including the stroke width transform (SWT) and Weber's local descriptors (WLD), as well as classical local features including intensity values, Gabor responses and vesselness measurements. Second, we encode context information by sampling the hybrid features from an orientation invariant local context. Third, we treat pixel-level vessel segmentation as a discriminative classification problem, and use a random forest to fuse the rich information encoded in the hybrid context-aware features. For evaluation, the proposed method is applied to retinal vessel segmentation using three publicly available benchmark datasets. On the DRIVE and STARE datasets, our approach achieves average classification accuracies of 0.9474 and 0.9633, respectively. On the high-resolution dataset HRFID, our approach achieves average classification accuracies of 0.9647, 0.9561 and 0.9634 on three different categories, respectively. Experiments are also conducted to validate the superiority of hybrid feature fusion over each individual component.

Improving the Effectiveness of Testing Pervasive Software via Context Diversity

Context-aware pervasive software is responsive to various contexts and their changes. A faulty implementation of the context-aware features may lead to unpredictable behavior with adverse effects. In software testing, one of the most important research issues is to determine the sufficiency of a test suite to verify the software under test. Existing adequacy criteria for testing traditional software, however, have not explored the dimension of serial test inputs and have not considered context changes when constructing test suites. In this article, we define the concept of context diversity to capture the extent of context changes in serial inputs and propose three strategies to study how context diversity may improve the effectiveness of the data-flow testing criteria. Our case study shows that the strategy that uses test cases with higher context diversity can significantly improve the effectiveness of existing data-flow testing criteria for context-aware pervasive software. In addition, test suites with higher context diversity are found to execute significantly longer paths, which may provide a clue that reveals why context diversity can contribute to the improvement of effectiveness of test suites.

Lesion segmentation from multimodal MRI using random forest following ischemic stroke

Understanding structure–function relationships in the brain after stroke is reliant not only on the accurate anatomical delineation of the focal ischemic lesion, but also on previous infarcts, remote changes and the presence of white matter hyperintensities. The robust definition of primary stroke boundaries and secondary brain lesions will have significant impact on investigation of brain–behavior relationships and lesion volume correlations with clinical measures after stroke. Here we present an automated approach to identify chronic ischemic infarcts in addition to other white matter pathologies, that may be used to aid the development of post-stroke management strategies. Our approach uses Bayesian–Markov Random Field (MRF) classification to segment probable lesion volumes present on fluid attenuated inversion recovery (FLAIR) MRI. Thereafter, a random forest classification of the information from multimodal (T1-weighted, T2-weighted, FLAIR, and apparent diffusion coefficient (ADC)) MRI images and other context-aware features (within the probable lesion areas) was used to extract areas with high likelihood of being classified as lesions. The final segmentation of the lesion was obtained by thresholding the random forest probabilistic maps. The accuracy of the automated lesion delineation method was assessed in a total of 36 patients (24 male, 12 female, mean age: 64.57±14.23yrs) at 3months after stroke onset and compared with manually segmented lesion volumes by an expert. Accuracy assessment of the automated lesion identification method was performed using the commonly used evaluation metrics. The mean sensitivity of segmentation was measured to be 0.53±0.13 with a mean positive predictive value of 0.75±0.18. The mean lesion volume difference was observed to be 32.32%±21.643% with a high Pearson's correlation of r=0.76 (p<0.0001). The lesion overlap accuracy was measured in terms of Dice similarity coefficient with a mean of 0.60±0.12, while the contour accuracy was observed with a mean surface distance of 3.06mm±3.17mm. The results signify that our method was successful in identifying most of the lesion areas in FLAIR with a low false positive rate.

Practical Context Awareness Information System for Wireless Vehicular Networks Based on a Core Telematics Platform

The popularization of consumer electronic products and the promotion of digital mobile devices have assisted telematics services in becoming more important and practicable. The global positioning system (GPS) is commonly used today and context awareness technologies are often adopted. To support efficient and effective telecommunications between a huge number of devices and the intelligent transport systems (ITS) services, a scalable networking platform is hence demanded. In this study, we present a novel approach to support an open architecture environment for telematics services based on international standards. The proposed core telematics platform (CTP) can optimize the data transmission under linearly increased time complexity by employing a high-scalable architecture design when a large number of devices connect to the application servers. The proposed context awareness application server (CAAS) approach can optimize context awareness features to provide personal tracking services to children and elders for safety purposes. The simulations and experiments in this study have verified the effectiveness of the proposed approach.

Context-aware features and robust image representations

Local image features are often used to efficiently represent image content. The limited number of types of features that a local feature extractor responds to might be insufficient to provide a robust image representation. To overcome this limitation, we propose a context-aware feature extraction formulated under an information theoretic framework. The algorithm does not respond to a specific type of features; the idea is to retrieve complementary features which are relevant within the image context. We empirically validate the method by investigating the repeatability, the completeness, and the complementarity of context-aware features on standard benchmarks. In a comparison with strictly local features, we show that our context-aware features produce more robust image representations. Furthermore, we study the complementarity between strictly local features and context-aware ones to produce an even more robust representation.

Traffic sign recognition based on a context-aware scale-invariant feature transform approach

A new context-aware scale-invariant feature transform (CASIFT) approach is proposed, which is designed for the use in traffic sign recognition (TSR) systems. The following issues remain in previous works in which SIFT is used for matching or recognition: (1) SIFT is unable to provide color information; (2) SIFT only focuses on local features while ignoring the distribution of global shapes; (3) the template with the maximum number of matching points selected as the final result is instable, especially for images with simple patterns; and (4) SIFT is liable to result in errors when different images share the same local features. In order to resolve these problems, a new CASIFT approach is proposed. The contributions of the work are as follows: (1) color angular patterns are used to provide the color distinguishing information; (2) a CASIFT which effectively combines local and global information is proposed; and (3) a method for computing the similarity between two images is proposed, which focuses on the distribution of the matching points, rather than using the traditional SIFT approach of selecting the template with maximum number of matching points as the final result. The proposed approach is particularly effective in dealing with traffic signs which have rich colors and varied global shape distribution. Experiments are performed to validate the effectiveness of the proposed approach in TSR systems, and the experimental results are satisfying even for images containing traffic signs that have been rotated, damaged, altered in color, have undergone affine transformations, or images which were photographed under different weather or illumination conditions.

Consumer Privacy Enforcement in Context-Aware Web Services

Privacy protection constitutes a genuine human right reflected both in the legislation and in different aspects of software engineering. Sensitive information needs to be protected in end-users interaction with Web Services especially in cases, where context-aware features are included. In this work the authors address the inclusion of consumer privacy preferences in the provision of context-aware Web Services. To achieve this the authors propose, on the one hand, a preferences language, where end-users can specify their privacy options, namely Consumer Privacy Language, and, on the other hand, a seamless enforcement mechanism that considers consumer preferences by intercepting and modifying appropriately Simple Object Access Protocol request and response messages. The enforcement approach has been evaluated based on various execution metrics for an example use case consisting of various Web Services and for different users configurations demonstrating the usefulness of the approach assisting towards the provision of privacy-aware environments.

Loosely coupled service composition for deployment of next generation service overlay networks

In order to provide various attractive services to users, a user-centric approach such as usergenerated contents and services and a platform which can create a service easily and efficiently have been studied and designed during the past two decades. The IEEE NGSON standard has defined a functional architecture that provides advanced overlay functions for value-added services by collaboration of services with features of context awareness, dynamic adaptation and self-organizing management. Its service composition mechanism is comparable with service oriented architecture. However, due to SOA's complexity, it has not become as popular for user-generated services and contents as so-called mashup technologies, which also provide attractive and simple service creation. In this article we propose the adoption of loosely coupled service composition into the NGSON framework that can easily integrate existing communications and data services with little to no dependence on the underlying implementation and service description language details. Furthermore, the NGSON prototype supporting our proposal is presented as a proof of concept and to verify its advantages for operators, service providers, and end users.