Semantic image segmentation based on Convolutional Neural Networks (CNNs) is one of the main approaches in computer vision area. In convolutional neural network-based approaches, a pre-trained CNN which is trained on the large image classification datasets is generally used as a backend to extract features (image descriptors) from the images. Whereas, the special size of output features from CNN backends are smaller than the input images, by stacking multiple deconvolutional layers to the last layer of backend network, the dimension of output will be the same as the input image. Segmentation using local image descriptors without involving relationships between these local descriptors yield weak and uneven segmentation results. Inspired by these observations, in this research we propose Context-Aware Features (CAF) unit. CAF unit generate image-level features using local-image descriptors. This unit can be integrated into different semantic image segmentation architectures. In this study, by adding the proposed CAF unit to the Fully Convolutional Network (FCN) and DeepLab-v3-plus base architectures, the FCN-CAF and DeepLab-v3-plus-CAF architectures are proposed respectively. PASCAL VOC2012 datasets have been used to train the proposed architectures. Experimental results show that the proposed architectures have 2. 7% and 1. 81% accuracy improvement (mIoU) compared to the related basic architectures, respectively.

Generalization is a prevalent concept in Cartography to which has been added new aspects such as model generalization with developments in GIS. Increasing demand for tailored and ubiquitous geospatial services like wayfinding, makes the context-aware generalization a noticeable research area in GI Science. Most of the wayfinding services use the network data model as the main spatial data model for their analyses. Whatever data or information that characterizes the situations relevant to users, systems and applications, can be considered as context. A context-aware service is as a service which can sense user, environment and device’s situations and respond to user requests concerning contexts to fulfill the user’s needs better. From the GI services perspective, the context of a query could be the location of the device, environmental settings that query is made in, the time, the activity of the user, user’s personal information, the user’s favorites and information needs, the user’s cognitive map of environment, the mode of travel, the purpose of travel and the device’s technological specifications.In this paper, we try to propose and implement a method for context-aware network generalization at the analysis level. For finding the best path, user contexts like user’s favorite streets will be used. These contexts are modeled as edge’s attributes and those edges which fulfill user’s needs, will be the generators of the Network Voronoi Diagram. With these diagrams, the network will be simplified into sub graphs using Delaunay diagram over the network. The path would be composed of the path between origin and destination to their corresponding generators and the path between generators. This method guarantees the maximum use of edges with user’s need context as well as decreasing computational cost.

Background and Aim: Among the places that context-aware systems and services would be very useful, are libraries. The purpose of this study is to achieve a coherent definition of context aware systems and applications, especially in digital libraries.Method: This was a review article that was conducted by using Library method by searching articles and e-books on websites and databases.Results: Findings of this study indicate that context-aware services in digital libraries by understanding specific conditions of each user - such as demographic characteristics, time position, location and collecting and analyzing of its data - could provide smart and appropriate services.Conclusion: Digital libraries are constantly evolving and moving forward and must coordinate themselves with on-going changes. Equipping libraries with context-aware services lead them to best meet information needs of their users, without limitation of time and place.

Context-aware systems serve the user by providing information that aligns with the preferences and context of the user. This study aims to recognize the contexts related to the entities of electronic theses and dissertations. Research is of the qualitative type that has used the documentary research method to find concepts, make coding, and classify contexts. The contribution of this research in increasing knowledge is to recognize the contexts of the users, electronic theses, and dissertation "ETD" and the system entities. The findings showed that the contexts of users are classified into 10 categories: user identity, interests and preferences, activities and history, social communication, computer context, user locations, user times, user access level, user status, and physical environment. The contexts of ETD have 8 categories: ETD identity, access level, audiences, activities and events, ETD content, time, place, and physical format of ETD. System contexts were grouped into 8 categories: system characteristics, user interaction capabilities, indexing capability, systems communication, user’s data storage, computing context, activity, and location. Conducting this research gives developers and decision-makers of ETD systems knowledge of contexts to redesign the system by considering the capabilities of context-aware systems to respond to users' needs, recommend documents, retrieve user-oriented documents, and develop system compatibility procedures. In the present study, we have tried to recognize all the contexts in an ETD system, and not only the user contexts but also the ETD contexts and the system contexts have been considered. context-aware ETD developers need some of the recognized contexts based on the type of context-aware services

Visual Question Answering (VQA) is a complex task that requires models to jointly analyze visual and textual inputs to generate accurate answers. Reasoning and inference are critical for addressing questions that involve relationships, spatial arrangements, and contextual details within an image. In this study, we propose a model based on the BLIP framework, as a generative model, that enhances contextual understanding by incorporating dense-captions -detailed textual descriptions generated for specific regions within an image- along with spatial information extracted from the image. The model focuses on emphasizing visual information and extracting additional context to improve answer accuracy. Experimental results on the GQA dataset demonstrate that the proposed approach achieves competitive performance compared to state-of-the-art methods

Stance detection is the process of identifying and classifying an author's point of view or stance towards a specific target in a given text. Most of previous studies on stance detection neglect the contextual information hidden in the input data and as a result lead to less accurate results. In this paper, we propose a novel method called ConSPro, which uses decoder-only transformers to consider contextual input data in the process of stance detection. First, ConSPro applies zero-shot prompting of decoder only transformers to extract the context of target in the input data. Second, in addition to target and input text, ConSPro uses the extracted context as the third type of parameter for the ensemble method. We evaluate ConSPro on SemEval2016 and the empirical results indicate that ConSPro outperforms the non-contextual approaches methods, on average 9% with respect to f-measure. The findings of this study show the strong capabilities of zero-shot prompting for extracting the informative contextual information with significantly less effort comparing to previous methods on context extraction.

Nowadays, in the age of communication, the production and transmission of information are increasing rapidly. There are also many destinations in the tourism sector that are worth traveling, but increasing the number of suitable tourist destinations through the high volume of choices makes it harder to choose the desired option. Therefore, the prioritization and personalization of the information have become a necessity for providing appropriate and effective recommendation. In this paper, considering the target user's location, a tourist recommendation system is proposed to suggest a perfect place for tourist accommodation using context information, trust criteria and graph clustering to increase the accuracy of recommendations. The experiments and evaluations on TripAdvisor dataset show that the proposed method suggests better results than other state-of-the-art methods.