Learning Geometry Information of Target for Visual Object Tracking with Siamese Networks

Recently, Siamese architecture has been widely used in the field of visual tracking, and has achieved great success. Most Siamese network based trackers aggregate the target information of two branches by cross-correlation. However, since the location of the sampling points in the search feature area is pre-fixed in cross-correlation operation, these trackers suffer from either background noise influence or missing foreground information. Moreover, the cross-correlation between the template and the search area neglects the geometry information of the target. In this paper, we propose a Siamese deformable cross-correlation network to model the geometric structure of target and improve the performance of visual tracking. We propose to learn an offset field end-to-end in cross-correlation. With the guidance of the offset field, the sampling in the search image area can adapt to the deformation of the target, and realize the modeling of the geometric structure of the target. We further propose an online classification sub-network to model the variation of target appearance and enhance the robustness of the tracker. Extensive experiments are conducted on four challenging benchmarks, including OTB2015, VOT2018, VOT2019 and UAV123. The results demonstrate that our tracker achieves state-of-the-art performance.

The Use of Location-Based Augmented Reality as Information Media in The Digital Map Application of Jakabaring Sport City

Jakabaring Sport City (JSC) is an integrated sports facility complex and a public tourist spot in Palembang. Due to its extensive area and numerous facilities, visitors need a map with distinct markers for the landmark to make it easier when walking around. Recent technology advancements and its expanding role in various fields invites the question on how to integrate technology in the tourism industry. This journal aims to create an android-based digital map application with augmented reality (AR) as a guiding medium that will display directions towards the Jakabaring exit gate. This app is built using Maps SDK for Android to build the digital maps, Navigation SDK for Android to run navigation, and Maps for Unity to display AR objects that will show the directions towards the exit. The development focuses on a location-based service using GPS to track the user's whereabouts and provide real-time location data around the user. This application has a location search feature, a guiding direction (route and navigation) to a designed location, and an AR camera feature to display where the JSC exit gate is located. Every feature is working well and able to help users while visiting JSC. However, this application could still be improved in the AR feature because on its current version, it is only able to display directions to the JSC exit gate when users are within a radius of one kilometer or less to the exit.

Location Selection Query in Google Maps using Voronoi-based Spatial Skyline (VS2) Algorithm

Google Maps is one of the popular location selection systems. One of the popular features of Google Maps is nearby search. For example, someone who wants to find the closest restaurants to his location can use the nearby search feature. This feature only considers one specific location in providing the desired place choice. In a real-world situation, there may be a need to consider more than one location in selecting the desired place. Assume someone would like to choose a hotel close to the conference hall, the museum, beach, and souvenir store. In this situation, nearby search feature in Google Maps may not be able to suggest a list of hotels that are interesting for him based on the distance from each destination places. In this paper, we have successfully developed a web-based application of Google Maps search using Voronoi-based Spatial Skyline (VS2) algorithm to choose some Point Of Interest (POI) from Google Maps as their considered locations to select desired place. We used Google Maps API to provide POI information for our web-based application. The experiment result showed that the execution time increases while the number of considered location increases.

Maqāṣid al-Sharī‘ah on Islamic banking performance in Indonesia: a knowledge discovery via text mining

Purpose In Indonesia, subjective issues towards the fundamental of Islamic banks (IBs) have been arising. For example, they are claimed to be not in line with the Shari‘ah (Islamic law). Furthermore, the existing scholarly works have not much gained knowledge from the local IBs explaining their efforts in promoting maqasid al-Shariah (objectives of Islamic law). Hence, because religiosity drives the fundamental establishment of IBs, this paper aims to explore the knowledge of how IBs in Indonesia promote maqasid al-Shariah via their published reports. Design/methodology/approach This paper performs text mining from 24 official reports of 5 IBs in Indonesia published from 2015 to 2017. The sample contains 7,162 digital pages and approximately 3,021,618 words. Traditional text mining via human intelligence is first performed to analyse for the numerical data required in the maqasid al-Shariah index (MSI) analysis. Furthermore, a computer-driven text mining using the ‘Text Search’ feature of NVivo 12 Plus is conducted to perform qualitative analysis. These approaches are made to gain relevant knowledge of how the sampled IBs promote maqasid al-Shariah from their published reports. Findings The analysis using the MSI explains a quantified maqasid al-Shariah on the sample’s performance, which indictes the lowest and the highest performing banks. Furthermore, a qualitative analysis supports the evidence from the quantitative analysis. It explains the authors’ coding process that results in 2 parent nodes and 20 child nodes, which contain 435 references coded from the sampled unstructured and bilingual texts. These nodes explain the information that associates with maqasid al-Shariah from the IBs’ reports. These findings explain how maqasid al-Shariah is measured mathematically and represent relevant knowledge of how maqasid al-Shariah is informed practically via digital texts. Research limitations/implications A positivist generalisation is neither intended nor established in this study. Practical implications This paper gains relevant knowledge of how the sampled IBs in Indonesia control and maintain the implementation of maqasid al-Shariah from large textual data. Such knowledge is practically important for IBs stakeholders in Indonesia; moreover to help navigate the Shari‘ah identity of Bank Syariah Indonesia (BSI), the new IB established from the merger of 3 state-owned IBs, which are among the sample of this study. Social implications This paper provides evidence that might best challenge the subjective issue of IBs claiming that they are not in line with the Shari‘ah, particularly in Indonesia. Originality/value This paper is among the pioneers that discover knowledge of how IBs promote maqasid al-Shariah in Indonesia’s banking sector via a text mining approach.

Penerapan Algoritma Sequential Search pada Aplikasi Kamus Bahasa Ilmiah Tumbuhan

Application of Sequential Search Algorithms in the Plant Scientific Language Dictionary Application. Scientific dictionaries such as plants dictionaries are thick and less attractive. Therefore, it is necessary to develop a scientific language dictionary application for plant which is attractive and can be accessed anytime and anywhere. The dictionary application is closely related to the search feature and the sequential search algorithm is used in the search feature at the application development. This algorithm performs the search by matching data in a row from all existing data. The application built had features such as a list of plants with scientific names equipped with A-Z alphabet menus, plant categories, and a search feature to make it easier for users to find the plant names. The results of the search speed performance test based on page load time with the microtime function show that sequential search algorithms can perform the searches quickly for relatively limited and not too much data.Keywords: Scientific Language Dictionary, Plant Dictionary, Sequential SearchAbstrak. Kamus ilmiah seperti kamus ilmiah tumbuhan berbentuk tebal dan kurang menarik. Oleh karena itu, diperlukan pengembangan aplikasi kamus Bahasa ilmiah tumbuhan yang menarik dan dapat diakses kapanpun dan dimanapun. Aplikasi kamus akan berhubungan erat dengan pencarian dan algoritma sequential search digunakan sebagai fitur pencarian pada pengembangan aplikasi kamus bahasa ilmiah tumbuhan ini. Algoritma ini melakukan pencarian dengan mecocokan data secara beruntun dari seluruh data yang ada. Aplikasi yang dibangun memiliki fitur-fitur seperti daftar tumbuhan dengan nama ilmiah yang dilengkapi menu abjad A-Z , kategori tumbuhan, dan fitur pencarian untuk memudahkan pengguna dalam menemukan nama tumbuhan. Hasil pengujian kecepatan pencarian berdasarkan page load time dengan fungsi microtime menunjukkan bahwa algortima sequential search dapat melakukan pencarian dengan cepat untuk data yang relatif terbatas atau tidak terlalu banyak.Kata Kunci: Kamus Bahasa Ilmiah, Kamus Tumbuhan, Sequential Search

Prediction of Hypertension Outcomes Based on Gain Sequence Forward Tabu Search Feature Selection and XGBoost

For patients with hypertension, serious complications, such as myocardial infarction, a common cause of heart failure, occurs in the late stage of hypertension. Hypertension outcomes can lead to complications, including death. Hypertension outcomes threaten patients’ lives and need to be predicted. In our research, we reviewed the hypertension medical data from a tertiary-grade A class hospital in Beijing, and established a hypertension outcome prediction model with the machine learning theory. We first proposed a gain sequence forward tabu search feature selection (GSFTS-FS) method, which can search the optimal combination of medical variables that affect hypertension outcomes. Based on this, the XGBoost algorithm established a prediction model because of its good stability. We verified the proposed method by comparing other commonly used models in similar works. The proposed GSFTS-FS improved the performance by about 10%. The proposed prediction method has the best performance and its AUC value, accuracy, F1 value, and recall of 10-fold cross-validation were 0.96. 0.95, 0.88, and 0.82, respectively. It also performed well on test datasets with 0.92, 0.94, 0.87, and 0.80 for AUC, accuracy, F1, and recall, respectively. Therefore, the XGBoost with GSFTS-FS can accurately and effectively predict the occurrence of outcomes for patients with hypertension, and can provide guidance for doctors in clinical diagnoses and medical decision-making.

The search for science and technology verses in Qur’an and hadith

Currently, the Vector Space Model algorithm has been widely implemented for the document search feature because of its reliability in retrieving information. One of them in the search for verses of the Qur'an based on the translation. However, if the phrase or word used is different (even though it has one meaning) with the word in the document in the database, the system will not display the verse. As we know that the Qur'an has a very deep meaning, so an interpretation of the verse is needed. Therefore, this research focuses on implementing the Vector Space Model (VSM) algorithm for searching verses and hadiths in science and technology by using the discussion parameters of these verses or hadiths. The test results obtained with 20 keyword samples using metric recall were 81% with an average time of 2.24 seconds.

Mobility Need-Adaptive Housing Platforms: The Benefit of a Commute Time Search Feature

The growing influx of people to urban areas has resulted in a tense housing market in many places, making the search for a suitable residence an increased challenge. Dedicated online platforms facilitate this process and offer two distinct approaches to find suitable accommodations concerning its location. Traditionally, users can search for a general area like a city to narrow down the results displayed. Additionally, some platforms offer searches based on the maximum commute time between apartments and points of interest. This paper investigates the benefit such approaches yield concerning technology acceptance and the fit of the task and information representation. Thus, a prototypically implemented online platform with and without a commute time search feature was evaluated in an online experiment. The treatment specification achieved significantly better results in terms of information quality and technology acceptance, implicating that such a design should be preferred for websites that facilitate the search for apartments. These insights can contribute to an enhanced understanding of visual system design to reduce the negative sustainability impacts of traffic induced by a divergence of residential and workplaces.