Fully convolutional network has predicted multiple class dense outputs in CT image labels and obtained significant improvements in segmentation tasks. In this paper, we present a joint multi-path fully convolutional network (MFCN) with random forests (RF) architecture for abdominal organs segmentation automatically. First, in coarse segmentation step, three FCNs are trained respectively with three orthogonal directions which consider contextual and spatial information of fusion layers adequately. In classification step, using features extracted from different layers of network and normalizing them to mean value as supervoxel representation to train RF. This allows the computation of supervoxel at each orientation achieve high efficiency. Finally, we aggregate the results of MFCN and RF on voxel-wise and perform conditional random fields (CRF) focuses on smoothing borders of fine segmentation regions. We exceeds the state-of-the-art methods and get achievable DSC values for our work is 90.1%, 88.4%, 88.0%, 88.6% represent liver, right and left kidney, spleen respectively.
Cyber-attacks represent a serious threat to public authorities and their agencies are an attractive target for hackers. The public sector as a whole collects lots of data on its citizens, but that data is often kept on vulnerable systems. Especially for Local Public Administrations (LPAs), protection against cyber-attacks is an extremely relevant issue due to outdated technologies and budget constraints. Furthermore, the General Data Protection Regulation (GDPR) poses many constraints/limitations on the data usage when “special type of data” is processed. In this paper the approach of the EU project COMPACT (H2020) is presented and the solutions used to guarantee the data privacy during the real time monitoring performed by the COMPACT security tools are highlighted.
Internets of Things (IoT) and Big Data applications and services have spread and are rapidly being deployed in the information services of the healthcare and financial industries, etc. However, the previous paper suggested that the current IoT services were individually developed, therefore, the open platform and architecture for the above IoT services of the healthcare industries should be deemed necessary, while the Big Data applications prevail in healthcare industry gradually. An open healthcare platform is expected to promote and implement the digital IT applications for healthcare communities efficiently. In this paper, we suggest that various IoT and Big Data applications will be designed and verified while the open platform for healthcare related IoT services should be proposed and verified by the research initiative named “Open Healthcare Platform 2030 – OHP2030”. In addition, the vision paper for enabling Digital Healthcare applications in the above OHP2030 research initiative is explained.
In this paper, a novel two-staged pavement image processing framework is presented. The pavement images are classified into four general categories in the first stage, so that the images can be processed using category-specific algorithms in the 2nd stage. The proposed algorithm first fuses a local contrast enhanced image with a global grayscale corrected image to obtain an enhanced distressed pavement image. The enhanced image is then decomposed with a three-layer wavelet transform to obtain three texture features of the entire image including High-Amplitude Wavelet Coefficient Percentage (HAWCP), the High-Frequency Energy Percentage (HFEP), and the Standard Deviation (STD). In the meantime, an improved P-tile method is used to obtain the binary image. From the binary image, three additional shape features are extracted including the Average Area of all Connected Components (AA), the Area of the Maximum Connected Component (AM), and the Equivalent Length of the longest Connected Component (EL). Finally, a BP neural network is used to fuse both the texture and shape features sequentially to achieve the initial classification. Experimental results show that for the four types of pavement images, the proposed algorithm achieves an effective classification of the pavement distress image with the accuracy rates of 96.5%, 91.4%, 95.2% and 98.1% respectively, which are higher than those of the classification algorithm with a single-type feature.
Due to the difference in operational characteristic, heavy vehicles have been viewed as a hindrance in traffic flow and capacity analysis. The emergence of passenger car equivalents (PCE) can assist traffic agencies in better understanding the impact of heavy vehicles on passenger vehicles in the mixed traffic stream, by converting a heavy vehicle of a subject class into the equivalent number of passenger cars. However, according to existing literature, most researchers have devoted to the estimation of PCE for basic freeway sections. Therefore, in this study, we explore the variation of heavy vehicle PCE for on-ramp adjacent zones under varying traffic volume. A one-lane on-ramp in Queensland, Australia, is selected for a case study and four existing PCE approaches are applied in the calculation of PCE. They are homogenization based method, time headway based method, traffic flow based method, and multiple regression method, respectively. The final PCE values are compared to those derived from VISSIM simulation model. The following conclusions are drawn: (1) homogenization based method cannot reveal the variation trend of PCE factors over traffic volume; (2) the results obtained through time headway and traffic flow based methods are more consistent with outcome from simulation model.
The Gold Coast Light Rail project represents a significant investment in public transportation infrastructure on the Gold Coast and is one of the key initiatives set out in City of Gold Coast Council’s Transport Strategy 2031. This paper conducts an impact analysis of Stage 2 on the peripheral transportation network. To assess the anticipated impacts, a traffic simulation was conducted in PTV Visum 16 for the base year of 2018 and forecasted for year 2031. For the base and forecasted models, a scenario was prepared both with and without Stage 2 of the system to allow a comparison of results between the variants. According to the results, it is found that Stage 2 has a positive impact on the transportation network by successfully implementing an integrated transport solution encouraging users to switch to public transport.
This paper proposes a novel approach, called dynamic integrated slack-based measure (DISBM) modeling approach, to evaluating multi-period non-radial slacks of non-storable production characterized with carry-over activities. The proposed modeling approach incorporates conventional dynamic slack-based measure (DSBM) technical efficiency and service effectiveness into data envelopment analysis (DEA) modeling such that multi-period input excesses, output shortages and consumption gaps can be simultaneously determined. Some important properties of the proposed DISBM modeling are explored. A case study on the efficiency and effectiveness of Taiwan’s intercity bus transport during 2007–2010 is presented. The results indicate that the proposed DISBM modeling is superior to conventional DSBM modeling in terms of benchmarking power, and that the non-radial slacks associated with input, output and consumption variables do provide rational information to rectify the inefficient and/or ineffective units throughout the production process.
One of the core features of social robotics system is a physical interaction between humans and humanoid robots. This provides additional challenges, both from safety and usability prospectives. When dealing with human-robot interaction, human safety has the highest priority. While in industrial environment we have robot cells to protect humans, in social robotics, that we consider, physical contact is possible, as well as other interactions, with consequences that might be in psychological areas. For example, the conversation with children might have different requirements in comparison to the conversation with adults, the behavioural assumptions might be different, etc. This paper summarises the core results of a project on social robotics system, where an autonomous humanoid robot guides visitors through a lab tour. The results of our work were implemented on the humanoid PAL REEM robot. The implementation includes a web-application to support the management of robot-guided tours. The application also provides recommendations for the users as well as allows for a visual analysis of historical data on the tours.
Remote consultation and diagnosis platform has been widespread in market for group diagnosis, education and etc. It provides convenience for people enjoying superior quality of medical service. However, most of platforms cannot meet increasing demands in practice. The most important reason is that they have troubles of navigating in complicated Internet condition. In general, remote rendering is preferred to support data transmission, by which server renders images locally, captures screenshot and delivers them to clients synchronously. As a result of neighborhood DICOM slices serving a high similarity, following frame is compressed according to the difference with last one. Even though necessary demand of bandwidth has been reduced a lot, remaining volume proves too large especially remote 3D volume rendering. In this paper, we proposed a novel method bit difference compression transmission. Compared with traditional algorithm, it redesigns a new data structure, which gives more significances to bit. Consequently, demands of network fall into a desirable amount. Specifically, we improved solution of 3D images sharing and collaboration. Instead of terrible screenshots, synchronization of cameras in clients has been introduced. With a minimal cost our platform realized the cumbersome process. Collected data from experiments demonstrated our proposed methods have an obvious superiority and run robust in different environment of Internet.
Shared-parking lot brings utilization improvement, but also has its disadvantage compared with traditional parking lot while they are competing for public users. In the market including both shared-parking lot and traditional parking lot, parking lot operators need to know how to deal with parking price to be competitive in the market. The Hotelling model is applied in this paper to study the product differentiation of traditional parking lot and shared-parking lot, with some equilibrium analyses to figure out equilibrium parking prices of both parking lots while considering their competition in the market. Two points of indifferent consumers exist in the competition of the traditional parking lot and the shared- parking lot.
This paper provides design and evaluation of two signal timing strategies with five cases respectively under different degree of weather condition. The main aim of this project is finding out the most suitable signal timing strategies depending on the weather condition by comparing the analysis results. The procedure mainly includes two parts in traffic signal design, which are signal timing plans design and performance analysis. Methods of design and performance analysis include Webster’s signal timing method, Akcelik’s signal timing method and HCM delay method. It is clear that the weather factor has huge impact on performance of signalized intersection. Apparently, compared to strategy B, strategy A is more suitable for bad weather condition.
Information technology has been focused to estimate growth degree of plants in agriculture. This paper focuses on leaf temperature that changes according to the activity of photosynthesis. Infrared cameras are a major method to measure leaf temperature in conventional methods. However, the expensive device price causes difficulty to install many sensors in practical fields. Infrared radiation sensors are new candidate device to estimate growth state by measuring leaf temperature. Since the price of infrared radiation sensors is inexpensive, we can install a lot of sensors into fields. Additionally, the consumed power of infrared radiation sensors is relatively small comparing to Infrared cameras. These features of infrared radiation sensors are appropriate for sensor networks working with a battery. This paper proposes a field sensor network to measure growth state of plants by infrared radiation sensors. Our goal is to realize a practical and inexpensive sensor network system with typical system on chip (SoC). Therefore, we employ a reasonable price SoC supporting IEEE 802.15.4 standard to design a unique device with various sensors. In order to realize multi-hop communication with low-power consumption, we propose a routing and media access control mechanisms for the developed system. The media access control technology realizes periodic sleep operation of all devices to enable long-term operation of the system. The routing control technology can construct a multi-hop network with the minimum number of hops. The experimental results demonstrated that the development system works in the practical fields.
Computer Graphics technology enables a three-dimensional representation of object’s shape and inner structure. It is widely used in the field of visualization and simulation such as computer-aided design, scientific visualization, and medical simulation. Recent studies on implicit surface generation from shape measured three-dimensional point cloud data provide precise and refined surface visualization for complex objects from buildings and tangible heritages to the internal structure of the human body. However, to understand and analyze the structural characteristics of complex shapes, conventional methods, which visualize the whole object with one criterion, could not produce satisfactory results. A more comprehensive visualization method that extracts and highlights the edges and feature regions of a complex object is desired. In this paper, we propose a fused visualization method that extracts and highlights the shape characteristics of three-dimensional volume data of the human body. For the implicit surface generation, volume stochastic process sampling method is applied. The surface curvature is then calculated by projecting the mathematically well-defined curvature information at a point on the iso-surface to its tangent plane. The high curvature area is extracted as the feature region and transparently fused with the original volume data. The proposed method, which realizes three-dimensional transparent fusion of feature-highlighted iso-surface visualization and volume visualization, comprehensively visualizes global structure of the target medical data as well as emphasizes the structural characteristics in the feature region.
Word representations are mathematical items capturing a word’s meaning and its grammatical properties in a machine-readable way. They map each word into equivalence classes including words sharing similar properties. Word representations can be obtained automatically by using unsupervised learning algorithms that rely on the distributional hypothesis, stating that the meaning of a word is strictly connected to its context in terms of surrounding words. This assessed notion of context has been recently reconsidered in order to include both distributional and morphological features of a word in terms of characters co-occurrence. This approach has evidenced very promising results, especially in NLP tasks, e.g, POS Tagging, where the representation of the so-called Out of Vocabulary (OOV) words represents a partially solved issue. This work is intended to face the problem of representing OOV words for a POS Tagging task, contextualized to the Italian language. Potential benefits and drawbacks of adopting a Bidirectional Long Short Term Memory (bi-LSTM) fed with a joint character and word embeddings representation to perform POS Tagging also considering OOV words have been investigated. Furthermore, experiments have been performed and discussed by estimating qualitative and quantitative indicators, and, thus, suggesting some possible future direction of the investigation.
Internets of Things (IoT) applications and services have spread and are rapidly being deployed in the information services of the healthcare and financial industries, etc. However, the previous paper suggested that the current IoT services were individually developed, therefore, the open platform and architecture for the above IoT services of the healthcare industries should be deemed necessary. An open healthcare platform is expected to promote and implement the digital IT applications for healthcare communities efficiently. In this paper, we suggest that the open platform for healthcare related IoT services will be proposed and verified by the research initiative named “Open Healthcare Platform 2030 – OHP2030”. In addition, the vision for the OHP2030 research initiative is expressed.
A time session variability between the enrollment data and the recognized data degrades speaker recognition performance. Hence, the time session variability is one of the most important issues in the speaker recognition technology. In this paper, we propose a robust speaker recognition method for the time session variability. The proposed method estimates a time session variability subspace. Then, the proposed method carries out the speaker recognition in the orthogonal complement of the time session variability subspace. In addition, we incorporate a linear discriminant analysis method into the proposed method. In order to evaluate the proposed method, we conducted a speaker identification experiment. Experimental results show that the proposed method improves speaker identification performance of baseline.
Cloud computing has become so popular that most sensitive data are hosted on the cloud. This fast-growing paradigm has brought along many problems, including the security and integrity of the data, where users rely entirely on the providers to secure their data. This paper investigates the use of the pattern fragmentation to split data into chunks before storing it in the cloud, by comparing the performance on two different cloud providers. In addition, it proposes a novel approach combining a pattern fragmentation technique with a NoSQL database, to organize and manage the chunks. Our research has indicated that there is a trade-off on the performance when using a database. Any slight difference on a big data environment is always important, however, this cost is compensated by having the data organized and managed. The use of random pattern fragmentation has great potential, as it adds a layer of protection on the data without using as much resources, contrary to using encryption.
Drones have been considered for use in various fields according to the performance improvement and the price down of devices. They are expected for some applications: disaster relief, farm field, security field, transportation field, etc. Some companies will employ the autopilot system for their business. However, they have to switch to manual operation in case of emergency due to the autopilot safety is not guaranteed. Therefore, a pilot must connect with the drone continuously by the network for remote monitoring. Cellular network systems are the candidate networks for remote monitoring. However, typical design of cellular networks does not assume user equipment devices in the air because antennas of cellular networks are usually aimed downward to reduce inter-cell interference. This means that drones may fly out a communication area of cellular networks. Therefore, business drones must communicate with some cellular networks to keep continuous communication. However, IP-based application will disconnect due to change of cellular networks. As a result, practical business drones’ operations require a continuous communication mechanism. This paper proposes a continuous remote control architecture for drone operations to improve safety of the autopilot function. The proposed architecture employs NTMobile technology as a seamless mobility protocol supporting continuous communication. Additionally, it also employs IP-based remote control application to control drones remotely. The evaluation system can acquire sensor information and exchange control information continuously when drones switch access networks. The proposed architecture can be a fundamental framework to realize a wide area drone operation service.
In this paper, we present a novel touchless interaction system for visualization of hepatic anatomical models in surgery. Real-time visualization is important in surgery, particularly during the operation. However, it often faces the challenge of efficiently reviewing the patient’s 3D anatomy model while maintaining a sterile field. The touchless technology is an attractive and potential solution to address the above problem. We use a Microsoft Kinect sensor as input device to produce depth images for extracting a hand without markers. Based on this representation, a deep convolutional neural network is used to recognize various hand gestures. Experimental results demonstrate that our system can significantly improve the response time while achieve almost same accuracy compared with the previous researches.
In this research report we present the current state of the transition from traditional, internal combustion engines vehicles, to electrical vehicles. The main characteristic of this transition is that new generation cars are matching the cost and performance of traditional petrol cars. Transition to electric vehicles is driven by the environmental sustainability, in the first place, economy, government policies, inherent automotive industry dynamics and consumer preferences. Transition is presented from global perspective in addition to specificities in Australian context. The conclusion is that such transition is a major disruption affecting the whole economy. It is characterized by the convergence of mobility and energy what can bring significant benefits to the entire society.
