Road area detection method based on DBNN for robot navigation using single camera in outdoor environments

2018 ◽  
Vol 45 (2) ◽  
pp. 275-286 ◽  
Author(s):  
K.M. Ibrahim Khalilullah ◽  
Shunsuke Ota ◽  
Toshiyuki Yasuda ◽  
Mitsuru Jindai
Keyword(s):  
Deep Learning ◽  
Robot Navigation ◽  
Cost Effective ◽  
General Purpose ◽  
Single Camera ◽  
Content Type ◽  
The Road ◽  
Outdoor Environments ◽  
Illuminant Invariant ◽  
Navigation Method

Purpose The purpose of this study is to develop a cost-effective autonomous wheelchair robot navigation method that assists the aging population. Design/methodology/approach Navigation in outdoor environments is still a challenging task for an autonomous mobile robot because of the highly unstructured and different characteristics of outdoor environments. This study examines a complete vision guided real-time approach for robot navigation in urban roads based on drivable road area detection by using deep learning. During navigation, the camera takes a snapshot of the road, and the captured image is then converted into an illuminant invariant image. Subsequently, a deep belief neural network considers this image as an input. It extracts additional discriminative abstract features by using general purpose learning procedure for detection. During obstacle avoidance, the robot measures the distance from the obstacle position by using estimated parameters of the calibrated camera, and it performs navigation by avoiding obstacles. Findings The developed method is implemented on a wheelchair robot, and it is verified by navigating the wheelchair robot on different types of urban curve roads. Navigation in real environments indicates that the wheelchair robot can move safely from one place to another. The navigation performance of the developed method and a comparison with laser range finder (LRF)-based methods were demonstrated through experiments. Originality/value This study develops a cost-effective navigation method by using a single camera. Additionally, it utilizes the advantages of deep learning techniques for robust classification of the drivable road area. It performs better in terms of navigation when compared to LRF-based methods in LRF-denied environments.

scholarly journals A deep learning approach to automatic gingivitis screening based on classification and localization in RGB photos

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Wen Li ◽  
Yuan Liang ◽  
Xuan Zhang ◽  
Chao Liu ◽  
Lei He ◽  
...  
Keyword(s):  
Deep Learning ◽  
Low Income ◽  
Dental Health ◽  
Area Under The Curve ◽  
Cost Effective ◽  
General Purpose ◽  
Dental Calculus ◽  
Task Learning ◽  
Large Populations ◽  
Dental Diseases

AbstractRoutine dental visit is the most common approach to detect the gingivitis. However, such diagnosis can sometimes be unavailable due to the limited medical resources in certain areas and costly for low-income populations. This study proposes to screen the existence of gingivitis and its irritants, i.e., dental calculus and soft deposits, from oral photos with a novel Multi-Task Learning convolutional neural network (CNN) model. The study can be meaningful for promoting the public dental health, since it sheds light on a cost-effective and ubiquitous solution for the early detection of dental issues. With 625 patients included in this study, the classification Area Under the Curve (AUC) for detecting gingivitis, dental calculus and soft deposits were 87.11%, 80.11%, and 78.57%, respectively; Meanwhile, according to our experiments, the model can also localize the three types of findings on oral photos with moderate accuracy, which enables the model to explain the screen results. By comparing to general-purpose CNNs, we showed our model significantly outperformed on both classification and localization tasks, which indicates the effectiveness of Multi-Task Learning on dental disease detection. In all, the study shows the potential of deep learning for enabling the screening of dental diseases among large populations.

Deep learning-based detection of tax frauds: an application to property acquisition tax

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Changro Lee
Keyword(s):  
Deep Learning ◽  
Learning Algorithm ◽  
Cost Effective ◽  
Tax Administration ◽  
Content Type ◽  
Effective Manner ◽  
Efficient Sampling ◽  
Tax Returns ◽  
Marketing Tool ◽  
Property Acquisition

PurposeSampling taxpayers for audits has always been a major concern for policymakers of tax administration. The purpose of this study is to propose a systematic method to select a small number of taxpayers with a high probability of tax fraud.Design/methodology/approachAn efficient sampling method for taxpayers for an audit is investigated in the context of a property acquisition tax. An autoencoder, a popular unsupervised learning algorithm, is applied to 2,228 tax returns, and reconstruction errors are calculated to determine the probability of tax deficiencies for each return. The reasonableness of the estimated reconstruction errors is verified using the Apriori algorithm, a well-known marketing tool for identifying patterns in purchased item sets.FindingsThe sorted reconstruction scores are reasonably consistent with actual fraudulent/non-fraudulent cases, indicating that the reconstruction errors can be utilized to select suspected taxpayers for an audit in a cost-effective manner.Originality/valueThe proposed deep learning-based approach is expected to be applied in a real-world tax administration, promoting voluntary compliance of taxpayers, and reinforcing the self-assessing acquisition tax system.

The D-NET software toolkit

2014 ◽  
Vol 48 (4) ◽  
pp. 322-354 ◽  
Author(s):  
Paolo Manghi ◽  
Michele Artini ◽  
Claudio Atzori ◽  
Alessia Bardi ◽  
Andrea Mannocci ◽  
...  
Keyword(s):  
Optimal Solution ◽  
Cost Effective ◽  
Heterogeneous Data ◽  
General Purpose ◽  
Data Models ◽  
Third Party ◽  
Content Type ◽  
Case Examples ◽  
Software Toolkit ◽  
Diverse Application

Purpose – The purpose of this paper is to present the architectural principles and the services of the D-NET software toolkit. D-NET is a framework where designers and developers find the tools for constructing and operating aggregative infrastructures (systems for aggregating data sources with heterogeneous data models and technologies) in a cost-effective way. Designers and developers can select from a variety of D-NET data management services, can configure them to handle data according to given data models, and can construct autonomic workflows to obtain personalized aggregative infrastructures. Design/methodology/approach – The paper provides a definition of aggregative infrastructures, sketching architecture, and components, as inspired by real-case examples. It then describes the limits of current solutions, which find their lacks in the realization and maintenance costs of such complex software. Finally, it proposes D-NET as an optimal solution for designers and developers willing to realize aggregative infrastructures. The D-NET architecture and services are presented, drawing a parallel with the ones of aggregative infrastructures. Finally, real-cases of D-NET are presented, to show-case the statement above. Findings – The D-NET software toolkit is a general-purpose service-oriented framework where designers can construct customized, robust, scalable, autonomic aggregative infrastructures in a cost-effective way. D-NET is today adopted by several EC projects, national consortia and communities to create customized infrastructures under diverse application domains, and other organizations are enquiring for or are experimenting its adoption. Its customizability and extendibility make D-NET a suitable candidate for creating aggregative infrastructures mediating between different scientific domains and therefore supporting multi-disciplinary research. Originality/value – D-NET is the first general-purpose framework of this kind. Other solutions are available in the literature but focus on specific use-cases and therefore suffer from the limited re-use in different contexts. Due to its maturity, D-NET can also be used by third-party organizations, not necessarily involved in the software design and maintenance.

scholarly journals A Deep Learning Approach to Automatic Gingivitis Screening based Onclassification and Localization in RGB Photos

2020 ◽  
Author(s):  
Wen Li ◽  
Yuan Liang ◽  
Xuan Zhang ◽  
Chao Liu ◽  
Lei He ◽  
...  
Keyword(s):  
Deep Learning ◽  
Low Income ◽  
Area Under The Curve ◽  
Cost Effective ◽  
General Purpose ◽  
Dental Calculus ◽  
Task Learning ◽  
Large Populations ◽  
Dental Diseases ◽  
Consistency Evaluation

Abstract Routine dental visit is the most common approach to detect the gingivitis. However, such diagnosis can sometimes be unavailable due to the limited medical resources in certain areas and costly for low-income populations. To increase the availability, this study proposes to screen the existence of gingivitis and its irritants, i.e., dental calculus and soft deposits, from oral photos with a novel Multi-Task Learning convolutional neural network (CNN) model. With 625 patients included in this study, the classification Area Under the Curve (AUC) for detecting gingivitis, dental calculus and soft deposits were 87.11%, 80.11%, and 78.57%, respectively; Meanwhile the box-wise localization sensitivity for gingivitis and dental calculus were 66.57% and 45.61%. Moreover, according to a consistency evaluation with three board-certificated dentists, the model achieved a median score of 3.0/5.0 for reasoning locations of soft deposits without any spatial supervision. By comparing to general-purpose CNNs, we showed our model significantly outperformed on both classification and localization tasks, which indicates the effectiveness of Multi-Task Learning on dental disease detection. The results show the potential of deep learning for enabling cost-effective screening of dental diseases among large populations.

Rent review 2014

2014 ◽  
Vol 32 (5) ◽  
pp. 530-533
Author(s):  
Malcolm Dowden
Keyword(s):  
Cost Effective ◽  
Case Law ◽  
General Purpose ◽  
Third Party ◽  
England And Wales ◽  
Content Type ◽  
House Of Lords ◽  
Special Rules ◽  
Review 2014 ◽  
Supreme Court Rulings

Purpose – The purpose of this legal update is to examine the recent case law relating to rent review in England and Wales. The paper argues that as rent terms have reduced in length, and as market conditions have tended to produce nil-uplifts, there have been relatively few review cases before the court. Cases that reach court tend to fall into two broad categories: contractual interpretation and challenges to third-party determination. Design/methodology/approach – Review and analysis of case law in England and Wales. Findings – There are no special rules for interpreting rent review clauses. The court's approach to contractual interpretation follows House of Lords and Supreme Court rulings culminating in Rainy Sky SA v Kookmin Bank (2011). There are also very limited circumstances in which the court will set aside an arbitrator's award, informed by a policy that favours upholding arbitration awards as a quick and cost-effective way to settle rent review disputes. Practical implications – Rent review clauses must be interpreted in accordance with the normal rules of contractual interpretation. The court is unlikely to be swayed by submissions asserting the “general purpose” of rent review. Originality/value – This is an original analysis of case law.

Radio frequency identification (RFID) stock control and geo-location data system from a moving vehicle

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Craig Proctor-Parker ◽  
Riaan Stopforth
Keyword(s):  
Radio Frequency Identification ◽  
Asset Management ◽  
Cost Effective ◽  
Content Type ◽  
Rfid Tag ◽  
Stock Control ◽  
The Road ◽  
Minimum Requirements ◽  
Frequency Identification ◽  
Locating System

Purpose The purpose of the research has been the primary consideration and evaluation of a cost effective, reliable, robust and simple process of radio frequency identification (RFID)-based stock control, asset management and monitoring of concrete safety bollards used in the road environment. Likewise, the consideration of the use of the same system and technology to other items in and around the general road infrastructure. Design/methodology/approach The research approach undertaken has been an evaluation of the use of currently available RFID technology, with a key emphasis on low cost, ease of use, reliability and convenience. Practical field exercises completed in considering the relevant RFID tags and readers and associated software and apps and necessary software integration and development have been undertaken. At the same time, evaluating the specific limits created in the specific environment is being applied. Of particular interest has been the use of a moving scan in a vehicle drive-through or pass-bye, type reading system. This has been determined to be viable and completely practical, drastically reducing the key issue of time-taken. Practical application of the system from idea to real life application has been undertaken. The integration of the use of the RFID tag and reader system with necessary and related software to database upload and storage has been established. The creation of an online facility to allow the appropriate use of the data and to include the convenient output of an asset report has been undertaken. Findings The findings have provided the necessary insight confirming the use of RFID technology as a simple yet reliable, cost effective and adaptable stock control, asset management and geo-locating system in the road environment. The use of such systems in this particular environment is in its infancy, and is perhaps novel and original in the specific aspect of using the system to stock control, manage and monitor road safety concrete bollards and other roadside objects in the road environment. Originality/value To establish if in fact, stock control geo-locating can be reliably undertaken with the use of RFID tags and readers in the specific road and road construction environment, particularly with the use of moving RFID reading of passive tags. To establish the minimum requirements of a field usable RFID tag and reader, specifically applicable to the concrete safety bollards, however to other roadside furniture. To identify the minimum requirements of a function, simple app to minimise general requirements of the overall stock control and monitoring of the RFID-tagged objects. To establish the possibility of reading the tag data, global positioning system (GPS) location and video imaging footage as a single operation function. To determine the basic parameters or limits of the GPS geo-locating, on the proposed products selected and overall system. To determine the current best practice in respect of reasonable accuracy and detail in relation to price considerations to a fully function stock control and monitoring system. To identify the minimum requirements of an online database to receive, house and provide ongoing access to and report on the data. To identify the key differences and benefits between traditional stock control and monitoring systems, against that of proposed RFID tag, read and geo-locating system.

Wheelchair robot navigation in different weather conditions using deep learning and evolved neural controller

2019 ◽  
Vol 46 (1) ◽  
pp. 146-158 ◽  
Author(s):  
K.M. Ibrahim Khalilullah ◽  
Shunsuke Ota ◽  
Toshiyuki Yasuda ◽  
Mitsuru Jindai
Keyword(s):  
Neural Network ◽  
Dimensionality Reduction ◽  
Robot Navigation ◽  
Weather Conditions ◽  
Single Camera ◽  
Road Detection ◽  
Google Maps ◽  
Neural Controller ◽  
Content Type ◽  
Road Junction

Purpose Wheelchair robot navigation in different weather conditions using single camera is still a challenging task. The purpose of this study is to develop an autonomous wheelchair robot navigation method in different weather conditions, with single camera vision to assist physically disabled people. Design/methodology/approach A road detection method, called dimensionality reduction deep belief neural network (DRDBNN), is proposed for drivable road detection. Due to the dimensionality reduction ability of the DRDBNN, it detects the drivable road area in a short time for controlling the robot in real-time. A feed-forward neural network is used to control the robot for the boundary following navigation using evolved neural controller (ENC). The robot detects road junction area and navigates throughout the road, except in road junction, using calibrated camera and ENC. In road junction, it takes turning decision using Google Maps data, thus reaching the final destination. Findings The developed method is tested on a wheelchair robot in real environments. Navigation in real environments indicates that the wheelchair robot moves safely from source to destination by following road boundary. The navigation performance in different weather conditions of the developed method has been demonstrated by the experiments. Originality/value The wheelchair robot can navigate in different weather conditions. The detection process is faster than that of the previous DBNN method. The proposed ENC uses only distance information from the detected road area and controls the robot for boundary following navigation. In addition, it uses Google Maps data for taking turning decision and navigation in road junctions.

The cost-effectiveness of insulin analogs and regular insulin for diabetes control: a case study in Iran

2020 ◽  
Vol 33 (4/5) ◽  
pp. 323-331
Author(s):  
Mohsen pakdaman ◽  
Raheleh akbari ◽  
Hamid reza Dehghan ◽  
Asra Asgharzadeh ◽  
Mahdieh Namayandeh
Keyword(s):  
Cost Effectiveness ◽  
Cost Effective ◽  
Regular Insulin ◽  
Diabetes Control ◽  
Diabetic Patients ◽  
Insulin Analogs ◽  
Content Type ◽  
Patients With Diabetes ◽  
The Government ◽  
The Cost

PurposeFor years, traditional techniques have been used for diabetes treatment. There are two major types of insulin: insulin analogs and regular insulin. Insulin analogs are similar to regular insulin and lead to changes in pharmacokinetic and pharmacodynamic properties. The purpose of the present research was to determine the cost-effectiveness of insulin analogs versus regular insulin for diabetes control in Yazd Diabetes Center in 2017.Design/methodology/approachIn this descriptive–analytical research, the cost-effectiveness index was used to compare insulin analogs and regular insulin (pen/vial) for treatment of diabetes. Data were analyzed in the TreeAge Software and a decision tree was constructed. A 10% discount rate was used for ICER sensitivity analysis. Cost-effectiveness was examined from a provider's perspective.FindingsQALY was calculated to be 0.2 for diabetic patients using insulin analogs and 0.05 for those using regular insulin. The average cost was $3.228 for analog users and $1.826 for regular insulin users. An ICER of $0.093506/QALY was obtained. The present findings suggest that insulin analogs are more cost-effective than regular insulin.Originality/valueThis study was conducted using a cost-effectiveness analysis to evaluate insulin analogs versus regular insulin in controlling diabetes. The results of study are helpful to the government to allocate more resources to apply the cost-effective method of the treatment and to protect patients with diabetes from the high cost of treatment.

scholarly journals A Generalization Performance Study Using Deep Learning Networks in Embedded Systems

Sensors ◽  
2021 ◽  
Vol 21 (4) ◽  
pp. 1031
Author(s):  
Joseba Gorospe ◽  
Rubén Mulero ◽  
Olatz Arbelaitz ◽  
Javier Muguerza ◽  
Miguel Ángel Antón
Keyword(s):  
Deep Learning ◽  
Embedded Systems ◽  
Embedded System ◽  
General Purpose ◽  
Learning Networks ◽  
Performance Study ◽  
Learning Techniques ◽  
Wide Range ◽  
Learning Architectures

Deep learning techniques are being increasingly used in the scientific community as a consequence of the high computational capacity of current systems and the increase in the amount of data available as a result of the digitalisation of society in general and the industrial world in particular. In addition, the immersion of the field of edge computing, which focuses on integrating artificial intelligence as close as possible to the client, makes it possible to implement systems that act in real time without the need to transfer all of the data to centralised servers. The combination of these two concepts can lead to systems with the capacity to make correct decisions and act based on them immediately and in situ. Despite this, the low capacity of embedded systems greatly hinders this integration, so the possibility of being able to integrate them into a wide range of micro-controllers can be a great advantage. This paper contributes with the generation of an environment based on Mbed OS and TensorFlow Lite to be embedded in any general purpose embedded system, allowing the introduction of deep learning architectures. The experiments herein prove that the proposed system is competitive if compared to other commercial systems.

Sign in / Sign up

Export Citation Format

Share Document