Search

The proportion of pet keeping has increased significantly. According to the survey results of Business Next, the proportion of Taiwan families keeping pets was 70% in 2020. Among them, the total number of fish pets was close to 33% of the overall pet proportion. Therefore, aquarium pets have become indispensable companions for families. At present, many studies have discussed intelligent aquarium systems. Through image recognition based on visual sensors, we may be able to detect and interpret the physiological status of the fish according to their physiological appearance. In this way, it can help to notify the owner as soon as possible to treat the fish or isolate them individually, so as to avoid the spread of infection. However, most aquarium pets are kept in groups. Traditional image recognition technologies often fail to recognize each fish's physiological states precisely because of fish swimming behaviors, such as grouping swimming, shading with each other, flipping over, and so on. In view of this, this paper tries to address such problems and then proposes a practical scheme, which includes three phases. Specifically, the first phase tries to enhance the image recognition model for small features based on the prioritizing rules, thus improving the instant recognition capability. Then, the second phase exploits a designed fish-ID tracking mechanism and analyzes the physiological state of the same fish-ID through coherent frames, which can avoid temporal misidentification. Finally, the third phase leverages a fish-ID correction mechanism, which can detect and correct their IDs periodically and dynamically to avoid tracking confusion, and thus potentially improve the recognition accuracy. According to the experiment results, it was verified that our scheme has better recognition performance. The best accuracy and correctness ratio can reach up to 94.9% and 92.67%, which are improved at least 8.41% and 26.95%, respectively, as compared with the existing schemes. [ABSTRACT FROM AUTHOR]

Globally, the incidence rate for breast cancer ranks first. Treatment for early-stage breast cancer is highly cost effective. Five-year survival rate for stage 0–2 breast cancer exceeds 90%. Screening mammography has been acknowledged as the most reliable way to diagnose breast cancer at an early stage. Taiwan government has been urging women without any symptoms, aged between 45 and 69, to have a screening mammogram bi-yearly. This brings about a large workload for radiologists. In light of this, this paper presents a deep neural network (DNN)-based model as an efficient and reliable tool to assist radiologists with mammographic interpretation. For the first time in the literature, mammograms are completely classified into BI-RADS categories 0, 1, 2, 3, 4A, 4B, 4C and 5. The proposed model was trained using block-based images segmented from a mammogram dataset of our own. A block-based image was applied to the model as an input, and a BI-RADS category was predicted as an output. At the end of this paper, the outperformance of this work is demonstrated by an overall accuracy of 94.22%, an average sensitivity of 95.31%, an average specificity of 99.15% and an area under curve (AUC) of 0.9723. When applied to breast cancer screening for Asian women who are more likely to have dense breasts, this model is expected to give a higher accuracy than others in the literature, since it was trained using mammograms taken from Taiwanese women. [ABSTRACT FROM AUTHOR]

Forcing pathways between urban surfaces (impervious and pervious pavers) and near-surface air temperature were measured and investigated with a network of multiple sensors. Utilizing field data measured between April 2021 and May 2022, and assuming that the influential variables follow the basic heat-transfer energy-balance equations, multiple regression-based statistical models were built to predict the surface temperature and near-surface air temperature (0.05 m, 0.5 m, 1 m, 2 m, and 3 m) of one impervious paver site and one pervious paver site in Taipei City, Taiwan. Evaporative cooling was found to be more influential on the pervious paver with a statistically significant influence on the microclimate up to 1.8 m (and up to 0.7 m for the impervious paver), using in situ data with an ambient air temperature higher than 24 °C. The surface temperature is mainly affected by solar shortwave radiation and ambient air temperature. As for near-surface air temperature, ambient air temperature is the most influential factor, followed by surface temperature. The importance of surface temperature indicates the influence of upwelling longwave radiation on the microclimate. The predictive equations show that pervious surfaces can help cities with hot and humid climates fight the changing climate in the future. [ABSTRACT FROM AUTHOR]

Fine aerosols with a diameter of less than 2.5 microns (PM2.5) have a significant negative impact on human health. However, their measurement devices or instruments are usually expensive and complicated operations are required, so a simple and effective way for measuring the PM2.5 concentration is needed. To relieve this problem, this paper attempts to provide an easy alternative approach to PM2.5 concentration estimation. The proposed approach is based on image processing schemes and a simple linear regression model. It uses images with a high and low PM2.5 concentration to obtain the difference between these images. The difference is applied to find the region with the greatest impact. The approach is described in two stages. First, a series of image processing schemes are employed to automatically select the region of interest (RoI) for PM2.5 concentration estimation. Through the selected RoI, a single feature is obtained. Second, by employing the single feature, a simple linear regression model is used and applied to PM2.5 concentration estimation. The proposed approach is verified by the real-world open data released by Taiwan's government. The proposed scheme is not expected to replace component analysis using physical or chemical techniques. We have tried to provide a cheaper and easier way to conduct PM2.5 estimation with an acceptable performance more efficiently. To achieve this, further work will be conducted and is summarized at the end of this paper. [ABSTRACT FROM AUTHOR]

Developing on-site earthquake early warning systems has been a challenging problem because of time limitations and the amount of information that can be collected before the warning needs to be issued. A potential solution that could prevent severe disasters is to predict the potential strong motion using the initial P-wave signal and provide warnings before serious ground shaking starts. In practice, the accuracy of prediction is the most critical issue for earthquake early warning systems. Traditional methods use certain criteria, selected through intuition or experience, to make the prediction. However, the criteria thresholds are difficult to select and may significantly affect the prediction accuracy. This paper investigates methods based on artificial intelligence for predicting the greatest earthquake ground motion early, when the P-wave arrives at seismograph stations. A neural network model is built to make the predictions using a small window of the initial P-wave acceleration signal. The model is trained by seismic waves collected from 1991 to 2019 in Taiwan and is evaluated by events in 2020 and 2021. From these evaluations, the proposed scheme significantly outperforms the threshold-based method in terms of its accuracy and average leading time. [ABSTRACT FROM AUTHOR]

In this study, a low-cost, software-defined Global Positioning System (GPS) and Satellite-Based Augmentation System (SBAS) Reflectometry (GPS&SBAS-R) system has been built and proposed to measure ocean-surface wave parameters on board the research vessel New Ocean Researcher 1 (R/V NOR-1) of Taiwan. A power-law, ocean-wave spectrum model has been used and applied with the Small Perturbation Method approach to solve the electromagnetic wave scattering problem from rough ocean surface, and compared with experimental seaborne GPS&SBAS-R observations. Meanwhile, the intensity scintillations of high-sampling GPS&SBAS-R signal acquisition data are thought to be caused by the moving of rough surfaces of the targeted ocean. We found that each derived scintillation power spectrum is a Fresnel-filtering result on ocean-surface elevation fluctuations and depends on the First Fresnel Zone (FFZ) distance and the ocean-surface wave velocity. The determined ocean-surface wave speeds have been compared and validated against nearby buoy measurements. [ABSTRACT FROM AUTHOR]

Kinmen, the famous Cold War island also known as Quemoy, is a typical island with isolated power grids. It considers the promotion of renewable energy and electric charging vehicles to be two essential strategies to achieve the goal of a low-carbon island and smart grid. With this motivation in mind, the main objective of this study is to design and deploy an energy management system for hundreds of current PV sites distributed on the island, energy storage systems, and charging stations on the island. In addition, the real-time acquisition of the data for power generation, power storage, and power consumption systems will be used for future demand and response analysis. Moreover, the accumulated dataset will also be utilized for the forecast or prediction of renewable energy generated by the PV systems or power consumed by the battery units or charging stations. The results of this study are promising since a practical, robust, and workable system and database are developed and implemented with a variety of Internet of Things (IoT), data transmission technologies, and the hybrid of on-premises and cloud servers. Users of the proposed system can remotely access the visualized data through the user-friendly web-based and Line bot interfaces seamlessly. [ABSTRACT FROM AUTHOR]

In recent years, the technology of artificial intelligence (AI) and robots is rapidly spreading to countries around the world. More and more scholars and industry experts have proposed AI deep learning models and methods to solve human life problems and improve work efficiency. Modern people's lives are very busy, which led us to investigate whether the demand for Bento buffet cafeterias has gradually increased in Taiwan. However, when eating at a buffet in a cafeteria, people often encounter two problems. The first problem is that customers need to queue up to check out after they have selected and filled their dishes from the buffet. However, it always takes too much time waiting, especially at lunch or dinner time. The second problem is sometimes customers question the charges calculated by cafeteria staff, claiming they are too expensive at the checkout counter. Therefore, it is necessary to develop an AI-enabled checkout system. The AI-enabled self-checkout system will help the Bento buffet cafeterias reduce long lineups without the need to add additional workers. In this paper, we used computer vision and deep-learning technology to design and implement an AI-enabled checkout system for Bento buffet cafeterias. The prototype contains an angle steel shelf, a Kinect camera, a light source, and a desktop computer. Six baseline convolutional neural networks were applied for comparison on food recognition. In our experiments, there were 22 different food categories in a Bento buffet cafeteria employed. Experimental results show that the inception_v4 model can achieve the highest average validation accuracy of 99.11% on food recognition, but it requires the most training and recognition time. AlexNet model achieves a 94.5% accuracy and requires the least training time and recognition time. We propose a hierarchical approach with two stages to achieve good performance in both the recognition accuracy rate and the required training and recognition time. The approach is designed to perform the first step of identification and the second step of recognizing similar food images, respectively. Experimental results show that the proposed approach can achieve a 96.3% accuracy rate on our test dataset and required very little recognition time for input images. In addition, food volumes could be estimated using the depth images captured by the Kinect camera, and a framework of visual checkout system was successfully built. [ABSTRACT FROM AUTHOR]

Online social media platforms play an important role in political communication where users can freely express and exchange their political opinion. Political entities have leveraged social media platforms as essential channels to disseminate information, interact with voters, and even influence public opinion. For this purpose, some organizations may create one or more accounts to join online political discussions. Using these accounts, they could promote candidates and attack competitors. To avoid such misleading speeches and improve the transparency of the online society, spotting such malicious accounts and understanding their behaviors are crucial issues. In this paper, we aim to use network-based analysis to sense influential human-operated malicious accounts who attempt to manipulate public opinion on political discussion forums. To this end, we collected the election-related articles and malicious accounts from the prominent Taiwan discussion forum spanning from 25 May 2018 to 11 January 2020 (the election day). We modeled the discussion network as a multilayer network and used various centrality measures to sense influential malicious accounts not only in a single-layer but also across different layers of the network. Moreover, community analysis was performed to discover prominent communities and their characteristics for each layer of the network. The results demonstrate that our proposed method can successfully identify several influential malicious accounts and prominent communities with apparent behavior differences from others. [ABSTRACT FROM AUTHOR]

This paper proposes the simulated and experimental results of a universal filter using the voltage differencing differential difference amplifier (VDDDA). Unlike the previous complementary metal oxide semiconductor (CMOS) structures of VDDDA that is present in the literature, the present one is compact and simple, owing to the employment of the multiple-input metal oxide semiconductor (MOS) transistor technique. The presented filter employs two VDDDAs, one resistor and two grounded capacitors, and it offers low-pass: LP, band-pass: BP, band-reject: BR, high-pass: HP and all-pass: AP responses with a unity passband voltage gain. The proposed universal voltage mode filter has high input impedances and low output impedance. The natural frequency and bandwidth are orthogonally controlled by using separated transconductance without affecting the passband voltage gain. For a BP filter, the root mean square (RMS) of the equivalent output noise is 46 µV, and the third intermodulation distortion (IMD3) is −49.5 dB for an input signal with a peak-to peak of 600 mV, which results in a dynamic range (DR) of 73.2 dB. The filter was designed and simulated in the Cadence environment using a 0.18-µm CMOS process from Taiwan semiconductor manufacturing company (TSMC). In addition, the experimental results were obtained by using the available commercial components LM13700 and AD830. The simulation results are in agreement with the experimental one that confirmed the advantages of the filter. [ABSTRACT FROM AUTHOR]

C-V2X (Cellular Vehicle-to-Everything) is a state-of-the-art wireless technology used in autonomous driving and intelligent transportation systems (ITS). This technology has extended the coverage and blind-spot detection of autonomous driving vehicles. Economically, C-V2X is much more cost-effective than the traditional sensors that are commonly used by autonomous driving vehicles. This cost-benefit makes it more practical in a large scale deployment. PC5-based C-V2X uses an RF (Radio Frequency) sidelink direct communication for low latency mission-critical vehicle sensor connectivity. Over the C-V2X radio communications, the autonomous driving vehicle's sensor ability can now be largely enhanced to the distances as far as the network covers. In 2020, 5G is commercialized worldwide, and Taiwan is at the forefront. Operators and governments are keen to see its implications in people's daily life brought by its low latency, high reliability, and high throughput. Autonomous driving class L3 (Conditional Automation) or L4 (Highly Automation) are good examples of 5G's advanced applications. In these applications, the mobile networks with URLLC (Ultra-Reliable Low-Latency Communication) are perfectly demonstrated. Therefore, C-V2X evolution and 5G NR (New Radio) deployment coincide and form a new ecosystem. This ecosystem will change how people will drive and how transportation will be managed in the future. In this paper, the following topics are covered. Firstly, the benefits of C-V2X communication technology. Secondly, the standards of C-V2X and C-V2X applications for automotive road safety system which includes V2P/V2I/V2V/V2N, and artificial intelligence in VRU (Vulnerable Road User) detection, object recognition and movement prediction for collision warning and prevention. Thirdly, PC5-based C-V2X deployment status in global, especially in Taiwan. Lastly, current challenges and conclusions of C-V2X development. [ABSTRACT FROM AUTHOR]

In this paper, a piezoelectric wave-energy converter (PWEC), consisting of a buoy, a frequency up-conversion mechanism, and a piezoelectric power-generator component, is developed. The frequency up-conversion mechanism consists of a gear train and geared-linkage mechanism, which converted lower frequencies of wave motion into higher frequencies of mechanical motion. The slider had a six-period displacement compared to the wave motion and was used to excite the piezoelectric power-generation component. Therefore, the operating frequency of the piezoelectric power-generation component was six times the frequency of the wave motion. The developed, flexible piezoelectric composite films of the generator component were used to generate electrical voltage. The piezoelectric film was composed of a copper/nickel foil as the substrate, lead–zirconium–titanium (PZT) material as the piezoelectric layer, and silver material as an upper-electrode layer. The sol-gel process was used to fabricate the PZT layer. The developed PWEC was tested in the wave flume at the Tainan Hydraulics Laboratory, Taiwan (THL). The maximum height and the minimum period were set to 100 mm and 1 s, respectively. The maximum voltage of the measured value was 2.8 V. The root-mean-square (RMS) voltage was 824 mV, which was measured through connection to an external 495 kΩ resistive load. The average electric power was 1.37 μW. [ABSTRACT FROM AUTHOR]

Bed is often the personal care unit in hospitals, nursing homes, and individuals' homes. Rich care-related information can be derived from the sensing data from bed. Patient fall is a significant issue in hospitals, many of which are related to getting in and/or out of bed. To prevent bed falls, a motion-sensing mattress was developed for bed-exit detection. A machine learning algorithm deployed on the chip in the control box of the mattress identified the in-bed postures based on the on/off pressure pattern of 30 sensing areas to capture the users' bed-exit intention. This study aimed to explore how sleep-related data derived from the on/off status of 30 sensing areas of this motion-sensing mattress can be used for multiple layers of precision care information, including wellbeing status on the dashboard and big data analysis for living pattern clustering. This study describes how multiple layers of personalized care-related information are further derived from the motion-sensing mattress, including real-time in-bed/off-bed status, daily records, sleep quality, prolonged pressure areas, and long-term living patterns. Twenty-four mattresses and the smart mattress care system (SMCS) were installed in a dementia nursing home in Taiwan for a field trial. Residents' on-bed/off-bed data were collected for 12 weeks from August to October 2021. The SMCS was developed to display care-related information via an integrated dashboard as well as sending reminders to caregivers when detecting events such as bed exits and changes in patients' sleep and living patterns. The ultimate goal is to support caregivers with precision care, reduce their care burden, and increase the quality of care. At the end of the field trial, we interviewed four caregivers for their subjective opinions about whether and how the SMCS helped their work. The caregivers' main responses included that the SMCS helped caregivers notice the abnormal situation for people with dementia, communicate with family members of the residents, confirm medication adjustments, and whether the standard care procedure was appropriately conducted. Future studies are suggested to focus on integrated care strategy recommendations based on users' personalized sleep-related data. [ABSTRACT FROM AUTHOR]

An accurate seismic response simulation of civil structures requires accounting for the nonlinear soil response behavior. This, in turn, requires understanding the nonlinear material behavior of in situ soils under earthquake excitations. System identification methods applied to data recorded during earthquakes provide an opportunity to identify the nonlinear material properties of in situ soils. In this study, we use a Bayesian inference framework for nonlinear model updating to estimate the nonlinear soil properties from recorded downhole array data. For this purpose, a one-dimensional finite element model of the geotechnical site with nonlinear soil material constitutive model is updated to estimate the parameters of the soil model as well as the input excitations, including incident, bedrock, or within motions. The seismic inversion method is first verified by using several synthetic case studies. It is then validated by using measurements from a centrifuge test and with data recorded at the Lotung experimental site in Taiwan. The site inversion method is then applied to the Benicia–Martinez geotechnical array in California, using the seismic data recorded during the 2014 South Napa earthquake. The results show the promising application of the proposed seismic inversion approach using Bayesian model updating to identify the nonlinear material parameters of in situ soil by using recorded downhole array data. [ABSTRACT FROM AUTHOR]

From the accident news, it is found that the occurrences of traffic accidents every year and the numbers of deaths and injuries have raised continually and have become a specific issue concerned in society in Taiwan. More seriously, the number of traffic accidents is positively increased with the increasing motorized vehicles. Thus, to reduce the incidence of traffic accidents through by some advanced real-time technologies is an important and interesting work. However, some serious problems against traffic safety are facing, such as the low-quality video saved by a camera, low efficiency facilities supported, inefficient management of surveillance resources, and low definition resolution for cameras, which is resulted in a dilemma problem caused from providing evidence-based images to a local authority either for criteria for judgment or basis for improvement. As a big effort to deal with the above defects for constructing a smart city, this paper makes a main purpose to develop an advanced system of intelligent cloud-based transportation vehicle surveillance (called ICTVSS) for license plate identification. This existing identification algorithm was studied and developed from a combination of improved differential algorithm and improved active contour algorithm. Given such a combination, a novel algorithm of dynamic license identification for smart monitoring was fully realized for constructing a well-defined smart city. The experimental results showed good performance and experienced that the proposed algorithm performed well in locating multi-license plate and differential methods, removing image noise of license plate, and processing constant-inconstant light source from complex environment cases, and guaranteed effective license plate identification from the benefit of high resolutions of digital cameras. [ABSTRACT FROM AUTHOR]

There is a critical need for detection of administrative regions through network map pictures in map censorship tasks, which can be implemented by target detection technology. However, on map images there tend to be numerous administrative regions overlaying map annotations and symbols, thus making it difficult to accurately detect each region. Using a RetinaNet-based target detection model integrating ResNet50 and a feature pyramid network (FPN), this study built a multi-target model and a single-target cascading model from three single-target models by taking Taiwan, Tibet, and the Chinese mainland as target examples. Two models were evaluated both in classification and localization accuracy to investigate their administrative region detection performance. The results show that the single-target cascading model was able to detect more administrative regions, with a higher f1_score of 0.86 and mAP of 0.85 compared to the multi-target model (0.56 and 0.52, respectively). Furthermore, location box size distribution from the single-target cascading model looks more similar to that of manually annotated box sizes, which signifies that the proposed cascading model is superior to the multi-target model. This study is promising in providing support for computer map reading and intelligent map censorship. [ABSTRACT FROM AUTHOR]

Air pollution is a global problem and can be perceived as a modern-day curse. One way of dealing with it is by finding economical ways to monitor and forecast air quality. Accurately monitoring and forecasting fine particulate matter (PM2.5) concentrations is a challenging prediction task but Internet of Things (IoT) can help in developing economical and agile ways to design such systems. In this paper, we use a historical data-based approach to perform PM2.5 forecasting. A forecasting method is developed which uses exponential smoothing with drift. Experiments and evaluation were performed using the real-time PM2.5 data obtained from large scale deployment of IoT devices in Taichung region in Taiwan. We used the data from 132 monitoring stations to evaluate our model's performance. A comparison of prediction accuracy and computation time between the proposed model and three widely used forecasting models was done. The results suggest that our method can perform PM2.5 forecast for 132 monitoring stations with error as low as 0.16 μ g/ m 3 and also with an acceptable computation time of 30 s. Further evaluation was done by forecasting PM2.5 for next 3 h. The results show that 90 % of the monitoring stations have error under 1.5 μ g/ m 3 which is significantly low. [ABSTRACT FROM AUTHOR]

Incidents of injuries caused by tiles falling from building exterior walls are frequently reported in Taiwan. Humidity is an influential factor in tile deterioration but it is more difficult to measure the humidity inside a building structure than the humidity in an indoor environment. Therefore, a separable microsensor was developed in this study to measure the humidity of the cement mortar layer with a thickness of 1.5-2 cm inside the external wall of a building. 3D printing technology is used to produce an encapsulation box that can protect the sensor from damage caused by the concrete and cement mortar. The sensor is proven in this study to be capable of measuring temperature and humidity simultaneously and the measurement results are then used to analyze the influence of humidity on external wall tile deterioration. [ABSTRACT FROM AUTHOR]

Programming is a skill that requires high levels of logical thinking and problem-solving abilities. According to the Curriculum Guidelines for the 12-Year Basic Education currently implemented in Taiwan, programming has been included in the mandatory courses of middle and high schools. Nevertheless, the guidelines simply recommend that elementary schools conduct fundamental instructions in related fields during alternative learning periods. This may result in the problem of a rough transition in programming learning for middle school freshmen. To alleviate this problem, this study proposes an augmented reality (AR) logic programming teaching system that combines AR technologies and game-based teaching material designs on the basis of the fundamental concepts for seventh-grade structured programming. This system can serve as an articulation curriculum for logic programming in primary education. Thus, students are able to develop basic programming logic concepts through AR technologies by performing simple command programming. This study conducted an experiment using the factor-based quasi-experimental research design and questionnaire survey method, with 42 fifth and sixth graders enrolled as the experimental subjects. The statistical analysis showed the following results: In terms of learning effectiveness, both AR-based and traditional learning groups displayed a significant performance. However, of the two groups, the former achieved more significant effectiveness in the posttest results. Regarding learning motivation, according to the evaluation results of the Attention, Relevance, Confidence, and Satisfaction (ARCS) motivation model, the AR-based learning group manifested significantly higher levels of learning motivation than the traditional learning group, with particularly significant differences observed in the dimension of Attention. Therefore, the experimental results validate that the proposed AR-based logic programming teaching system has significant positive effects on enhancing students' learning effectiveness and motivation. [ABSTRACT FROM AUTHOR]

The technique of active ionospheric sounding by ionosondes requires sophisticated methods for the recovery of experimental data on ionograms. In this work, we applied an advanced algorithm of deep learning for the identification and classification of signals from different ionospheric layers. We collected a dataset of 6131 manually labeled ionograms acquired from low-latitude ionosondes in Taiwan. In the ionograms, we distinguished 11 different classes of the signals according to their ionospheric layers. We developed an artificial neural network, FC-DenseNet24, based on the FC-DenseNet convolutional neural network. We also developed a double-filtering algorithm to reduce incorrectly classified signals. That made it possible to successfully recover the sporadic E layer and the F2 layer from highly noise-contaminated ionograms whose mean signal-to-noise ratio was low, SNR = 1.43. The Intersection over Union (IoU) of the recovery of these two signal classes was greater than 0.6, which was higher than the previous models reported. We also identified three factors that can lower the recovery accuracy: (1) smaller statistics of samples; (2) mixing and overlapping of different signals; (3) the compact shape of signals. [ABSTRACT FROM AUTHOR]

Taiwan is an island, and its economic activities are primarily dependent on maritime transport and international trade. However, Taiwan is also located in the region of typhoon development in the Northwestern Pacific Basin. Thus, it frequently receives strong winds and large waves brought by typhoons, which pose a considerable threat to port operations. To determine the real-time status of winds and waves brought by typhoons near the coasts of major ports in Taiwan, this study developed models for predicting the wind speed and wave height near the coasts of ports during typhoon periods. The forecasting horizons range from 1 to 6 h. In this study, the gated recurrent unit (GRU) neural networks and convolutional neural networks (CNNs) were combined and adopted to formulate the typhoon-induced wind and wave height prediction models. This work designed two wind speed prediction models (WIND-1 and WIND-2) and four wave height prediction models (WAVE-1 to WAVE-4), which are based on the WIND-1 and WIND-2 model outcomes. The Longdong and Liuqiu Buoys were the experiment locations. The observatory data from the ground stations and buoys, as well as radar reflectivity images, were adopted. The results indicated that, first, WIND-2 has a superior wind speed prediction performance to WIND-1, where WIND-2 can be used to identify the temporal and spatial changes in wind speeds using ground station data and reflectivity images. Second, WAVE-4 has the optimal wave height prediction performance, followed by WAVE-3, WAVE-2, and WAVE-1. The results of WAVE-4 revealed using the designed models with in-situ and reflectivity data directly yielded optimal predictions of the wind-based wave heights. Overall, the results indicated that the presented combination models were able to extract the spatial image features using multiple convolutional and pooling layers and provide useful information from time-series data using the GRU memory cell units. Overall, the presented models could exhibit promising results. [ABSTRACT FROM AUTHOR]

Facility management platforms are widely used in the facility maintenance phase of the building life cycle. However, a large amount of complex building information affects facility managers' efficiency and user experience in retrieving specific information on the facility management platform. Therefore, this research aims to develop a conversation-based method to improve the efficiency and user experience of facility management information delivery. The proposed method contains four major modules: decision mechanism, equipment dataset, intent analysis, and knowledge base. A chatbot prototype was developed based on the proposed method. The prototype was then validated through a feasibility test and field test at the Shulin Arts Comprehensive Administration Building in Taiwan. The results showed that the proposed method changes the traditional information delivery between users and the facility management platform. By integrating natural language processing (NLP), building information modelling (BIM), and ontological techniques, the proposed method can increase the efficiency of FM information retrieval. [ABSTRACT FROM AUTHOR]

Smartwatches are being increasingly used in research to monitor heart rate (HR). However, it is debatable whether the data from smartwatches are of high enough quality to be applied in assessing the health impacts of air pollutants. The objective of this study was to assess whether smartwatches are useful complements to certified medical devices for assessing PM2.5 health impacts. Smartwatches and medical devices were used to measure HR for 7 and 2 days consecutively, respectively, for 49 subjects in 2020 in Taiwan. Their associations with PM2.5 from low-cost sensing devices were assessed. Good correlations in HR were found between smartwatches and certified medical devices (rs > 0.6, except for exercise, commuting, and worshipping). The health damage coefficients obtained from smartwatches (0.282% increase per 10 μg/m3 increase in PM2.5) showed the same direction, with a difference of only 8.74% in magnitude compared to those obtained from certified medical devices. Additionally, with large sample sizes, the health impacts during high-intensity activities were assessed. Our work demonstrates that smartwatches are useful complements to certified medical devices in PM2.5 health assessment, which can be replicated in developing countries. [ABSTRACT FROM AUTHOR]

The chicken industry, in which broiler chickens are bred, is the largest poultry industry in Taiwan. In a traditional poultry house, breeders must usually observe the health of the broilers in person on the basis of their breeding experience at regular times every day. When a breeder finds unhealthy broilers, they are removed manually from the poultry house to prevent viruses from spreading in the poultry house. Therefore, in this study, we designed and constructed a novel small removal system for dead chickens for Taiwanese poultry houses. In the mechanical design, this system mainly contains walking, removal, and storage parts. It comprises robotic arms with a fixed end and sweep-in devices for sweeping dead chickens, a conveyor belt for transporting chickens, a storage cache for storing chickens, and a tracked vehicle. The designed system has dimensions of approximately 1.038 × 0.36 × 0.5 m3, and two dead chickens can be removed in a single operation. The walking speed of the chicken removal system is 3.3 cm/s. In order to enhance the automation and artificial intelligence in the poultry industry, the identification system was used in a novel small removal system. The conditions of the chickens in a poultry house can be monitored remotely by using a camera, and dead chickens can be identified through deep learning based on the YOLO v4 algorithm. The precision of the designed system reached 95.24% in this study, and dead chickens were successfully moved to the storage cache. Finally, the designed system can reduce the contact between humans and poultry to effectively improve the overall biological safety. [ABSTRACT FROM AUTHOR]

This study proposes a shape approximation approach to portray the regions of interest (ROI) from medical imaging data. An effective algorithm to achieve an optimal approximation is proposed based on the framework of Particle Swarm Optimization. The convergence of the proposed algorithm is derived under mild assumptions on the selected family of shape equations. The issue of detecting Parkinson's disease (PD) based on the Tc-99m TRODAT-1 brain SPECT/CT images of 634 subjects, with 305 female and an average age of 68.3 years old from Kaohsiung Chang Gung Memorial Hospital, Taiwan, is employed to demonstrate the proposed procedure by fitting optimal ellipse and cashew-shaped equations in the 2D and 3D spaces, respectively. According to the visual interpretation of 3 experienced board-certified nuclear medicine physicians, 256 subjects are determined to be abnormal, 77 subjects are potentially abnormal, 174 are normal, and 127 are nearly normal. The coefficients of the ellipse and cashew-shaped equations, together with some well-known features of PD existing in the literature, are employed to learn PD classifiers under various machine learning approaches. A repeated hold-out with 100 rounds of 5-fold cross-validation and stratified sampling scheme is adopted to investigate the classification performances of different machine learning methods and different sets of features. The empirical results reveal that our method obtains 0.88 ± 0.04 classification accuracy, 0.87 ± 0.06 sensitivity, and 0.88 ± 0.08 specificity for test data when including the coefficients of the ellipse and cashew-shaped equations. Our findings indicate that more constructive and useful features can be extracted from proper mathematical representations of the 2D and 3D shapes for a specific ROI in medical imaging data, which shows their potential for improving the accuracy of automated PD identification. [ABSTRACT FROM AUTHOR]

Zoning conflicts have transformed Old Street fabrics in terms of architectural style and construction material in Lukang, Taiwan. This transformation should be assessed as a contribution to digital cultural sustainability. The objective of this study was to compare the evolved façades resultant from the changes made by the development of architectural history and urban planning. A combination of 3D scan technology and a smartphone augmented reality (AR) app, Augment®, was applied to the situated comparison with direct interaction on-site. The AR application compared 20 façades in the laboratory and 18 façades in four different sites using a flexible interface. The comparisons identified the correlation of evolved façades in real sites in terms of building volumes and components, pedestrian arcades on store fronts, and previous installations. The situated comparisons were facilitated in a field study with real-time adjustments to 3D models and analyses of correlations across details and components. The application of AR was demonstrated to be effective in reinstalling scenes and differentiating diversified compositions of vocabulary in a remote site. [ABSTRACT FROM AUTHOR]

Partially fermented tea such as oolong tea is a popular drink worldwide. Preventing fraud in partially fermented tea has become imperative to protect producers and consumers from possible economic losses. Visible/near-infrared (VIS/NIR) spectroscopy integrated with stepwise multiple linear regression (SMLR) and support vector machine (SVM) methods were used for origin discrimination of partially fermented tea from Vietnam, China, and different production areas in Taiwan using the full visible NIR wavelength range (400–2498 nm). The SMLR and SVM models achieved satisfactory results. Models using data from chemical constituents' specific wavelength ranges exhibited a high correlation with the spectra of teas, and the SMLR analyses improved discrimination of the types and origins when performing SVM analyses. The SVM models' identification accuracies regarding different production areas in Taiwan were effectively enhanced using a combination of the data within specific wavelength ranges of several constituents. The accuracy rates were 100% for the discrimination of types, origins, and production areas of tea in the calibration and prediction sets using the optimal SVM models integrated with the specific wavelength ranges of the constituents in tea. NIR could be an effective tool for rapid, nondestructive, and accurate inspection of types, origins, and production areas of teas. [ABSTRACT FROM AUTHOR]

This research mainly aims at the construction of the novel acceleration pedal, the brake pedal and the steering system by mechanical designs and mechatronics technologies, an approach of which is rarely seen in Taiwan. Three highlights can be addressed: 1. The original steering parts were removed with the fault tolerance design being implemented so that the basic steering function can still remain in case of the function failure of the control system. 2. A larger steering angle of the front wheels in response to a specific rotated angle of the steering wheel is devised when cornering or parking at low speed in interest of drivability, while a smaller one is designed at high speed in favor of driving stability. 3. The operating patterns of the throttle, brake, and steering wheel can be customized in accordance with various driving environments and drivers' requirements using the self-developed software. The implementation of a steer-by-wire system in the remote driving control for a go kart is described in this study. The mechatronic system is designed in order to support the conversion from human driving to autonomous driving for the go kart in the future. The go kart, using machine vision, is wirelessly controlled in the WiFi frequency bands. The steer-by-wire system was initially modeled as a standalone system for one wheel and subsequently developed into its complete form, including front wheel steering components, acceleration components, brake components, a microcontroller, drive circuit and digital to analog converter. The control output section delivers the commands to the subsystem controllers, relays and converters. The remote driving control of the go kart is activated when proper commands are sent by the vehicle control unit (VCU). All simulation and experiment results demonstrated that the control strategies of duel motors and the VCU control were successfully optimized. The feasibility study and performance evaluation of Taiwan's go karts will be conducted as an extension of this study in the near future. [ABSTRACT FROM AUTHOR]

No matter your experience level or budget, there is a great ski goggle waiting to be found.Goggles are an essential part of skiing or snowboarding gear to protect your eyes from harsh environmental elements and injury. In the ski goggles manufacturing industry, defects, especially on the lens surface, are unavoidable. However, defect detection and classification by visual inspection in the manufacturing process is very difficult. To overcome this problem, a novel framework based on machine vision is presented, named as the ski goggles lens defect detection, with five high-resolution cameras and custom-made lighting field to achieve a high-quality ski goggles lens image. Next, the defects on the lens of ski goggles are detected by using parallel projection in opposite directions based on adaptive energy analysis. Before being put into the classification system, the defect images are enhanced by an adaptive method based on the high-order singular value decomposition (HOSVD). Finally, dust and five types of defect images are classified into six types, i.e., dust, spotlight (type 1, type 2, type 3), string, and watermark, by using the developed classification algorithm. The defect detection and classification results of the ski goggles lens are compared to the standard quality of the manufacturer. Experiments using 120 ski goggles lens samples collected from the largest manufacturer in Taiwan are conducted to validate the performance of the proposed framework. The accurate defect detection rate is 100% and the classification accuracy rate is 99.3%, while the total running time is short. The results demonstrate that the proposed method is sound and useful for ski goggles lens inspection in industries. [ABSTRACT FROM AUTHOR]

The in situ scattering and transmissometry laser (LISST-100X), equipped with an acoustic wave and current (AWAC) meter and conductivity, temperature, and depth (CTD) instruments over the seabed in the East China Sea, was used to monitor the variation in suspended particles in the bottom sea layer, including particle size distribution (PSD) and volume concentration. The power law approximation was tested to describe the variability in PSD based on the field data. The results show that the power law was robust in processing continuous data, accompanied with the same optimal reference particle size (~63 μm) and little change in the corresponding exponent (~3.4) in both periods. Suspended particles were divided into three types: macroflocs (>133 μm), microflocs (36–133 μm), and single grains (<36 μm). Particle sizes were coarse during the two seasons, with macroflocs representing more than 60% of all the suspended particles, especially in February, when the particle size spectra were usually open-ended. Results from the harmonic analysis method indicate that tidal-induced resuspension and advection are the major reasons for the diurnal dynamics of sediments. Due to the tidal asymmetry in the region, we only found one mode in volume concentration at the moment of maximum velocity. However, the ratios of macroflocs were bimodal, with maximum floods and ebbs in one tidal cycle in February, when the higher mode at the maximum ebbs may be contributed to by the flocculation of finer particles considering the decreasing ratios of finer particles. Due to the enhanced stratification and the clean barrier built up by the Taiwan Warm Current in the southeast corner, the significant daily variation in suspended particles observed in February weakened in September. The influence of waves was uncertain, although the correlation coefficient between significant wave height and volume concentration was about 64% in February. [ABSTRACT FROM AUTHOR]

The study continues the theoretical derivation from Part 1, and the experiment is carried out at a bus station equipped with six water-cooled chillers. Between 2012 and 2017, historical data collected from temperature and humidity sensors, as well as the energy consumption data, were used to build artificial intelligence (AI) assisted heating ventilation and air conditioning (HVAC) control models. The AI control system, in conjunction with a specifically designed prior information notice (PIN) sensor, was used to improve the prediction accuracy. This data collected between 2012 and 2016 was used for AI training and PIN sensor testing. During the hottest week of 2017 in Taiwan, the PIN sensor was used to conduct temperature and humidity data predictions. A model-based predictive control was developed to obtain air conditioning energy consumption data. The comparative results between the predictive and actual data showed that the temperature and humidity prediction accuracies were between 95.5 and 96.6%, respectively. Additionally, energy savings amounting to 39.8% were achieved compared to the theoretical estimates of 44.6%, a difference of less than 5%. These results show that the experimental model supports the theoretical estimations. In the future, a PIN sensor will be installed in a chiller to further verify the energy savings of the AI assisted HVAC control. [ABSTRACT FROM AUTHOR]

Magnetic resonance imaging (MRI) offers the most detailed brain structure image available today; it can identify tiny lesions or cerebral cortical abnormalities. The primary purpose of the procedure is to confirm whether there is structural variation that causes epilepsy, such as hippocampal sclerotherapy, local cerebral cortical dysplasia, and cavernous hemangioma. Cerebrovascular disease, the second most common factor of death in the world, is also the fourth leading cause of death in Taiwan, with cerebrovascular disease having the highest rate of stroke. Among the most common are large vascular atherosclerotic lesions, small vascular lesions, and cardiac emboli. The purpose of this thesis is to establish a computer-aided diagnosis system based on small blood vessel lesions in MRI images, using the method of Convolutional Neural Network and deep learning to analyze brain vascular occlusion by analyzing brain MRI images. Blocks can help clinicians more quickly determine the probability and severity of stroke in patients. We analyzed MRI data from 50 patients, including 30 patients with stroke, 17 patients with occlusion but no stroke, and 3 patients with dementia. This system mainly helps doctors find out whether there are cerebral small vessel lesions in the brain MRI images, and to output the found results into labeled images. The marked contents include the position coordinates of the small blood vessel blockage, the block range, the area size, and if it may cause a stroke. Finally, all the MRI images of the patient are synthesized, showing a 3D display of the small blood vessels in the brain to assist the doctor in making a diagnosis or to provide accurate lesion location for the patient. [ABSTRACT FROM AUTHOR]