Taiwan Association of Engineering and Technology Innovation: E-Journals
Not a member yet
887 research outputs found
Sort by
A Framework for Crop Disease Detection Using Feature Fusion Method
Get PDFCrop disease detection methods vary from traditional machine learning, which uses Hand-Crafted Features (HCF) to the current deep learning techniques that utilize deep features. In this study, a hybrid framework is designed for crop disease detection using feature fusion. Convolutional Neural Network (CNN) is used for high level features that are fused with HCF. Cepstral coefficients of RGB images are presented as one of the features along with the other popular HCF. The proposed hybrid model is tested on the whole leaf images and also on the image patches which have individual lesions. The experimental results give an enhanced performance with a classification accuracy of 99.93% for the whole leaf images and 99.74% for the images with individual lesions. The proposed model also shows a significant improvement in comparison to the state-of-art techniques. The improved results show the prominence of feature fusion and establish cepstral coefficients as a pertinent feature for crop disease detection
Defining Effective Gain for Evaluation of Orbital Angular Momentum Links
Get PDFIn this paper, a communication link based on circular phased array antennas generating orbital angular momentum (OAM) beams at radio frequency is investigated. The presence of a null in the radiation pattern of OAM antennas is the main drawback of them. This problem makes it difficult to establish a telecommunication link using OAM systems and calculate the link budget for such a system. To solve this problem, we have defined two new gain parameters by using Friis Transmission Equation. The new formulas can help to calculate the effective gain of OAM antennas. Also, we have defined the effective OAM gain in detail for the first time in order to evaluate the performance of the OAM links. By using the proposed formulas, a capable and secure link based on the orthogonality of OAM beams can be designed
Tea Verification Using Triplet Loss Convolutional Network
Get PDFTo solve tea image classification problems, this study focuses on triplet loss convolutional neural network to classify six high-mountain oolong tea classes. In the experiment, instead of using traditional deep learning training approach for local feature of tea images, an innovative image verification approach is proposed to learn the global feature of tea images by integrating the distributed tea leaves’ features of all tea sub-images and using a majority voting mechanism to do classification. The results show that the proposed approach can work for small sample size dataset and have higher accuracy than normal transfer learning approach. The average accuracy of the proposed approach achieves 99.54%
Vehicle Path Planning with Multicloud Computation Services
Get PDFWith the development of artificial intelligence, public cloud service platforms have begun to provide common pretrained object recognition models for public use. In this study, a dynamic vehicle path-planning system is developed, which uses several general pretrained cloud models to detect obstacles and calculate the navigation area. The Euclidean distance and the inequality based on the detected marker box data are used for vehicle path planning. Experimental results show that the proposed method can effectively identify the driving area and plan a safe route. The proposed method integrates the bounding box information provided by multiple cloud object detection services to detect navigable areas and plan routes. The time required for cloud-based obstacle identification is 2 s per frame, and the time required for feasible area detection and action planning is 0.001 s per frame. In the experiments, the robot that uses the proposed navigation method can plan routes successfully
Adaptive Vibrarthographic Signal Denoising via Ant Colony Optimization Using Dynamic Denoising Filter Parameters
Get PDFThis study proposes an ant colony optimization (ACO) denoising method with dynamic filter parameters. The proposed method is developed based on ensemble empirical mode decomposition (EEMD), and aims to improve the quality of vibrarthographic (VAG) signals. It mixes the original VAG signals with different white noise amplitudes, and adopts a hybrid technology that combines EEMD with a Savitzky-Golay (SG) filter containing the dynamic parameters optimized by ACO. The results show that the proposed method provides a higher peak signal-to-noise ratio (PSNR) and a smaller root-mean-square difference than the regular methods. The SNR improvement for the VAG signals of normal knees can reach 13 dB while maintaining the original signal structure, and the SNR improvement for the VAG signals of abnormal knees can reach 20 dB. The method proposed in this study can improve the quality of nonstationary VAG signals
5G F-OFDM Waveform Based Software-Defined Radio Technology
Get PDFFiltered-orthogonal frequency division multiplexing (F-OFDM) is a quasi-orthogonal waveform candidate for the applications of the fifth generation (5G) communication system. In this study, an F-OFDM waveform with unequal sub-band sizes is proposed to improve the spectrum efficiency (SE) of the 5G system. The proposed waveform is modeled with the Blackman window-sinc filter and is developed based on the software-defined radio (SDR) technology for practical implementation. The result shows that the F-OFDM performance of the simulation and hardware implementation is approximately the same. The SE using the proposed F-OFDM waveform is 6% and 5.8% higher than the SE using the conventional OFDM waveform under the simulation in the LabVIEW NXG simulator and under the practical use in the universal software radio peripheral (USRP) platform, respectively
Implementation of 20 nm Graphene Channel Field Effect Transistors Using Silvaco TCAD Tool to Improve Short Channel Effects over Conventional MOSFETs
Get PDFIn recent years, demands for high speed and low power circuits have been raised. As conventional metal oxide semiconductor field effect transistors (MOSFETs) are unable to satisfy the demands due to short channel effects, the purpose of the study is to design an alternative of MOSFETs. Graphene FETs are one of the alternatives of MOSFETs due to the excellent properties of graphene material. In this work, a user-defined graphene material is defined, and a graphene channel FET is implemented using the Silvaco technology computer-aided design (TCAD) tool at 100 nm and scaled to 20 nm channel length. A silicon channel MOSFET is also implemented to compare the performance. The results show the improvement in subthreshold slope (SS) = 114 mV/dec, ION/IOFF ratio = 14379, and drain induced barrier lowering (DIBL) = 123 mV/V. It is concluded that graphene FETs are suitable candidates for low power applications
Tunable Lossy and Lossless Grounded Inductors Using Minimum Active and Passive Components
Get PDFIn this contribution, nine new Grounded Inductance Simulators (GISs) using a single Multiple-Output Current Controlled Current Conveyor Transconductance Amplifier (MO-CCCCTA) and one grounded capacitor are proposed. Among them, two are lossless types and seven are lossy types. The use of a single grounded capacitor makes the circuits suitable for fabrication. All the proposed circuits are electronically tunable through the bias currents of MO-CCCCTA. Furthermore, no component matching conditions are needed for realizing them. The designed circuits are verified through PSPICE simulator with ± 0.9 V power supply. The simulation results show that for all the proposed circuits: maximum operating frequencies are about 12 MHz, power dissipation is less than 0.784 mW, Total Harmonic Distortions (THDs) are under 8.09%, and maximum output voltage noise at 1 MHz frequency is 14.094 nV/√Hz. To exhibit the workability of the proposed circuits, they are used to design band-pass, low-pass filter, parallel RLC resonator, and parasitic inductance cancelator
Fish Species Detection Application (FiSDA) in Leyte Gulf Using Convolutional Neural Network
Get PDFThis study presents an application that employs a machine-learning algorithm to identify fish species found in Leyte Gulf. It aims to help students and marine scientists with their identification and data collection. The application supports 467 fish species in which 6,918 fish images are used for training, validating, and testing the generated model. The model is trained for a total of 4,000 epochs. Using convolutional neural network (CNN) algorithm, the best model during training is observed at epoch 3,661 with an accuracy rate of 96.49% and a loss value of 0.1359. It obtains 82.81% with a loss value of 1.868 during validation and 80.58% precision during testing. The result shows that the model performs well in predicting Malatindok and Sapsap species, after obtaining the highest precision of 100%. However, Hangit is sometimes misclassified by the model after attaining 55% accuracy rate from the testing results because of its feature similarity to other species
Driving Assistance System with Lane Change Detection
Get PDFIn this study, a simple technology for a self-driving system called “driver assistance system” is developed based on embedded image identification. The system consists of a camera, a Raspberry Pi board, and OpenCV. The camera is used to capture lane images, and the image noise is overcome through color space conversion, grayscale, Otsu thresholding, binarization, erosion, and dilation. Subsequently, two horizontal lines parallel to the X-axis with a fixed range and interval are used to detect left and right lane lines. The intersection points between the left and right lane lines and the two horizontal lines can be obtained, and can be used to calculate the slopes of the left and right lanes. Finally, the slope change of the left and right lanes and the offset of the lane intersection are determined to detect the deviation. When the angle of lanes changes drastically, the driver receives a deviation warning. The results of this study suggest that the proposed algorithm is 1.96 times faster than the conventional algorithm