We propose a semantic-guided framework (FashionOn+) that generates image-based virtual try-on results with arbitrary poses. FashionOn+ contains three stages: (I) conducts the semantic segmentation to have the prior knowledge of body parts for rendering the corresponding texture in stage (II). (III) refines two salient regions, i.e., faceclothes, to generate high-quality results. With the novel architecture, we win first place in the Multi-pose Virtual Try-on Challenge in CVPR, 2020. Further, we tackle the low-resolution limitation (256x192)achieve high-resolution results (640x480).

We propose a novel AI-based few-shot self-supervised learning method for automatic optical inspection image quality assessmentcomponent detection based on only few training images. Our method iteratively learns the feature representations of the components by using self-similarity of these components. With the large number of self-learned representations, the appearance variations of each component are then effectively learned in the AI model for component detectionmeasurement. The computation complexity of our method is significantly lower than that of deep learning methods.

The Artificial Intelligence Underwater MonitoringManagement system is a system that integrates aquacultureinformation technology, which can collect image data of shrimp 24/7 by using IoT devices, then next enhance the images that collected under different light, water qualityturbidity. The system will recordanalyze the number, size, healtheating status of white shrimp, checking the amount of leftovers feed on the feeding tray, then manage the feeding strategycontrol water quality based on the above analysis.

Swelling at the extremities of the extremities is quite common in KD,it is one of the five major diagnostic criteria for KD. According to the results of our research, about 90 of the children will have this symptomis the most important symptoms of KD suggested by American Heart Association. Our research also found that albumin decreased,the degree of reduction is obviously positively related to the severity of the disease, so we have developed a theory that can detect water, hemethe optical method of oxygen. Provides a non-intrusive method for quick differentiation.

This technique combines deep learningdigital twin technologies to connect the designconstruction processto achieve construction-site rebar inspection, issue tracking, verificationmaintenance. This core component of this method is based on the image collection of rebar frames, digital twin creation, cooperative machine learning,the rebar inspection module development. The BIM model was automatically matchedcompared with the 4D digital twin with features of point cloudtime sequence.

The self-powered sensing system realizes self-powered wireless wearable pressure sensing modules, self-powered intelligent environment sensing robots,self-powered smart window, which can carry out self-powered sensing signals combined with the transmission of the Internet of Things. Functions such as medical monitoring aidsautomated environmental quality sensing. In the future, this system opens a window towards automatic monitoring of environmental analysispersonal smart healthcare.

1. CNN model architecture optimization techniques: Propose an Agilev3-L architecture, with a performance index of mAP@50:95FPS:FP16 up to 171.79 2. GOP-mode acceleration scheme for real time inference: employ GOP-modeTRACKING algorithms to enhance the processing rate. 3. Propose a rapid data labeling system HiTag.

NCREE has originally developed 5D digital space—on the basis of 3D city modelsconnections of different kinds of sensors around the world— is an online to offline virtual space with a combination of rising 5G technology advantages. Collectingreorganizing various 3D cartographic data with Building Information Modeling (BIM), satellite imagery, UAV 3D modeling, LiDar point cloud data, etc., can increase the diversity of the buildinglandscape. 5D+5G smart city platform would accelerate the 5D smart city to become the digital twin of a real city.

This project uses AIoT technology to face the occurrence issue of incorrect preｓｃｒｉｐｔion drugs of the drug dispensing phase in hospital dispensing Room. The proposed learning-object-detection based deep technology AI mechanism can automatically identifyverify the correctness of the drugs in the drug bag, thereby significantly improving the accuracy of drug dispensing in Hospital.

The technology of 「Guard Patch」includes three parts. 1. Wearable devices, 2. Intelligent wireless monitoring system. 3. Cloud server with AI to avoid the happen of cardiovascular diseases (CVD). The products of 「Guard Patch」also includes three parts: 1. Medical Market: (1) YuGuard, (2) YuSound, (3) YuRine. 2. Wearable Market: (1) YuCloth, (2) YuPet, (3) YuBelt. 3.Educative Market: TriAnswer. The mentioned seven products developed by Yutech can be employed in medical, wearable, educative markets to satisfy the different demands of customersachieve the dream of 「Bioelectronics Garden」

The optical detection device is fabricated based on the nitrogen dioxide absorbs the wavelength of light from 400nm to 550nm. The detection range, resolution and sensitivity are related to design of light path. The light signals can be received by the photodetector convert to concentrations of nitrogen dioxide which transfer to computer or mobile immediately.；The all-solid sulfur dioxide gas sensor developed by SSID team. The key technology is the special functionality of solid electrolyte and electrode materials, and then the layer material sequentially patterned design and planning. It can be used in high temperature and highly corrosive exhaust gas environment to achieve real-time monitoring of pollution emissions.；The 6'~8'MEMS technology service platform of TSRI was used to develop

This technology is a new microarray chip sensing technology that integrates "organic light-emitting diodes", "liquid crystal chemical sensors" and "mobile communication system" into "multifunctional liquid crystal smart cloud sensors". The obtained signals are uploaded to the cloud for multi-dimensional analysis to achieve multiplex detection in one single sample. This sensor device is lightweight, low-cost, simple to operate, and the target-of-interest can be custom-made according to user needs.

With the astronomically growing trade flows, customs administrations need effective and explainable methods to detect suspicious transactions. This project presents a novel artificial intelligence-based model named DATE that ranks trade flows in the order of fraud risk and to maximize customs revenue. We confirm the superiority of DATE over state-of-the-art AI models, with a remarkable precision of 92.7% on illegal cases and a recall of 49.3% on revenue after inspecting only 1% of all trade flows. Predictions of DATE are also interpretable from the attention mechanism. We are deploying DATE in Nigeria and Malawi Customs Services, in collaboration with the World Customs Organization (WCO). DATE has been published in ACM KDD 2020, which is an AI top conference.

This technology uses in-situ optical photointerrupt sensorin-situ acoustic ultrasonic transducer to measure the growth of fouling layer thickness during membrane water treatment. Being the world’s first to combine real-time filter speed attenuation dataphysical mode of clogging, the clogging of film can be instantly analyzed in a smart way which in turn effectively extends its lifetime. T

Due to the fast development of 5G networks, it is critical to identify users service types to allocate resources intelligently. Our technology focuses on users mobility type identification by extracting practical features from users cellular information. We proposed a system architecturehave collected 700-hour data with 150 GB. By using our dataother datasets in the world, we show that our technology can achieve 95 accuracy,reduce 16 energy consumption compared to traditional methods.

The Air/Ground cooperation for optimal rice harvesting model is established to provide a visual harvesting decision service on a cloud platform. Drones and mobile devices are employing to estimate grain moisture and forecast the variation of harvest moisture content (HMC) in the coming days by huge amounts of imagery data, deep learning algorithms, and weather forecasts. This model can benefit in several aspects, such as setting an accurate and comprehensive optimal harvest schedule, reducing the cost of agricultural apparatus and barn ovens, ensuring the rice quality, and maximizing farmers' benefits. The potential value of the model practice could be more than a billion in Taiwan.

The system consists of a racket sensor, a smartphone App, and the IoT shuttlecock serving machines. Data of the 3-axis inertial sensor will be sent to the smartphone via Bluetooth for analysis. The App can recognize 7 stroke activities, including clear, cut, drive, lob, rush, net play, and smash using machine learning algorithms, while the speed, force, and stroke times are recorded at the same time. The App also enables the user to set parameters of the serving machines for assigning shot placements and serving schedules.

"It is expected that human will work with robots in the coming years. However, it is still unknown how both can collaborate together. Meanwhile, it is still very time-consuming to deploy intelligent robots in production lines. For this work, we investigate deep comprehension for 3D (point cloud)text (voice) signals to enable novel human-robotic object referring for robotic arms. For the task

Six synchronized high-speed cameras were installed in the baseball stadium in NTUS. We established the calibration equation of the multi-camera systemestablished an electronic strike zone system, KarmaZone, that can automatically coordinates when the pitch crossed the home plate. From the slow motion videos recorded, the markerless 3D human posture estimation has been developed with the deep

”AI2 Robot City” is a game-based learning kit for primary and secondary school students, which combines AI image-recognition teaching tool of MIT App Inventor and the computational thinking board game named as "Robot City." Through this learning kit, users will learn to make smart cars, create image-recognition models, write and perform mobile application. Users will learn to write a program to recognize the personal cards in the board game, and furthermore to control the smart cars with blue-tooth and compete in the computational thinking board game.

This system help doctors to distinguish age-related macular degeneration (ARMD)polypoidal choroidal vasculopathy (PCV) by color fundus photography. We have collected large image dataset from Taipei Veterans General Hospitalother hospitals to establishvalidate CNN model, which had been tested by external image datasetmodified to improve the accuracy of the AI model.

AI model is consisted of detection, data pre-processingclassification ensemble model. We applied YOLO 3 to detect vertebral body from radiography,used data pre-processing to reduce image noisyenhance contrast, integrated ResNet34, DenseNet121, DenseNet201 into AI mode. Our result reveals the ensemble model has better performance in accuracy, sensitivityspecificity.

A transfer learning-assisted glaucoma detection system was build first with training images from the fundus images database of Taipei Veterans General Hospital,then fine-tuned to improve its applicability on fundus images from different datasets. This system aims to assist clinic-based glaucoma screening to increase the diagnostic rate of glaucoma by using existing healthcare facility.

TWCC provides services with up to date container technology. Through optimized AI software, developers can quickly deploy, which increases efficiency by 40 compared with the past. Also, it supports high-speed parallel computing. A further increase of 30 compared with current cloud services. In addition, TWCC will integrate AI programstools from different fields to become the largest model m

1. CDX is to provide an environment for long periods of time in coordination with school courses, including the simulation of a real company＇s network environment with randomly selected vulnerabilities in cyber defense contests. 2. HoneynetMalware Knowledge Base simulates systemapplication vulnerabilitiesthen monitors the cyber-attacksthe corresponding traffic flow.

The CPS-OFDM technology by NTU enjoys the best spectral efficiencydesign flexibility among the 5G candidate new waveform designs. CPS-OFDM is integratedjointly optimized with the high performance SCMA by Univ of Surrey to achieve high communication efficiency, high reliability, low latency, for boosting the productivity of the factories of the future (FoF).

This technology integrates 1000+ times of UAV imaging experiences with labeled rice lodging images for training. A rice lodging recognition model using deep learning reaches 90 accuracy. The recognition model can be deployed in a microcomputer mounted on UAVs to implement edge computing. While taking aerial images, the inference can be completedreveal lodging areadamage level in-time.

This project integrated with industrial foresight technologies, including UAV, artificial intelligenceimage recognition, to collect real-time images, apply algorithm in evaluation, link the technology of IOT (Internet of Things)environment sensing,use unmanned vehicle to conduct controlling work.

Our system uses wireless insole in the shoes to measure foot plantar pressure of athletes.

This technology multiple research to integrates the developed production system of the "smart integrated production system". Using AI to identify pestsdiseases,identify the high temperature resistant varieties,developed Root temperature control system,using IOT to establish Big data. An adopted robot will be used to assist sprayingharvesting in the future.