The urban traffic control optimization platform technology is developed by combining vehicle tracking and identification technology with traffic control factor simulation technology, the system is equipped with vehicle counting and vehicle type classification, regional intersection traffic simulation, light optimization and other functions. Through the combination of traffic information and deep learning analysis to help optimize traffic signals at each intersection, it promotes smoother driving for the public.

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.

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.

This research team develops unmanned 3D intelligent marine aquaculture technology, using omni AIoT aquaculture technology, especially strengthening the integration of gyroplane, ROV, unmanned shipcentral control room,using long-distance intelligent control to make employees shift from traditional field work to basecentral control room. Due to the huge cross domain system of omni AIoT, it modularizes it, At the same time, edge computing technology is developed to facilitate commercial promotion.

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.

As we continue to face the rapid increase in confirmed Coronavirus cases around the world, we created an AI-based pneumonia detection platform for COVID-19. The system is able to automatically detect high-risk patients with pneumonia that will then send information to doctors. With that information, the doctors are then able to make follow-up decisions and provide a treatment plan after the diagnosis. In specific, doctors from the Department of Medical Imaging provided us thousands of positive and negative chest x-rays for pneumonia as a training set. Our system has already been tested with and adopted by doctors at the NCKU Hospital. The system achieved 95% accuracy to detect the pneumonia symptom, based on 1400 test images.

"Coach Box" ideatechnology are derived from "CoachAI：金準羽球" project of the Ministry of ScienceTechnology "精準運科". The integrated technologies of "Coach Box" including computer vision, deep learning, machine learning, IoT, wearable applications, automation equipment,big data. The expected outcomes of "Coach Box" to show player performancegive expert advice through activities analysis of vision data. Simultaneously, it can be used as a medium for productive communication between coachesstudents.

With the rise in computing power, deep-learning based computer-aided diagnosis systems have gained interest in the research community. Our system process the images to assist doctors to determine whether the patients have nodules in lungs. Meanwhile, we utilized the Feature Pyramid Network to extend the receptive field on the convolutional kernel, which improved the performance on the nodule detection with various locations in CXR. The semi-supervised learning mechanism also achieves the way of soft-annotation to reduce human effort in medical image annotation.

Induced pluripotent stem cells are a promising new technology in translation medicine,AI technology is an automaticpowerful platform that can help identify cell images with unique characteristics, predictprovide information about cell types, differentiation stages, cell quality,normal Or disease-specific feature information. The combination of this induced pluripotent stem cell differentiation platformAI-based artificial intelligence image analysis technology will have great benefits for future cell product applications.

Our team construct an AIoT smart aquaculture management system. The management system mainly consists of: (1) Image Behavior MonitoringAnalysis Subsystem (2) Smart Feeding Subsystem (3) IOT Subsystem including underwater sensors, ROV,Drone (4) Cloud Subsystem (5) Big Data Analysis Subsystem.

Similar to the critical point tracking technology of human vision, the split attentionspatial grouping enhancement module, combining the multi-path grouping architecturespatial attention technology, can accurately extract important information from the imageimprove the network performance. Moreover, adopting neural architecture search technology to automatically search for the most suitable network architecture based on current moduleshardware devices can balance diagnosis speedhigh accuracy.

The RL model develops a control policy directly from experience to predict statesrewards during a learning procedure. Hence, we designed a medical image environment including US images, different actions,rewards, agent learns in this environment to extract the ALN regionevaluates the status. The performance of our proposed method achieves an accuracy of 83.6, a sensitivity of 88.6,a specificity of 89.0.

This Artificial Intelligence model interprets 12-lead electrocardiograms of adults to diagnose heart failure accuratelyefficiently. It is a fabulous heart failure screening tool that can detect more asymptomatic patients. Our goal is to help patients get early treatmentprevent disease deterioration.

The main purpose of this invention is to provide a training system with the surrounding photographing unit, the display unit,a force platform. The system can immediately capture the motion of the user from various angles,measure the force of the user from both legs, in order to provide the useful biofeedback information to coachesathletes.

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.

Inspired by the human nature that a child can learn by taking an objectthen observing it, we proposed two novel methods: (1) Object Detection by Interactive Perception (ODIP), where a few-shot object detector gradually learns unseen instances by interacting with a well-developed object grasping system, collecting required visual dataannotations in an automatic manner. (2) an efficienteffective few-shot object detection model with novel attention mechanism called Dual-Awareness Attention (DAnA).

The technology is an integrated solution incorporating the retrievable personality for multimodal emotion sensing for spoken services. This framework provides a real-world flexibility that enables the estimation of the target speaker emotion states without manual personalization.

This technology integrates acoustic-optical sensing calculationrecognition information,uses hydrophones to estimate the state of dynamic targetsvisually recognizelocate the dynamic target for vehicle motion control. The fusion acoustic-optical sensing architecture can be used according to the target statecharacteristics. By calculatingadapting to the environmenttarget state characteristics, the exploration tasks will be performed.

"The Crowdsourcing UAV Platform for AI Agricultural Survey(AIAS) provides automatic mosaicing serviceis equipped with four breakthrough technologies, including (1)massive images gridㄒding, (2)parallel computing technology, (3)UAV task specification standardization,(4)UAV task matchmaking. AIAS platform is committed to creating aerial UBER crowdsourcing services in various aspects, such as (1)crop distribution surveys, (2)large-scale lodging surveys, (3)agricultural insurance investigation, (4)surveys on illegal use of farmland,(5)surveys on the subsidy of fallow."

The dialogue system is the main subsystem of the visually impaired navigation system, which provides destinations for the navigation system through multiple dialogues. We use the knowledge graph as the basis for reasoning. In terms of close-range navigation, deep learning technology is used to develop RGB camera detection depth algorithm, indoor semantic cutting algorithm, integrated detection depth estimationindoor semantic cutting in indoor obstacle avoidance, etc. The whole system uses the CellS software design framework to integrate distributed AIoT systems.

The invention focuses on the early fry culture, including algae, live-food organism, water quality control, fine image detection,automated intelligent culture technology. By connecting environmental parameter sensing equipment, fine image observation equipment, automatic feedback/delivering equipment,central intelligent learning/processing platform, an intelligent production system of aquatic fry can be completed,the big data on production parameters can be collected, optimizing breeding strategiesefficiency,increasing the breeding profitability more than 10 times.

In the context of FinTech, we present Financial Graph Attention Networks (FinGAT) to recommend high-profitable stocks in terms of return ratio using time series of stock pricessector info. Our FinGAT can learn the long-short-term price tendency,model the latent interactionsinfluence between stockssectors without any hand-crafted effort. Experiments conducted on Taiwan Stock, S&P 500,NASDAQ stock markets exhibit remarkable accuracy of FinGAT, comparing to state-of-the-arts (by 11, 14,12 performance improvement). The tech is published in IEEE TKDE 2021.

In order to solve the undiscovered problem that caused by the fall ofelders, we develop a deep learning based systemnamed it 「SkyEye」,which includes our own sensor 「AI Falling Image Sensor」, a cloud sever that storespushes information about fall eventsa mobile app that communicates with users.

This system uses the innovative AIoT application to target annoying scalp maintenance problemsdevelops an intelligent scalp detectionmanagement system. The core of the technology is to use deep learning-based object detection models. Tens of thousands of microscopic image data sets that are labeledtrained for scalp symptoms. As a result, the scalp image recognition module is develop, such as dandruff, hair loss, oil, and inflammation, etc. Hence this system can provide scalp detection, and maintenance effectiveness tracking functions lead scalp maintenance services to a new level of intelligent management.

In Taiwan, there are almost 96 million people look for medical treatment due to diabetes. The project successfully used the infrared light measurement system to observe the differences between the bilateral feet,collocating with the Chaos Synchronization to measure the differences between different signals, to decrease the probability of the attendance as the ultimate objective .

"1. Automatically generate Verilog code tools based on the hardware architecture of convolutional neural networks: 4 different hardware architectures (output stationary, weight stationary, Tree architecture, NVDLA) can be generated for the currently more commonly used DNN networks. 2. Visual performance index analysis tool: According to the selected DNN model,the choice of hardware architecture specifications, analyze the performance index."

"For video anomaly detection, we apply pretrained models to obtain the foregroundthe optical flow as ground truth. Then our model estimates the information by taking only a single frame as input. For human behaviors, we take the human poses as inputuse a GCN-based model to predict the future poses. Both the anomaly scores of these two works are given by the error of the estimation. For defect detection, our model takes patches of the image as inputlearns to extract features. The anomaly score of each patch is given by the distance between the patchall the training patches."

We provides a wearable device for the visually impaired to walk outdoors. By the deep learning network, the system can recognizeguide the visually impaired to walk on safe areas such as sidewalkscrosswalk. In addition, it can recognize the types of common obstaclesguide the visually impaired to avoid it in advance. Finally, we can convert the Google Maps route into easy-to-understand voice prompts instruction to guide the visually impaired to move in the right direction.

Super fast terapixel server allows multiple users to viewanalyze gigapixelterapixel images with mobile devices, such as iPhones, iPads, Android mobiles,laptops at the same time. Users could annotateanalyze the data using built-in tools such as 3d effect, color deconvolution, color quantification, nuclei/object detection, lengtharea measurement,quantitative reporting system.