:::

Special Cases

14
2022.3
【2022 Application Example】 Even the United Nations is on board! Yoyo Data Application captures global business opportunities with agricultural data

Nearly 2,000 days in the fields have made Yoyo Data Application a top player in Taiwan’s agricultural data sector Their comprehensive grasp of crop yields, production periods, and prices has enabled them to collaborate with the United Nations The service area for agricultural land skyrocketed from 24 hectares to over 6,000 hectares in less than three years—a 250-fold increase For Wu Junxiao, founder and CEO of Yoyo Data Application, aligning with global environmental trends and becoming a data company at the intersection of climate technology and the green economy to serve the global market is his ultimate entrepreneurial goal Wu Junxiao, originally an engineer, joined the Industrial Technology Research Institute in 2010, where he honed his profound technical and data science analytic skills 'At that time, I was working in data analysis engineering, and almost all data-related materials would be directed to me Additionally, I worked on indoor cultivation boxes, planting vegetables and mushrooms, hence planting the seed of entrepreneurship by integrating agriculture with data analysis,' Wu recalls Since 2016, Wu Junxiao has been frequently visiting farms to 'embed' himself among farmers and agricultural researchers, chatting and sharing information systematically, which quickly established his agricultural know-how Solid data analysis capabilities have even convinced the United Nations In 2017, he left the Institute to start his own business and founded Yoyo Data Application in 2019 Today, many agricultural businesses are his clients, with service areas rapidly climbing from 24 hectares to over 6,000 hectares, expected to surpass 7,000 hectares in 2022 His clientele includes markets in Japan, Central America, and even entities under the United Nations like the World Farmers Organization, which utilizes the 'Yoyo Crop Algorithm System' supported by Yoyo Data How exactly does Yoyo Data Application manage to impress even UN agencies The 'Yoyo Crop Algorithm System' developed by Yoyo Data Application accurately predicts the production period, yield, and prices Firstly, due to Wu Junxiao's precise mastery over agricultural data, Yoyo Data Application's clients don't necessarily need sensors or other hardware devices 'Sensors are expensive and if you buy cheap devices, you just collect a lot of noise or flawed data, which is useless,' Wu explains He continues, 'Collecting data doesn't necessarily require sensors our data solutions can solve problems more directly and effectively' For instance, one of Yoyo Data Application's products, the Yoyo Money Report Agri-price Linebot, developed in collaboration with LINE in 2020, gathers data on origin, wholesale, and terminal prices spanning over 10 years, driven by Yoyo Data’s proprietary AI algorithms This enables the system to autonomously learn about agricultural product trading prices, using big data and AI to perform price prediction analysis, thereby helping buyers reduce transaction risks and expanding the data application to the entire agricultural supply chain Regarding banana prices, the accuracy of price predictions increased from the original 70 to 998 Wu Junxiao notes that both buyers and farmers are very sensitive to prices Now, through the Yoyo Money Report service, both buyers and farmers can precisely understand the fluctuations in agricultural product prices Yoyo Data can also provide customers with optimal decision-making advice based on predictive models for crop growth, yield, and price estimations Currently, price predictions cover 28 types of crops Precise estimates of production periods and price fluctuations allow Yoyo Data to provide differentiated services based on data analysis The 'Yoyo Crop Algorithm System' provided by Yoyo Data Application incorporates a 'Parameter Bank', usually collecting 200-300 parameters, not just straightforward data like temperature and humidity, but also data divided according to the physiological characteristics of the crops Through effective dynamic data algorithms, it can accurately calculate when crops will flower and when they can be harvested, what the yield will be, and so forth For instance, the prediction accuracy of the broccoli production period is 0-4 days, with the flowering period predicted this year to be precisely 0 days, perfectly matching the actual flowering time in the field In these dynamic calculations, a 7-day range is considered reasonable, and the average error value of Yoyo Data's predictions typically ranges from 2-4 days, with most crop production period accuracies above 80 Through effective dynamic data algorithms, over 120 global crops can have their production periods and yields accurately estimated Using these effective dynamic data algorithms can set estimates for production quantities, helping adjust at the production end Yoyo Data Application's clientele primarily includes exporters of fruit crops like pineapples, bananas, guavas, mangos, pomelos, sugar apples, Taiwan's agricultural production is highly homogenized, often leading to a rush to plant the same crops and resulting in price crashes Yoyo Data Application helps clients differentiate their offerings Thus, Wu Junxiao positions his company as a boutique digital consultant, carefully selecting clients for quality over quantity He notes that Taiwanese agricultural clients focus on how to improve yield rates, even categorizing yield rates by quality, aiming for high-quality, specialized export markets whereas international clients prioritize maximizing per-unit yields, showing different operational approaches in domestic and international markets In addition to agricultural fruit, Yoyo Data Application has also extended its services to the fisheries sector, including species like milkfish, sea bass, and white shrimp, all using the same system to establish various parameters related to the growth of fish and shrimp, such as when to feed and when to harvest, and the anticipated yield, timing, and prices Yoyo Data Application harnesses the power of data to create miracles in smart agriculture In response to the company's rapid development, Yoyo Data Application introduced venture capital funds in 2021 to expand its staff and promote its business Wu Junxiao states that in response to the global trend towards net zero carbon emissions by 2050, he plans to help clients plant carbon in the soil, effectively retaining carbon in the land while also connecting clients to carbon trading platforms, creating environmental business opportunities together Wu Junxiao says that from the start of his entrepreneurial journey, he positioned the company as a global entity, thus continuous international collaborations are planned As a data company serving a global clientele and focused on climate technology and the green economy, this represents Wu’s expectations for himself and his company's long-term goals Yoyo Data Application founder and CEO Wu Junxiao「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

2022-03-14
【2021 Application Example】 Life-saving is as urgent as a spark AI critical illness system monitors and grasps the golden rescue period

60-year-old Mr Huang was admitted to the hospital due to a stroke After lying in the intensive care unit for two weeks, his condition suddenly took a turn for the worse After rescue, he was lucky enough to survive In fact, with the assistance of AI critical illness early warning technology, hospitals can detect signs and take timely and accurate medical measures 6-8 hours before a patient's heart stops, which can greatly reduce the chance of death in the hospital The deterioration of the condition is a process that evolves over time, and its subtle changes are by no means without context Previous research reports show that about 60 to 70 of inpatients who experience unexpected in-hospital cardiac arrest had symptoms 6 to 8 hours before their cardiac arrest, but only a quarter of them were recognized by clinical staff Detection and discovery, therefore, there is a need for a risk warning tool or system that can be used earlier and continuously to monitor the condition, alert medical staff to pay attention to subtle changes in the patient's condition at any time, and take timely and accurate intervention measures before the condition progresses to effectively reduce adverse events or the risk of serious adverse events Unexpected deterioration cannot be detected early Acute and severe patients often undergo unpredictable changes, and timely detection or prediction of potential acute and severe patients is an important issue The currently commonly used clinical assessment method is Modified Early Warning Score MEWS, which uses simple physiological parameter assessment including heartbeat, respiratory rate, systolic blood pressure, body temperature, urine output and state of consciousness to screen out high-risk patients, and has been proven to be predictive Patient clinical prognosis MEWS is a scoring mechanism with a single time point and a standardized formula However, the AI crisis warning system developed by Boxin Medical Electronics - Hospital Emergency and Critical Care Early Warning Index System EWS is designed to predict patient status with immediate response , collect the physiological data of patients over time for deep learning, find the best prediction model, and improve the overall accuracy Boxin Medical Electronics uses a big data analysis model to build an early warning system EWS, IoT Internet of Things and 5G communication technology, allowing medical staff to remotely monitor the physiological status of patients through communication equipment, and monitor emergency and severe cases quickly The patient's condition changes and the golden rescue period of 6-8 hours before cardiac arrest can be grasped After Boxin Medical Electronics introduces AI visual interpretation, unmanned operation can greatly reduce medical manpower The AI technology developed by Boxin Medical Electronics is the Gradient Boosting Ensemble Learning System GBELS to build an early warning system It is a learning-based EWS prediction algorithm developed by the company, which is an integrated learning Ensemble Learning and is classified as supervised learning, providing the following three functions 1 Early warning risk notification is used to analyze representative data using GBELS to provide an early risk score so that medical staff can conduct immediate clinical assessment and provide appropriate medical treatment 2 Reduce medical manpower Collect continuous physiological monitoring data, such as heartbeat, respiration, blood pressure and blood oxygen concentration, etc, to reduce the time for medical staff to write cases 3 Combine IOT logistics network and 5G communication technology to quickly transmit medical data such as monitoring parameters and imaging data, and assist medical staff to monitor changes in patients' condition remotely through communication equipment AI critical illness system monitoring to master the golden treatment period Boxin Medical Electronics stated that assessing the severity of the disease in acute and severe patients is a complex task, and patients often experience unpredictable changes Clinical medical staff often judge the condition based on their own clinical experience or intuition, which lacks science and objectivity, resulting in the inability to correctly identify and timely detect potentially acute and severe patients, resulting in or misdiagnosis leading to increased in-hospital mortality of patients The introduction of an AI early critical illness warning system can assist emergency and critical care medical staff to correctly predict the patient's condition and allow patients to receive the care they need immediately This can reduce the manpower arrangement of the emergency and critical care ward at the same time and reduce labor costs In addition, the easy-to-carry design will help the system be introduced into ambulances, home care and other places in the future, so that emergency patients can receive appropriate care earlier Other departments within the hospital can also develop new applications around this system, which can effectively accelerate the development and promotion of smart medical technology With the COVID-19 epidemic still raging in many countries around the world, this system can also help hospitals in various places to operate more effectively Caring for and monitoring the condition of critically ill patients In addition to AI critical illness warning, Boxin Medical Electronics has also developed AI image interpretation - Medical Physiological Monitor Life Cycle Compliance Testing AVS, which uses AI image interpretation technology to develop automated quality inspection of life support medical equipment The instrument solves the time-consuming problem of medical instrument testing It can reduce testing time by 70, increase the number of tests by 3 times, and effectively reduce labor costs by 50 At the same time, it is 100 compliant with regulatory requirements, and gradually solves the shortage of manpower and medical resources in the medical field , medical work overload and other issues It has now taken root in mainland China and is actively preparing for its launch in Europe It will develop towards the Japanese and American markets in the future Boxin Medical Electronics develops AI image interpretation-medical physiological monitor life cycle compliance testing AVS to solve the time-consuming problem of medical instrument testing and can reduce testing by 70 time At this stage, Boxin Medical's smart medical technology has been introduced into medical hospitals including Hsinchu MacKay, Changkei, Dongyuan General Hospital, Kaohsiung University of Technology Affiliated Hospital, Zhenxin Hospital, Hsintai Hospital, Taipei Medical University Affiliated Hospital, etc GE HealthcareInc, an internationally renowned medical materials manufacturer, and Mindray Medical, China's largest medical materials manufacturer, are both representative customers of Boxin Medical Electronics 「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

2021-10-12
【2020 Application Example】 AI Bread Recognition System, machine scans, and the price is instantly calculated for you!

A brilliant idea transforming AI facial recognition technology As artificial intelligence develops, more and more industries are embracing AI technology, even subtly entering into people's lives As most bakeries sell freshly made bread and pastries, which typically do not have barcodes, they rely on cashiers to visually identify each item and enter the type and price of the bread Thus, inspired by AI facial recognition technology, if such artificial intelligence could identify hundreds of types of bread, it could enhance checkout efficiency Diverse handmade breads delight customers but challenge clerks A local bakery has over 100 types of bread, regularly updating or adding new products, offering customers a variety of choices this poses a challenge for cashiers It takes two months to train a cashier, but even after they start, there's still a 5 to 10 error rate due to bread recognition mistakes each month, especially during peak checkout times after work, causing bottlenecks and further errors due to the stress on cashiers The difficulty in training cashiers and the lack of precision in the checkout process have long troubled businesses When baking meets artificial intelligence, it sparks a marvelous retail experience In typical bakeries, bread is sold 'naked' immediately after baking and then 'packaged' when it cools to room temperature Both methods require cashiers to recognize and remember the prices and undergo two months of training before they can work the cash register Even then, there is still a 5 to 10 error rate each month My Dee Bakery, with its extensive range of over 100 bread types, poses a significant challenge for cashiers Due to Yun Kui Technology Co, Ltd's expertise in developing iPad POS systems, which are designed to be simple, convenient, and easy to use, they allow businesses to check out efficiently and accurately Therefore, integrating the existing POS system with AI image recognition capabilities enables businesses to carry out transactions more efficiently and precisely AI bread recognition model operational schematic Image provided by Yun Kui Technology The execution can be simplified into eight steps, which include 1 Data collection Take bread image data at bakeries 2 Image annotation The image data is handed over to Mu Kesi Co, Ltd for manual annotation 3 AI modeling and training Managed by Mu Kesi, who adjusts AI models and training 4 iPad POS adjustment Simultaneous adjustments of the UI interface on the POS side and backend integration with the AI model 5 Start testing Once Mu Kesi reaches over 95 recognition accuracy with current data, formal integration testing begins 6 Real scene testing Move to the bakery to gather data and verify the correct recognition rates 7 Planning real scene application accessories When recognition accuracy exceeds 98, design accessories for on-site checkout, such as remote cameras and projection light sources 8 Official Application Integration with electronic receipts goes live POS machine AI bread recognition checkout process Start recognition - Recognition complete - Checkout - Confirm checkout, takes only 3 seconds Image provided by Yun Kui Technology AI bread recognition system, making multitasking easy After adding AI capabilities, not only can it save upfront training time and costs for bakery cashiers and reduce costs from recognition errors, but it can also speed up the checkout process and efficiency, increasing customer satisfaction This can later be promoted to various retail industries, expanding the new map of smart retail Before and after comparison chart of the bread checkout process with AI valuation Image provided by Yun Kui Technology「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

2020-05-21
【2020 Application Example】 "AI Color Recognition and Cost Optimization Control System" automatically recognizes colors, breaks through the traditional color grading model, significantly reduces costs, and improves yield!

Mixing new colors relies on the experience of master craftsmen The so-called "computer color matching" in the paint industry is simply the selection of "existing colors" for mixing, but there is actually no way to mix paint for a ldquonew colorrdquo and it all relies on the experience of master craftsmen Hence, it is necessary to start from scratch when a new color is encountered, which consumes a lot of manpower and time Moreover, due to the different color mixing habits of each master craftsman, the cost can be significantly different despite producing the same result The trilogy when paint factories face the crises of transformation I Lack of color mixing standards Generally, when traditional paint factories produce new colors, they will use a "spectrophotometer" to measure the LAB value of the sample color, and then the paint mixer will mix the paint of that color based on past experience After color mixing is completed, the instrument will be used to test the LAB value and C and H wavelength This process does not have a complete system and database records, and there are not standards for color mixing II Production costs are difficult to control Paint factories produce many pigments with different materials and functions, and the cost of paint will vary depending on the "color masterbatch material" used Even if the color number of the masterpiece is the same, the cost will be different if the ratio of the color masterbatch is different Paint mixers do not have a set of color mixing standards when mixing paint, making it difficult to control production costs III The color grading process is lengthy and personnel training is difficult As instruments cannot replace manual color mixing, the training of a paint mixer requires years of experience in paint mixing, familiarity with chromatology, as well as basic understanding of hue, saturation, and brightness If there is no basic reference color values when mixing paint, the paint mixer must spend a lot of time repeatedly mixing colors, resulting in a loss from time cost Developing an "AI Color Recognition and Cost Optimization Control System" The paint factory engaged in industry-academia collaboration with the Department of Computer Science amp Information Engineering of Chaoyang University of Technology through CDIT Information Co Ltd, and utilized the university's AI research capabilities to jointly develop the "AI Color Identification and Cost Optimization Control System" It established a database of "paint color numbers" and "color masterbatch material cost," and analyzes the optimal color mixing and optimal cost formula through data mining methods The paint mixer can refer to the formula analyzed by the system for color mixing, and then input the formula into the system after paint mixing is completed The formula is fed back to the basic database and an "artificial neural network model" is used by the system for deep learning, establishing a color grading standardization system for cost control and data collection, so as to solve the current difficulties faced by paint factories In the early stages of system development, CDIT planned the system requirements of the paint factory, established the system architecture and system database, and then worked with Chaoyang University of Technology on the implementation of model functions for the application of data mining and artificial neural network After the system is completed, CDIT will assist the paint factory in system testing and correction The system will be introduced after correction and testing are completed, and training on system use will be provided to ensure the correct use of the system System Screen Differences before and after using the system Expand new markets for the paint industry to see the paint industry thrive The "AI Color Recognition and Cost Optimization Control System" collects the color mixing formulas of paint mixers, establishes a paint color masterbatch formula database, and records the cost of each color number The system's deep learning function is then used with a spectrophotometer to analyze the optimal color mixing formula for each data entry, so that the paint factory can control the cost of paint mixing The optimal color mixing formula recommended by the system increases the speed of paint mixing and increases output value Future benefits include The improvement in product yield reduces customer complaints and improves customer satisfaction The breakthrough in the traditional color mixing model improves corporate image Improves the efficiency of paint mixing, and allows the remaining time to be invested in training to enhance the professional capabilities of personnel It will also allow the joint expansion of new markets with the paint industry and learning of new application technologies, and promote them to other paint companies, enhancing the industry's overall competitiveness to see the paint industry thrive

2020-08-11

Records of Application Example

【導入案例】AOI封銲製程全面檢測AI化，減少50篩檢量
【2020 Application Example】 Full Inspection AI Implementation in AOI Sealing Process, Reducing Screening Volume by 50%

Miniaturization of products, client demands full inspection A listed electronic component manufacturer in Taichung, responding to the 5G era injecting new growth momentum into the quartz component industry, especially under the explosion of 5G opportunities, the importance of quartz components will play a more crucial role than in the past in consumer products As frequency components move towards miniaturization and at the same time demand high precision, the manufacturing processes are more susceptible to subtle factors, necessitating manufacturers to manage comprehensive data across all aspects including human, machine, material, method, and environment to quickly identify key defective factors in complex production environments Differing perceptions of defects, difficulty in enhancing quality consistency With the trend of miniaturization and complexity of electronic components, visual inspection on the production line has four main functions including measurement, identification, positioning, and inspection, with inspection being the most challenging part as most electronics manufacturers still rely on traditional manual visual inspection Taking the PCB industry, where Automated Optical Inspection AOI technology has the highest penetration rate, as an example, a research institution once investigated and found that when two individuals inspect the same PCBA board four times, their mutual agreement rate was less than 28, and the self-agreement rate was only about 44 Due to differing perceptions of defects among on-site personnel, even automated machine vision can still lead to inconsistencies in product quality due to system settings or differences among quality control staff 偲捷科技檢測AI化，降低過篩率2030 With the support of the advisory team, collaboration with Sijie Technology aimed at the defects in the sealing process Based on CNN Convolutional Neural Network, the integration of multiple models introduced an AI recognition module to aid in the optimization of subsequent AOI tests, aiming to improve the accuracy of inspection equipment It is estimated that after introducing AI visual recognition, the over-screening rate could be effectively reduced to 2030 Thus, the industry, needing smarter inspection systems, has started applying AI technology to assist AOI equipment in optimizing subsequent screening tests AI-powered AOI Inspection Solution Cross-Model Design Concept Sealing AOI Inspection Trial Results Reducing false rejects, cutting manual screening workload by 50 The project, through a deep learning network architecture, reclassifies defects detected including true and false defects, and further classifies them to reduce the false reject rate of the traditional AOI solution This is anticipated to further aid manual inspectors in reducing more than 50 of the inspection screening volume, addressing current production line issues of relying heavily on manual re-inspection and low efficiency Future goals include integrating robotic arms for automatic loading and unloading, and analyzing defect causes to optimize production process parameters「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】處方箋智慧辨識 社區藥局藥師的小幫手
【2020 Application Example】 Intelligent Prescription Recognition: A Helpful Tool for Community Pharmacy Pharmacists

AI is thriving in healthcare services, where pharmacists in community pharmacies are essential for providing drug knowledge and pharmaceutical services However, these pharmacists often spend much time manually processing prescription entries into systems, which takes away from the time they could spend on drug education, medication effectiveness tracking, and other professional services How can AI help community pharmacies to support pharmacists Tedious, time-consuming, and repetitive tasks, and AI solutions Pharmacy operations are under threat from new market dynamics and limited profit-making modes, making digital upgrades challenging for single-pharmacist community pharmacies Pharmacists, taking on multiple roles to understand the health levels of community residents, face tedious, time-consuming, and highly repetitive tasks that hinder the quality of service and make it difficult to respond to customers non-stop all year round Smart Pharmacist Assistant Service Platform Enabled by Jiankangli Technology's smart pharmacist assistant service platform's system architecture, paired with the mobile application 'Smart Good Doctor' and the backend system 'Smart Good Pharmacist', along with the integration of external development feature resources 'OCR Prescription Recognition' and 'RPA Process Automation Training Module RPA library' Primarily applied in clinics and pharmacies at the primary healthcare level, this aims to solve various operational challenges and pain points It includes using digital technology to improve work efficiency, bridging the gap between the public and medical institutions, enhancing the medical relationship, achieving better operational and manpower benefits Additionally, it enhances medication safety for the public and improves their knowledge on medications, while also reducing the daily burden on pharmacists in pharmaceutical services Smart Pharmacist Assistant Project In the current stage, the Institute for Information Industry's team is guiding the integration of pharmacy information system vendors with AI startups, focusing on the development of intelligent prescription image recognition technologies, along with drug image recognition and smart drug scheduling reminder technologies as key research areas This has led to the implementation of practical deployments in 12 community pharmacies in Greater Taipei With the help of the Taiwan Young Pharmacists Association in promoting these technologies, over 100 community pharmacy proprietors have expressed interest in adopting such technologies Once the integration of these service platform systems is complete, it will become a model for promoting AI services in Taiwan's community pharmacy pharmaceutical services「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】AI助被動元件建構最佳AOI參數模型，降低過篩元件生產成本，年省250萬元
【2020 Application Example】 AI Helps Establish the Best AOI Parameter Model for Passive Components, Reducing Production Costs of Over-Screened Components, Saving NT$2.5 million Annually

Traditional AOI uses limited sample images for inspection, facing the problem of high over-screening rates In the electronic component manufacturing industry, AOI Automated Optical Inspection equipment is often used to measure defects in product appearance For a long time, AOI measurement equipment has used limited sample images in image processing to compare the appearance of products from different external light sources and angles This comparison method can automatically screen for defects in product appearance However, due to current technical limitations, there are often problems with light source parameter adjustment between product batches If an inexperienced technician handles these adjustments, it will lead to a decrease in machine utilization rate and high over-screening rates The maturity of AI image machine learning has brought new opportunities for the AOI process In terms of Taiwan's passive components, chip resistors and MLCC currently rank in the top two worldwide in terms of market share in 2019 In the long term, various car manufacturers have launched electric vehicles and smart vehicles, and various countries have also developed 5G-related equipment, which will further increase future shipments of passive components Therefore, besides expanding new product lines, how to help existing products enhance their competitiveness will be the key to the industry's future international competition AOI inspection is one of the common stations in the passive component process, limited sample images are used in the current stage to compare the appearance However, when switching between product batches, there are often problems with light source parameter adjustment, and the condition of these adjustments will affect the over-screening mis-screening of good products in each batch In each batch of defective products in the industry, on average over-screening mis-screening occurs 20 of the time Relying on the guidance capabilities of the Southern Taiwan Industry Promotion Center, which has been deeply involved in Southern Taiwan for more than a decade, the company was matched with the AI image recognition technology unit of the Industrial Technology Research Institute ITRI to address the pain points of the passive component industry, reducing over-screening in the AOI process and also reduce errors caused by manual adjustment Using image recognition technology to reduce the occurrence of AOI over-screening The technical unit of the ITRI that participated this time used image recognition technology to develop AOI technology for passive component processes in the establishment of AI modules In the development process, the company in this case first provided product appearance images and corresponding adjustment parameters, and then used the adjustment logic of current production line personnel to construct a product data set and further establish an AI model When planning the production line test, the first priority is image recognition rate Image detection and tag search are combined with comparison by an AI module to output AOI adjustment parameters for reference by online personnel Image analysis diagram In the future, we also hope to use the help of machine learning to complete the AI learning curve for machine parameter adjustment, further reduce the over-screening rate of product appearance defect detection, simultaneously solve the gap in on-site professional and technical talent, and increase product yield Scenarios before and after implementing machine learning Implementing AI applications in processes to lay the foundation for developing unmanned factories In the future, we hope the guidance of AI HUB will accelerate the application of advanced process technology and establish AI indicators for each station of the passive component process, which will help domestic production of high quality passive component products and increase product yields and prices It will use innovative thinking to increase the added value of the industry and continue to lead the passive component industry forward

【導入案例】動態車牌辨識系統 省時省力方便管理
【2020 Application Example】 Dynamic License Plate Recognition System: Time-Saving and Convenient for Management

Jiude Songyi Company, with 40 years in motor-related equipment manufacturing, introduced a dynamic license plate recognition system with an accuracy rate of 989 to effectively monitor vehicles entering and exiting the factory area The system uses AI technology, making vehicle management both time-efficient and effortless License plate recognition systems are a fundamental application of intelligent image analysis Using cameras to capture images of license plates, the system then analyzes and processes these images to recognize the plates Kangqiao Technology, established in 2008 by a team of LED developers and software engineers, specializes in LED product applications, developed license plate recognition and Etag integrated systems, primarily for domestic and international public works projects Recently, the III AI Team collaborated with the Taiwan Energy Technology Service Industry Development Association to explore real-world applications of license plate recognition technology They identified three major issues faced by Jiude Songyi Company at this stage 1 Currently, the company gate has no barrier machine or other control equipment Vehicle entry and exit rely entirely on manual control and recording If no personnel are present, vehicle movements cannot be controlled 2 When issues arise, the existing surveillance system has to slowly search for data to locate the problematic vehicle, which is very time-consuming and inconvenient 3 When the footage is found, it is often difficult to clearly identify the license plate, and even if found, it is not possible to verify the vehicle owner Solving Three Major Problems, Providing Four Major Functions After understanding the actual needs of the enterprise, according to the license plate recognition system architecture established by Kangqiao Technology, real-world validation was conducted on-site, with monitoring computers set up in the control room Kangqiao Technology License Plate Recognition System Architecture After installation, the main functionalities of the license plate recognition system are as follows 1 When vehicles enter or exit, high-resolution smart cameras can identify license plates and capture images, recording the license number and vehicle status 2 When file retrieval is needed, vehicle data can be searched by time or license plate information, allowing quick access to the required video files, saving considerable time 3 The use of high-resolution smart cameras significantly improves image quality, which helps in clear identification in case of incidents 4 With registered license plate data, a blacklist and whitelist database can be set up, facilitating the management by security personnel The advantage of license plate recognition is that it fully automates vehicle entry and exit control, reducing labor costs The software helps to prevent misuse of license plates and eliminates the issues of remote control, induction buckle loss, and borrowing by unauthorized persons Vehicles can enter and exit without using a remote control or rolling down the window The long-distance license plate recognition allows gates to open while the vehicle is still moving, eliminating the waiting time for parking Kangqiao Technology License Plate Recognition System Setup in the Management Room The III AI Team states continually collaborating with relevant associations, from identifying corporate needs, setting topics, linking teams, introducing real-world validations, to systematically assisting enterprises in need to adopt AI technology and solve industrial problems, aiming for the AI transformation of industries In the future, it will continue to help enterprises harness technology tools to overcome business challenges「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】核電廠「不玩了」 安全管理智慧化更重要
【2020 Application Example】 Nuclear Power Plant Calling It Quits: Elevating Importance of Smart Safety Management

Plant safety is a crucial aspect of industrial security Currently, many surveillance cameras are used in conjunction with manual monitoring by security personnel to provide information However, manual monitoring has its limitations Implementing an AI system to assist in detecting abnormal behaviors and facial recognition can significantly aid security personnel by covering blind spots in manual monitoring Located in Shimen District, New Taipei City, the Jinshan Nuclear Power Plant is nestled between mountains and the sea, boasting picturesque scenery However, this first nuclear power plant in Taiwan is entering its decommissioning phase and will soon become a part of history With the decommissioning process underway, numerous external contractors will be entering and exiting the complex, complicating access management The need for continual safety monitoring of external construction to ensure nuclear safety is critical Additionally, although the Lungmen Nuclear Power Plant is currently mothballed, it still contains sensitive areas and requires a reduction in staff presence, thus prompting an urgent demand for smarter safety management With assistance from the Taiwan Nuclear Level Industrial Development Association, the AI team at the Institute for Information Industry aims to tackle the issues of safety and occupational safety at the Jinshan Nuclear Power Plant with minimal staffing Based on interviews, the technology needs identified for AI implementation at the plant include personnel access control and safety monitoring of personnel and the plant area Facial Recognition AI Solves Two Major Challenges Personnel Access Control and Plant Safety Monitoring For personnel access control, a facial recognition system is deployed at the nuclear power plant Utilizing the uniqueness of human faces and AI's high recognition rate, the effectiveness of the plant's personnel access control is enhanced In terms of personnel operations and plant safety, an abnormal behavior detection system is also deployed This system utilizes AI to recognize abnormal or dangerous behaviors from the postures of individuals captured by surveillance cameras, promptly providing feedback to safety personnel for action Selected by the Institute for Information Industry, the solution from Wantech Intelligent Sensing abbreviated as Wantech focuses on developing facial and posture recognition functionalities After several discussions with Wantech, Google's Facenet and Posenet algorithms were chosen for implementation Facenet, requiring only 128 dimensions per face image, achieves optimal performance with just a few photos, making it particularly suitable for building industrial-grade facial recognition systems Posenet, used for motion detection, transforms data via a Data Processing Unit DPU into a format suitable for machine learning algorithms—Support Vector Machine SVM—for binary classification of human postures into falling or not falling categories Utilizing Visual Pages for Clear Management Interfaces The user interfaces for both systems are implemented using Python's web framework Flask, which provides web services adaptable across different operating systems, achieving a cross-platform purpose The Glasses App is developed using Unity to access web data In recent years, advancements in AI technology have increasingly incorporated facial recognition into safety management The unique characteristics of facial features eliminate the risks associated with RFID forgery and offer higher accuracy compared to other biometric recognitions fingerprints, voiceprints, complete objectivity devoid of personal bias, easy system setup and maintenance, and fully automated operations requiring no additional manpower Undoubtedly, incorporating facial recognition into safety management systems can significantly enhance the safety factor of the plant while reducing management complexities Body Posture Recognition Operating in the Laboratory Taiwan has four nuclear power plants, bearing significant management costs Continued implementation of AI technology solutions can not only reduce labor costs but also significantly enhance the effectiveness of safety management「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】AI導入營建業 減少工安意外 安心看得見
【2020 Application Example】 AI Implementation in Construction Industry Reduces Workplace Accidents: Safety Visibility Enhanced

The construction industry is Taiwan's leading industry, supporting the architecture, decoration, and repair sectors However, the high incidence of occupational accidents in this sector is a major concern for both employers and workers The introduction of AI for equipment recognition in the construction industry reassures companies and protects workers, creating a win-win situation According to the Ministry of Labor's 2017 statistics on occupational injuries, the average rate of occupational injuries per thousand workers across various industries is 2773 However, the construction industry tops the list with a rate of 10036, which is 36 times the average and categorizes it as a high-risk group for occupational injuries Proactive early warning measures can significantly reduce the rate of workplace accidents In light of this, the Institute for Information Industry, under the mandate of the Ministry of Economic Affairs' Industrial Development Bureau, has initiated an AI project that prioritizes the implementation of AI technology in the construction industry Selecting well-known construction firms in Taiwan, the project applies Canon's safety helmet proper wearing recognition solution to reduce occupational accident rates Smart Recognition of Safety Helmet Wearing A Solution for Employers Senior executives in the construction industry emphasize that compared to other industries, construction workers face higher health and safety risks primarily at construction sites Many risks arise from the workers not properly wearing or using personal protective equipment, such as safety helmets Relying solely on human supervision for ensuring safety gear compliance is time-consuming and often ineffective Implementing AI technology for smart monitoring on construction sites can save corporate resources while ensuring worker safety, achieving dual benefits Indeed, to protect workers during operations, construction plants require workers to properly wear safety helmets Wearing a helmet does not imply it is worn correctly To prevent the helmet from falling off during operations, it is necessary to securely fasten the chin strap directly under the chin after putting on the helmet 工地用安全帽正確佩戴方法 At construction sites, many foreign workers often do not follow proper safety protocols, such as not wearing safety helmets correctly If supervisory personnel were to be assigned, it would entail excessive use of human resources With the assistance of the information strategy team, major construction companies have adopted Canon's image recognition technology To determine the optimal placement of image recognition cameras, both teams first conduct site surveys and collect various types of safety helmets used on-site Subsequently, standard cameras are installed at entry points of construction sites and work zones to capture footage of the site personnel This footage helps Canon develop models for correctly and incorrectly worn helmets, aiding the image recognition software in its learning phase Canon's engineers regularly visit the site to retrieve footage, and once the image recognition software achieves a certain accuracy level, the image recognition cameras are then installed at the construction site 佳能工地安全帽資料搜集攝影機設置 Improving Recognition Accuracy for Concrete Implementation of Workplace Safety Currently, no local technology can accurately recognize the proper wearing of safety helmets Therefore, Canon has developed and trained its own recognition software The complex environment at the actual installation sites can impact the effectiveness of recognition In the future, machine learning will significantly enhance the overall recognition accuracy, ensuring that safety measures involving the wearing of safety helmets are concretely implemented While AI recognition technology is introduced in the construction industry's safety domain, it can also be integrated with mobile devices for early warning In practice, once a camera captures recognition data and processes it, the results can be pushed immediately to specific individuals such as safety managers on their mobile phones, tablets, or even linked to access control systems If a worker is detected without a properly worn safety helmet, relevant personnel can be alerted promptly Access can be denied until the worker correctly wears the safety helmet, offering considerable potential for future applications「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】AI點點名 掌握長者進出 解決日照中心人力荒
【2020 Application Example】 AI Roll Call to Monitor Elderly Entry/Exit and Solve Staffing Shortages in Daycare Centers

The silver storm is coming Taiwan will enter an 'ultra-aged society' by 2026 Daycare centers across Taiwan are facing a 'staffing shortage,' with AI facial recognition introduced to monitor entries and exits, making reliance on AI for roll calls a comforting solution for day centers How serious is the aging population issue in Taiwan Let's consider a figure by 2018, the proportion of the elderly population aged 65 and above in Taiwan had already exceeded 14, officially entering an aged society Moreover, according to estimates by the National Development Council, Taiwan will enter an 'ultra-aged society' where the elderly population exceeds 20 by 2026, aging even faster than Japan The council also predicts that by 2065, Taiwan's elderly population will exceed 40, implying that every 12 working individuals will need to support one elderly person Faced with a massive elderly population, daycare centers are bound to experience severe staffing shortages The informatization of Taiwan’s long-term care institutions is insufficient and urgently needs AI technology to address the staffing crisis Ian Wen, the vice-president of the Taiwan Long-Term Care Association National Federation, which has 800 members, states that unlike the medical industry continuously incorporating cutting-edge technology, Taiwan's long-term care sector has not benefited from Taiwan's world-class technological advances Small and medium-sized institutions depend heavily on manual labor With the introduction of AI technology to solve transformation problems, there can be substantial benefits for both the institutions and the elderly Responding to the industry's urgent calls, the Ministry of Economic Affairs' Industrial Development Bureau and the Institute for Information Industry have been actively seeking solutions Initially, the Institute focused on needs, collaborating with the Long-term Care Association to visit multiple institutions and understand their issues Most venues claimed that controlling exact attendance of elderly residents daily is necessary to comply with the long-term care subsidies Just before 7 AM, care recipients come in wheelchairs, with canes, driven by family members at the back door, and some who are supposed to arrive yet remain unseen The chaos at the entrance -- elders, families, and caregivers talking and bustling around -- makes it impossible to even hear each other By the time the roster call finishes, breakfast bought early in the morning is still sitting on the table This is a typical morning for caregivers at the daycare centers AI roll calls help solve current issues of staffing shortage and information inaccuracy Daycare centers commonly face issues with seniors having irregular attendance and check-in times Current operations only manage these through manual registration With multiple entrances, large and multi-level premises, and complex traffic including caregivers, administrative staff, elders, their families, and visitors, it's challenging to effectively manage them Additionally, manual roll calls can lead to errors during busy hours and even create misunderstandings regarding subsidy counts, causing problems for both the Ministry of Health and Welfare and the providers Thus, industry stakeholders are keen on using AI-enhanced devices to help healthcare staff, reduce manual documentation, and free up administrative staff time to assist more elderly care recipients With the mediational and advisory support from the Information Industry Institute, security monitoring providers Qizhuo Technology and Hangte Electronics have integrated facial recognition technology into long-term care institutions By setting up facial recognition devices at entrances and creating an innovative long-term leasing business model, they not only solve budget and staffing issues for small and medium-sized institutions but also help electronic device providers find suitable field verification sites, effectively solving problems for both supply and demand sides Qizhuo Technology solution implementation, left shows discussions with venue staff about installation details, right shows the detection screen Hangte Electronics solution recognition screen Facial recognition technology in long-term care progresses rapidly, capable of replacing the manual roll call systems and assisting caregivers during nighttime inspections, ensuring the whereabouts of elderly residents The application within daycare centers is expected to continue expanding「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】防範於未然 跌倒及危險區域偵測維護長者安全
【2020 Application Example】 Proactive Prevention: Fall and Hazardous Area Detection to Safeguard Elderly Safety

We all know that falls are a major concern for the elderly Once a fall occurs, it could lead to injuries or even life-threatening consequences that may be irreversible such as remaining undiscovered after a fall To counteract this, early warning through AI technology for fall and hazardous area detection can greatly enhance the safety of the elderly According to international statistics, the fall incidence rate among people aged 65 and above is 30-40 This implies that out of ten elderly individuals, 3 to 4 might experience a fall annually Indeed, falls are the most common cause of injury among the elderly Additionally, detections and warnings of risky behaviors in hazardous areas, such as scalds or slipping in the bathroom, can significantly reduce injury risks for elderly individuals To ensure that the elderly lead a long and healthy life with minimized accidental injuries, the AI team from the Institute for Information Industry actively collaborates with long-term care centers and AI device manufacturers Their goal is to meet the most urgent needs of the elderly, addressing areas where care centers, due to limited staff and resources, can't provide comprehensive care Accidents and injuries are among the top ten causes of death The establishment of an early warning system is urgently needed Statistics show that among the top ten causes of death for people over 65, in both Taiwan and the United States, accident injuries such as falls are included Post-fall, elderly individuals often experience a decline in mobility and quality of life In addition to physical injuries like fractures and bleeding, psychological impacts can also occur, causing them to avoid going out and leading to further physical decline Thus, preventing falls and providing immediate warnings to minimize fall-related injuries are crucial issues in elderly care Currently, the Institute for Information Industry's team is guiding collaborations between elderly care providers and AI device manufacturers The focus includes developing AI technologies for elderly facial recognition, along with technologies for detecting falls and hazardous behaviors, which are now being implemented in three elderly care facilities across northern, central, and southern regions for practical validation Collaboration between smart surveillance manufacturers and facilities effectively enhances recognition rates Mr Wu Jiachen, Vice President of Chiztech, stated that their smart surveillance technologies, including fall detection, facial recognition, and electronic fencing, have been well-developed but require practical validation sites to accumulate big data Introduced by the Institute for Information Industry, demonstrations in long-term care settings significantly improve recognition rates, greatly benefiting future applications Chiztech's developed fall detection solution Moreover, Mr Guo Hongda, Vice President of Hantech Electronics, who has been involved in safety surveillance for over 30 years, pointed out that the greatest key to successful smart surveillance lies in data accumulation and smart image analysis Establishing an AI database for various applications is crucial For instance, detected wandering can initially indicate whether the person's movement suggests discomfort or an anomaly, allowing immediate alerts to the monitoring center If an elderly person approaches potentially dangerous areas like a water dispenser or water heater, service personnel can be notified quickly to assist and prevent possible accidents, thus effectively facilitating early warning measures Hantech Electronics' developed fall detection solution With the assistance of the National Federation of Taiwan Long-Term Care Association, which has about 800 members, approximately 100 small and medium-sized care institutions have expressed interest in adopting the technology Once these facilities are fully equipped, they will become the seedbeds for advancing the AI transformation of Taiwan's eldercare sector「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【導入案例】從一顆包子窺看如何應用AI減少50報廢率，為冷凍食品提升60生產效能
【2020 Application Example】 Peeking into a Baozi to See How AI Reduces Scrap Rates by 50% and Boosts Production Efficiency by 60% for Frozen Foods

From production line to dining table, who oversees the hygiene management of what we eat In recent years, there has been a continuous stream of news reports concerning food safety, such as repackaging expired goods, and poisoning incidents at Hong Rui Zhen It's clear that people are increasingly concerned about the hygiene of their food However, due to various quality control methods in food processing, there are inherent risks The World Health Organization WHO has pointed out that unsafe food and water cause physical harm to 2 million people each year Hence, international markets demand that food processing companies must establish a traceability system for products This is why major domestic food processors also aim to set up a production traceability system to quickly trace back to problematic raw materials and initiate recall and destruction of problematic food Visible assurance, implementing production transparency A major domestic food manufacturer producing frozen food and instant meals has expanded its market presence to North America, New Zealand, Japan, etc They are also at the forefront in promoting food management domestically, having obtained certifications such as HACCP, ISO22000, ISO14001 Since food production is labor-intensive, it is prone to quality impacts caused by worker fatigue Additionally, the production lines often have unclear records of production quantities, processes, and timing This obscurity in traceability makes it difficult to track production information when defects occur, leading to food safety management gaps that result in the scrapping of entire batches To address this, the Production Development Center at National Sun Yat-sen University utilized its advisory resources to help the food manufacturer tackle food safety management challenges, planning the use of AI technology to collect production data and establish anti-fraud and traceability for food production Intelligent manufacturing boosts food safety Although the level of automation is not high in the processing of bakery products, the food plant in this case is keen to enhance the automation of its production lines and introduce smart manufacturing For businesses, a traceability system not only helps establish brand image and increase product and brand value, but also gives consumers peace of mind due to the transparency of production lines Therefore, the Production Development Center at National Sun Yat-sen University matched AI technology service providers, Hong Ge Technology, in the first phase to plan the introduction of data collection devices to link food work orders information, reducing human operational omissions and capturing real-time production information through dashboards to ensure the consistency of production stage information potentially affected by human factors Schematic for intelligent production line planning The second phase involves using deep learning during the dough fermentation stage to calculate size and volume, analyze the relationship between temperature, humidity, fermentation time, and product volume, and assess whether to introduce AOI foreign object detection after freezing as a second quality control step Schematic of AI-integrated quality control for finished products Food processing ID card, launching the AI-era of food safety tracing In Taiwan, the understanding and acceptance of production history by consumers is gradually improving From the supply of raw materials, processing, production, to distribution and sales, it is necessary to have complete control and provide transparent information Publicly disclosing the production history not only increases trust between enterprises and consumers, but also aligns Taiwan's food safety environment with international standards In 2020, the Production Development Center at National Sun Yat-sen University will assist enterprises with the adoption of advanced AI technology, documenting the entire data process from industry to dining table and supervising food production processes to successfully implement product tracing, prevention of adulteration, and the establishment of high standards for products, thus advancing food processing products to international standards「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

rows
Rows：67, 8 pages