:::

【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

▲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.

Before and after scenario of incorporating machine learning

▲ 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.

Recommend Cases

【導入案例】巨量遙測空間數據AI分析雲端服務平台 使衛星遙測影像順利落地應用
AI Analysis Cloud Service Platform for Remote Sensing Big Data Enables the Smooth Application of Satellite Remote Sensing Images

Although satellite remote sensing images can make all surface objects visible, it still requires a lot of time and manpower to be truly applied to the industry In order to effectively solve the problems that customers face in digesting huge amounts of image data and eliminate technical obstacles for cross-domain users to process satellite remote sensing images, ThinkTron has developed an "AI Analysis Cloud Service Platform for Remote Sensing Big Data" as a new beginning for cross-domain AI applications for spatial information In recent years, in response to the impact of industrial globalization, Taiwan's agriculture has continued to transition towards technology-based and higher quality, improving the yield and quality of crops by solving problems, such as microclimate impacts and pest and disease control The demand of agriculture on images has expanded endlessly to accurately grasp the growing environment of crops In the early years when UAVs unmanned aerial vehicles were not yet popular, manual field surveys were the most basic but most labor-intensive work With the emergence of UAV drones, aerial photography operations might not be difficult, but the range that can be photographed is limited Furthermore, surveying expertise is required to accurately capture spatial information At this time, the use of satellite remote sensing data may break away from the past imagination of using image data Taiwan Space Agency TASA ODC data warehouse services In the past ten years, with the breakthrough of modern satellite remote sensing application technology, Digital Earth has become a new trend in global data collection Countries have developed data cube image storage technology, and the development of smart agriculture has become one of the largest image users Determining planting distribution is the first step in understanding crop yields Free satellite remote sensing images, powerful data warehousing support, and the team's robust image recognition technology are important supports for accelerating agricultural transformation Using satellite remote sensing image data can accelerate the development of smart agriculture However, in the past, it was difficult to extract large-area crop distribution through satellite remote sensing images, not to mention the cost If you wanted to use free information, you had to visit all websites of international space agencies, look through the wide variety of satellite specifications, and carefully evaluate the sensor specifications, image resolution, and revisit cycle After finding suitable images, you had to look at them one by one to filter the ones you need Next is downloading dozens of images that are often several hundreds of Megabytes MB each, which might exceed the capacity of your computer Also, after overcoming image access and preparing data, you must then start to confirm the downloaded image products and which bands you want, because the image you see is not just an image file jpg or png, but rather complex multi-spectral information, attribute fields and coordinate information It takes a lot of effort just to confirm the correct information Facing GIS software packages with complex functions is the start of another trouble The complex image pre-processing process and the inflexible machine learning package greatly reduce the efficiency of analyzing data After finally getting the results of crop identification, you might find that the best time for using map information may have already passed The above-mentioned complex and time-consuming satellite image processing problems are precisely the pain points of the market ThinkTron expanded from traditional machine learning to modern deep learning applications, and developed an "AI Analysis Cloud Service Platform for Remote Sensing Big Data" under the GeoAI framework, breaking through the constraints of details in the spatial information for customers Differences between the process before and after introducing the AI analysis cloud service platform ThinkTron said that Taiwan's ODC Open Data Cube system has been completed and began providing services after years of efforts from the Taiwan Space Agency TASA, formally becoming aligned with international trends The powerful warehousing technology allows users to easily capture and use image data of a specific time and spatial range according to their needs The warehouse stores multiple satellite image resources from international space agencies, including the ESA's Sentinel-1 one image every 6 days, Sentinel-2 one image every 6 days, USGS's Landsat-7 one image every 16 days, Landsat-8 one image every 16 days, and the domestic Formosat-2 one image every day and Formosat-5 one image every 2 days ThinkTron develops satellite image recognition tools based on Python Breaking free from the limitations of GIS Geographic Information System software packages, ThinkTron integrated GDAL Geospatial Data Abstraction Library based on Python, and considered computing efficiency and parallel processing when developing all tools required for satellite image processing and image recognition modeling, including coordinate system and data format conversion, grid and vector data interaction, and data intra-difference and normalization All of the tools are designed with AI applications in mind, and some commonly used tools are packaged into an open source package under the name TronGisPy to benefit the technical community ThinkTron utilized the team's understanding of satellite remote sensing images and the collected tagged data crop distribution maps to preset the image recognition modeling process, the required training data specifications, and dataset definitions This is imported into the machine learning LightGBM or deep learning CNN framework that was completed in advance, and the entire training process to be performed in the Web GIS interface, providing users with partial flexibility to freely filter images, confirm spatial and temporal ranges, select models, and adjust hyperparameters In addition to the operation of training models, it also provides historical models to output identification results, and finally displays the identification results of crop distribution on the Web GIS map In fact, agriculture is not the only industry that needs satellite remote sensing applications AI applications of spatial information have also appeared in various fields as companies in different industries aim to enhance their global competitiveness For example, surveying and mapping companies that have a large amount of map data can use the AI analysis cloud service platform to store map data while also accelerating the efficiency of digital mapping Under the severe global climate change and the risk of strong earthquakes, there is a wide variety industrial insurance, agricultural insurance, financial insurance, or disaster insurance are all inseparable from spatial information The use of remote sensing image recognition to understand insurance targets has long been an international trend AI Analysis Cloud Service Architecture for Remote Sensing Big Data

【導入案例】防患於未然 麗臺科技研發心臟衰竭AI辨識技術可及早發現病徵
Preventing Problems Before They Arise: Leadtek Research Develops AI Technology for Early Detection of Heart Failure Symptoms

With the increase in the elderly population, the incidence of various chronic diseases is rising daily Among these, heart failure is not only a silent killer it has a very long disease course with a high recurrence rate, leading to increased burden on healthcare personnel However, by using medically certified electrocardiography acoustics devices, coupled with AI predictive assessment of heart failure risk and remote care systems, diagnosis can be aided significantly, helping doctors make accurate diagnoses for subsequent patient medical care or referrals Heart failure has a lengthy course and medical expenditure is five times that of diabetes If you find yourself short of breath even with minimal movement, or if you wake up from sleep needing to sit up to feel comfortable, or if you have symptoms such as swollen lower limbs, anxiety, restlessness, fatigue, or a loss of appetite, be cautious These could be signs of heart failure According to statistics, there are about 60 million people with heart failure worldwide, with 5 million new cases every year In China, nearly 290 million people suffer from cardiovascular diseases, accounting for the second leading cause of death among urban residents around 12 million of these are heart failure patients, accounting for over 59 of cardiac-related deaths The disease course of heart failure is exceptionally long, and both its recurrence and rehospitalization rates are exceedingly high, resulting in medical costs that are twice that of hypertension and five times those of diabetes According to US research statistics, the 30-day mortality rates for patients with myocardial infarction and heart failure are respectively 166 and 111, and the rehospitalization rates within 30 days are 199 and 244 The symptoms of heart failure, because they are similar to those of other diseases such as chronic obstructive pulmonary disease and asthma, have an 185 misdiagnosis rate, which poses a challenging problem for healthcare institutions Leadtek, a major graphics card manufacturer, has been investing in the medical and healthcare sector since 2000 Following two heart attacks in 2011 and 2015 experienced by Chairman Lu Kunshan, Leadtek has focused on health big data, independently developing AI technology for heart failure recognition This AI application reads patients' electrocardiograms and phonocardiograms to perform anomaly detection and model prediction of heart failure risk, enabling early detection of disease symptoms Leadtek independently developed heart failure AI recognition technology to predict medical history and risk Leadtek's independently developed heart failure AI recognition technology has the following three judgment functions 1 Prediction of heart failure history Classifies electrocardiogram and phonocardiogram data into 'with hospitalization history of heart failure' and 'no history of heart failure' 2 Risk prediction of heart failure Provides a predictive risk value of heart failure occurrence based on the electrocardiogram and phonocardiogram data 3 Prediction of heart failure recurrence risk For patients with heart failure, it reads their phonocardiogram and electrocardiogram data, assessing the risk prediction of heart failure recurrence Leadtek states that the application of heart failure AI recognition technology can assist doctors in making more efficient and accurate diagnoses, facilitating subsequent medical treatment or referrals for patients As an instance, in studies of heart failure patients discharged from Taipei Veterans General Hospital, using the EMAT Electromechanical Activation Time index and SDI Systolic Dysfunction Index calculated by the synchronized electrocardiography-acoustic device as treatment guidelines resulted in a higher survival rate compared to those treated based on traditional symptoms This research has also been published in the authoritative international cardiology journal JACC, receiving recognition in the international market System manufacturers can apply heart failure AI recognition technology for other value-added applications Leadtek states that cooperating system manufacturers can choose to build their own heart failure AI risk prediction engine, uploading their system's electrocardiogram and phonocardiogram data to Leadtek's heart failure AI risk prediction engine, which then returns risk prediction values for integration by system manufacturers cooperating manufacturers as a value-added application input Not just for clinical use, the heart failure AI risk prediction engine can also be extended for use at home or in the workplace Additionally, this system can be extended to other applications, including One, hospital outpatient screening Doctors can use the electrocardiogram and phonocardiogram recorder along with the heart failure AI risk prediction model to conduct a 10-second rapid test in outpatient and emergency departments to assess a patient's cardiac history and heart failure risk Two, discharge risk assessment Doctors can use the electrocardiogram and phonocardiogram recorder along with the heart failure AI risk prediction model to assess the heart failure risk during a patient's hospital stay The test data can serve as a pre-discharge risk assessment and prognostic indicator Three, continuous home care Patients can use the electrocardiogram and phonocardiogram recorder, wearable electrocardiogram recorder, and transmit through a home transmission box gateway to measure electrocardiogram and phonocardiogram signals at home and upload them to the amor health cloud platform for heart failure AI risk prediction analysis Patients can manage their health autonomously via an APP, reviewing historical physiological trends disease management nurses can manage member health through the health management backend Web Four, home rehabilitation training Patients can wear a health bracelet to monitor activity, fatigue, circulation, and sleep, autonomously managing their health through the mobile APP and observing the risk of heart failure, engaging in exercise and rehabilitation training to aid in swift recovery The heart failure AI recognition technology system can also be extended to employee home care applications Additionally, in factories or offices, this system can also achieve employee health management goals, with applications including One, workplace safety units Provide employees with wearable electrocardiogram recorders before they start work duties Two, physiological monitoring for business executors While executing business duties or training, employees wear wearable electrocardiogram recorders for fatigue warnings, signaling whether physiological conditions allow continued execution of tasks Task segments can use data transmission boxes or apps to upload physiological monitoring information to the health management platform, assessing the heart failure risk for operations staff, with test data serving as an indicator for enterprise resource human units and public safety Three, workplace physiological monitoring center care The workplace physiological monitoring center can inspect and record employees' historicalphysiological trends through the health cloud platform Four, workplace nursing units Nursing units receiving instructions from the physiological monitoring center can provide health management advice based on employees' physiological trends nursing centers can manage employee health through the health management backend Web Five, employees can wear health bracelets to monitor activity, fatigue, circulation, and sleep, autonomously managing their health and observing the risk of heart failure through the mobile APP, engaging in exercise and rehabilitation training to aid in rapid recovery Workplace application of heart failure cloud care and big data center diagram「Translated content is generated by ChatGPT and is for reference only Translation date:2024-05-19」

這是一張圖片。 This is a picture.
AI Assists the Red Cross for Smarter Emergency Response

More Preparation Less Loss The Taiwan Food Bank Association, a non-profit organization, collects donations daily from wholesalers, retailers, manufacturers, and even kind-hearted individuals across Taiwan They also rescue consumable materials that are about to be discarded, properly allocate and deliver to households in need, aiding local underprivileged populations When natural disasters such as earthquakes, landslides, mudslides, typhoons, floods, and droughts occur in Taiwan, the food bank's resources can be immediately deployed for disaster relief This field verification unit is the Nantou County Red Cross AssociationOne of the food bank locations, hereinafter referred to as the Nantou Red CrossIs responsible for tasks like pre-disaster supplies preparation and disaster relief material distribution, helping the government bear the responsibility of disaster relief and aid In Taiwan, various natural disasters have characteristics of different duration and spatial coverage, wide or narrow With the normalization of extreme weather, the scale and number of disasters are gradually increasing and becoming harder to predict The required amount and type of materials differ by disaster, and they must address the lifestyles of the affected areas, rescue needs, traffic conditions, geographical restrictions, and other factors for varied material allocation, facing numerous challenges Typhoon Kanu severely damaged transportation in Nantou mountain areas Nantou County Red Cross planned the mountainous route Puli gt Fazhi Elementary School gt Qin'ai Village gt Aowanda to deliver supplies Disasters happen repeatedly We need to be prepared at all times Effective disaster preparedness can mitigate the impact, including swift response to material needs in affected areas, aid distribution, and even psychological support, providing added security for life and property of those in disaster zones Lack of Timeliness in Disaster Information To improve the living conditions and address the lack of supplies in remote areas, the Taiwan Food Bank Association has partnered with the Nantou Red Cross and has successively established food bank points in Nantou City, Puli, and Ren'aiLixing, Ruiyan, XinyiWangmei, Tongfu, Shuili, Lugu and Caotun among others9establish food bank locations, providing supplies worth a certain amount per household every month6001000in New Taiwan Dollars However, many challenges still need to be overcome during natural disasters For example, when typhoons, earthquakes, and landslides occur, the information source for disaster relief dispatch systems relies on post-disaster reports The time lag between reporting, response, and execution prevents timely adjustment and distribution of 'disaster relief' supplies based on the needs of affected areas, affecting rescue efficiency due to lack of timely information The 'preparedness' supplies of the Nantou Red Crosssuch as dry food, water, instant noodles, etc,are recorded manually in terms of stock, expiration dates, and distribution,When a disaster occurs, there is a chance that 'preparedness' supplies have expired and cannot become 'disaster relief' supplies It’s also possible that both conditions mentioned above occur simultaneously, leading to a need for more time to reassign 'preparedness' supplies into usable 'disaster relief' materials On the other hand, upon receiving information about shortages in disaster areas, the supplies donated by the public often grossly differ from the actual needs of the disaster zone, leading to an excess of supplies The Process of Material Operations Before and After a Natural Disaster AIAnticipating Natural Disasters Reinforcing the Accuracy of Preparedness Material Dispatch Application API Technology connects to compute the state of the climate, the intensity of disaster rescues, prioritizing the main tasks of the Nantou Red Cross and the needed areas of search and rescue Coordinated with the existing heavy rain and typhoon simulation disaster training of the Nantou Red Cross, a 'Natural Disaster Emergency Preparedness Material Dispatch and Supplement Decision System' is establishedreferred to as the Emergency Preparedness Material System。 In material management, inventory data along with immediate supply data are entered into the Emergency Preparedness Material System for comparison and analysis, helping the Nantou Red Cross quickly recognize materials like cookiesdry food, beverages, frozen food, toilet paper, etc, and determining whether they should be 'preparedness' materials or regularly distributed materials Adding to this, information forecasting understands the potential disaster conditions in remote areas, facilitating food delivery, addressing both front-end food wastage and backend practical needs When a natural disaster occurs, it enables faster response and decision-making, completing material deployment, hence increasing the speed of material operation transition20。 AI Emergency Preparedness Material System Helps Rapidly Adapt Material Distribution Through the field verification of the Nantou Red CrossAIthe system, material management, and related applications are promoted to more emergency response organizations in different areas, while continuously improving the alert functions within the Emergency Preparedness Material System, strengthening the technological foundation for alerts, enhancing prediction accuracySystem immediacy, and optimizing the data collection and analysis process Simultaneously, it can collaborate with government agencies, meteorological departments, or other rescue teams to discuss integrating more data sources, establishing a mechanism to share resources and data promptly, sharing information in real-time, helping more emergency response organizations enhance their disaster response abilities, seizing the golden rescue time 「Translated content is generated by ChatGPT and is for reference only Translation date:2024-12-12」