:::

【2020 Solutions】 AI Enhancements, AOI Inspection Miss Rate at 0.1% Surpasses Manual Effort by 10 Times

Did you know a single golf ball can have up to 28 defect inspections? Manually, one can inspect 500 balls in an hour, but with AI, up to 6,000 balls can be inspected in the same time. Huiwen Technology has developed AOI (Automated Optical Inspection) technology that achieves a miss rate of 0.1%, which is ten times better than human inspection. Besides the golf ball industry, Huiwen Technology's AOI inspections are also being introduced to the textile industry and others.

Geng Cheng Lin, the founder and general manager of Huiwen Technology, has been an expert in artificial intelligence (AI) since 2013, recognizing the future potential and explosive power of deep learning (DL) and AI-based image recognition. AOI has always been a strong demand in the manufacturing industry, mainly to improve product quality for business owners, stabilize the quality of delivered goods, and use data from AOI inspections to improve processes, thus creating a virtuous cycle and further cost reductions.

Due to uncontrollable factors such as human eye fatigue and inconsistent standards, inspection encounters bottlenecks. The limit of human inspection miss rate after training is about 1-2%, and the situation worsens over time. AOI is a stable and capable of mass inspection device, achieving a miss rate of 0.1%, which is ten times that of human eyes, implying a detection rate of 99.9%. Of course, AOI also results in a 5%-10% over-inspection rate, which can be further screened manually. With the help of AOI, the burden of quality inspection is reduced, saving a considerable amount of labor time.

AOI Golf Ball Defect Inspection, Inspection Capacity Increased 12 Times per Hour

The first litmus test of Huiwen Technology's AOI technology was on golf balls. With their highly reflective, uneven surfaces, golf balls were previously inspected manually for defects. A tiny golf ball can have up to 28 defects, and traditionally, only 500 balls could be inspected per hour. A major domestic golf ball manufacturer, meeting the demands of Japanese customers, introduced AOI inspection two years ago. The high-speed, high-precision AOI system combined with AI deep learning image recognition technology conducts defect detection on golf ball surfaces, fully automates the feeding and outfeed process, replacing manual recognition of missed defects, and can immediately record defect conditions and report back, inspecting up to 200,000 packs of golf balls per year per machine, greatly enhancing customer satisfaction. However, this step took Huiwen Technology more than two years.

Golf Ball AOI Recognition Image

▲ Golf Ball AOI Recognition Image

Golf Ball AOI Recognition, 28 Surface Defects Unveiled

▲ Golf Ball AOI Recognition, 28 Surface Defects Unveiled

Lin Geng Cheng says, from data assessment and consulting, followed by data organization and tagging, selecting and verifying AI algorithms to AI training services, the golf ball data is like starting from zero, accumulating one by one. Thankfully, with full support from golf ball manufacturers, the efforts have finally bore fruit. With AI inspection, while manually one might inspect 500 balls in an hour, AI can handle 6,000, achieving effectiveness 12 times greater.

Unlike other companies, Lin believes that AI needs to delve deep into domains to scrape professional data since only with such domain data can AI perform well. Therefore, the company starts from individual projects, rather than setting an AI product from the beginning. Without quality data or a focused domain, the best algorithms cannot succeed in AI. Over the years, Huiwen Technology has accumulated project experience, gradually developing products while focusing on domain data and providing the latest AI algorithms to customers, growing together, creating a tighter collaboration, which is why, different from external fundraising, Huiwen's investors are customers or partners.

Evaluation to Official Launch: AI Introduction Requires Six Phases

The projects undertaken by Huiwen Technology are divided into several phases: 1. Evaluation period, 2. Initial Validation (POC) period, 3. Data Collection period, 4. Repeated Verification period, 5. AI Positive Cycle period, 6. Official Launch. The evaluation period involves understanding and assessing the Domain conditions of the demand side beforehand, followed by POC verification, extensive data collection after POC, entering repeated verification stage, and finally allowing AI to enter a positive cycle phase, achieving a certain level of effectiveness before the official launch. Generally, a project takes about six months to a year to develop. However, with more familiar PCB AOI projects, the first two stages are skipped, starting from data collection, thus significantly reducing the time.

'Regardless of this project or others, common questions from customers are: 'How much data is enough? When will AI learn?' Facing such questions, Lin points out that the reasons for these questions are: 1. The inexplicability of deep learning technology, as it is a black box; 2. Generally, customers lack the concept of AI technology. Thus, the company must patiently verify data repeatedly, identify the data needed by AI, accumulate and test it, and clarify and resolve all Domain conditions, which requires a lot of time and patience.

During the AI introduction process, customers have high expectations of integrating AI services, thinking that it can immediately replace human labor. Lin points out that this is not the case; the real value of AI lies in accumulating large volumes of high-quality data, which is then transformed and analyzed to establish AI training and verification models to fully address problems generated by manual processes.

Apart from inspecting golf balls, Huiwen Technology is currently targeting the textile industry for items such as fabric and shoelaces, and many industries have conducted POC trials through Huiwen, including the semiconductor industry, PCB industry, and other traditional industries.

AOI Fabric Defect Detection, Top Image Shows Before AOI Inspection, Bottom Image Shows After AOI Inspection

▲AOI Fabric Defect Detection, Top Image Shows Before AOI Inspection, Bottom Image Shows After AOI Inspection

Lin Geng Cheng indicates that the most difficult aspect of entrepreneurship is nurturing talent and customer recognition; customers often demand quick results, not realizing that AI adoption requires data accumulation and repeated verification, processes that cannot show results in less than six months.

Affected by the COVID-19 pandemic, the trend of globalization and centralization of the manufacturing supply chain has been disrupted, replaced by 'short region' supply chains, suggesting small, beautiful factories will flourish everywhere, potentially bringing new opportunities for AOI. Lin notes that high automation indeed offers opportunities for automatic inspection AI, however, relatively high capital investments, including automation equipment, mainframes, GPUs, and sufficient AI maintenance talents, are burdens that small and medium enterprises or small factories cannot bear, requiring government financial resources and input to facilitate smooth transformation.

「Translated content is generated by ChatGPT and is for reference only. Translation date:2024-05-19」

Recommend Cases

這是一張圖片。 This is a picture.
Smart Construction Site Security Platform

In construction site operations, implementing safety protection measures and establishing related processes are essential for controlling workplace safety Every business owner strives to minimize industrial safety risks To reduce the probability of workplace accidents, it is particularly important to inspect personal protective equipment PPE and safety measures The Yongyi Smart Construction Site Security Platform utilizes an AI-embedded system, not only to detect whether workers are properly wearing helmets, but also to manage access control at construction site entrances and verify worker identity The Smart Construction Site Security Platform is also a part of the government's push for the Smart Construction Label Initiative 'Smart Site Management' is one of the three main items under the 'Maintenance Management' indicator, highlighting the importance of 'Smart Site Management' This solution includes access management, surveillance management, safety management, and environmental monitoring as aspects of its AIOT solution Feature Highlights 「Translated content is generated by ChatGPT and is for reference only Translation date:2024-11-09」

【解決方案】瑕疵辨識率達百分百 耐銳利科技獲面板大廠青睞
Defect identification rate reaches 100%, Nairi Technology is favored by major panel manufacturers

On the machine tool production line, there are some slight differences in the first step of assembly Accumulated tolerances will cause the assembly work to be repeated, which is time-consuming and labor-intensive, resulting in shipment delays that will impact the company's reputation Narili Technology Company focuses on the field of smart manufacturing and provides various AI solutions It uses machine learning models to inherit the experience of old masters In the CNC processing machine assembly and casting process, it uses AI to analyze production line data, accurately adjust various data, and improve Production accuracy is 25 This AI production line data analysis system is called "Master 40" by Huang Changding, chairman of Naruili Technology It is the most evolved version of the master plus artificial intelligence It has been used in machine tool processing factories with remarkable results In addition, Nairi Technology used AI defect detection technology to participate in the 2021 AI Rookie Selection Competition of the Industrial Bureau of the Ministry of Economic Affairs, assisting AUO in advanced panel image defect detection, with an accuracy rate of 100, and won the award Assisted panel manufacturer AUO to solve problems with 100 accuracy in defect detectionHuang Changding further explained that during the production of general panels, edges and corners are There may be defects in the corners Although the defects are visible to the naked eye, AOI is often difficult to identify, causing the detection error rate to often exceed 30 Therefore, re-inspection must be carried out with manpower to improve the accuracy rate However, in response to the demand for a small number of diverse products and insufficient manpower, using AI detection is indeed a good method Nairui Technology, founded in 2018, has been able to win the favor of major panel manufacturers with its AI technology in just three years In fact, it has been honed in the field of CNC machine tools for a long time Tang Guowei, general manager of Narili Technology, pointed out that the top three CNC machine tool factories in Taiwan hope to introduce AI into the two production lines of assembly and casting Among them, on the assembly line, in order to maintain the accuracy of assembly, every part of the component is designed Tolerances are designed During assembly, each component is within the tolerance However, the cumulative tolerance still fails the final quality inspection and must be dismantled and reassembled This is not only time-consuming and labor-intensive, but also causes waste "After entering the production line, I realized that some masters have accumulated a lot of experience and are good at adjustment After his adjustment, the accuracy rate has improved a lot and the speed is faster" On the contrary, the new engineers did not Based on experience, it takes a long time to adjust and may not pass the quality inspection The yield rate of Master 40 system has increased significantly from 70 to 95Tang Guowei then said that the original size data set by Master during assembly All were recorded on paper After the information was written, it was stored in the warehouse and sealed No one studied the relationship between the dimensions Narili assists customers in designing the Fu 40 system Through the human-machine panel, the master can directly input the measured dimensions and related data during assembly After collecting data from different masters, AI algorithms are used to analyze the relationship between the data and create an AI model The AI model automatically notifies the operator what size to adjust to, and the quality inspection will definitely pass In this way, the yield rate will be improved It has increased significantly from 70 to more than 95 Narili Technology Company focuses on the field of smart manufacturing and provides various AI solutionsTang Guowei added, assembling the spindle of a CNC processing machine It took four hours In the first step, the machine made measurement errors, including vibration, temperature, speed, etc that were out of range It had to be dismantled and reinstalled, which took another four hours How to adjust after disassembly depends on the experience of the master At first, the master may have done the best assembly method based on experience, but the error rate was also 30, and the assembly took several days With the assistance of AI masters, the assembly time only takes half a day, and the yield rate reaches over 95, saving a lot of time and manpower "Use the AI model of machine learning to collect the experience of all the masters and provide it for AI learning The first step is digitalization, and the second step is knowledgeization This is the transformation of the enterprise "An important key", Huang Changding believes that Narili Technology is an important partner in the transformation of traditional manufacturing from automated production to digital transformation In addition, another industry that Naili Technology focuses on is the smart car dispatching system of the leading brand of elevator manufacturers The so-called car dispatch referring to the elevator car means that if there are more than two elevators, group management is required In the past, car dispatching was based on fixed rules If the elevator was closer to the requested car, that elevator would be automatically dispatched On the one hand, it did not take into account that dispatching a car if the elevator was called too many times might make other people wait longer The previous vehicle dispatching model did not take into account the usage characteristics of the building, resulting in a lot of waste For example, in an office building, there are peak hours in the morning, lunch break, and afternoon after work AI smart car dispatch can be flexibly adjusted according to off-peak and peak hours, increasing the efficiency of car dispatch, reducing waiting time, and reducing wasted electricity Introducing elevator smart dispatch to improve transportation efficiency and have environmental protection functionsHuang Changding added that just like the previous traffic lights at intersections, the system has already The number of seconds to stop and pass on highways, sub-trunks and small streets is programmed Smart traffic lights are now used to flexibly adjust waiting times to make road sections prone to congestion smoother Using AI to learn usage scenarios and introducing a smart dispatch system into elevators will improve transportation efficiency and make it more environmentally friendly In addition to introducing smart elevator dispatching, Nairili also introduced AI into the smart production and shipment scheduling system of elevator factories Elevator factories often cannot accurately estimate the customer's elevator delivery date For example, office buildings or stores must be completed to a certain extent before the elevator can be installed on the construction site If affected by unexpected factors such as delays in the customer's construction period, the elevator factory will often be idle or the schedule will be difficult to arrange Tang Guowei pointed out that generally those who understand the progress of client projects may be from business or engineering, but overall, the accuracy rate of shipments is only about 60, which means that 40 of them will not be shipped as scheduled Therefore, if the shipping schedule can be accurately estimated, the production line can be freed up for emergency orders or other product production needs The AI smart scheduling system will analyze past shipment data, about 20-30 parameters such as climate, distance between the factory and the construction site, and customer credit, and put them into the AI algorithm to accurately predict whether shipments can be made as scheduled goods Huang Changding also specifically stated that the machine learning of Naili Technology is not ordinary machine learning, but also incorporates various calculation methods such as traditional image processing technology and statistics Only by being very familiar with the domain knowledge can we make good products AI models are also where the company’s competitiveness lies He emphasized that the data that general SaaS platforms can process is very limited, and the accuracy rate has increased from 70 to 75 at most Naili’s strength lies in AI algorithms and machine learning, and it must be coupled with in-depth industry knowledge to produce output Good AI model Narili Technology started with the AI project, gradually deepened the technology, chose to start with the more difficult tasks, and accumulated rules of thumb It is expected to develop SaaS services this year 2022, based on customer needs starting point, gradually gaining a foothold and becoming an important partner in smart manufacturing The picture left shows the general manager of Naruili Technology Tang Guowei and Chairman Huang Changding right「Translated content is generated by ChatGPT and is for reference only Translation date:2024-05-19」

【解決方案】2秒鐘完成結帳動作 Viscovery AI影像辨識助攻智慧零售
Complete checkout in 1 second, Viscovery AI image recognition assists smart retail

Artificial intelligence AI has gradually changed the way various industries operate in recent years However, most of the work is still done by humans, with AI playing a supporting role This has led to emergence of the term "AI Copilot," which stands for "AI-driven tools or assistants" that aim to assist users in completing various tasks and improve productivity and efficiency The concept of AI Copilot comes from the role of "co-pilot" During flight, the co-pilot assists the main pilot in completing various tasks to ensure flight safety and efficiency In fact, there have been signs of various "machines" beginning to play the role of "copilot" in different fields since the Industrial Revolution, assisting humans in completing heavy physical and repetitive tasks, greatly improving factory production efficiency, and driving rapid economic development Following the advancement of computing equipment and breakthroughs in machine learning, deep learning, and image recognition technologies, the concept of AI Copilot has gradually taken shape The development of AI Copilot marks the transition from "machine-assisted to AI-assisted" Early robots could only complete preset repetitive tasks, but today's AI copilot can learn and adapt to new environments and tasks, and continuously optimize its performance in practical applications This transformation not only changes human-machine interactions, but also has a profound impact on various industries The application scope of AI copilot covers various industries, including finance, healthcare, manufacturing, education, retail, etc, and are everywhere to be seen Application of AI copilot in the retail industry AI image recognition checkout In the retail industry, the application of AI copilot has begun to show concrete results Take Viscovery's AI image recognition checkout system as an example This system is a type of AI copilot model that helps store clerks speed up checkout or assists consumers in simplifying the self-service checkout process The store clerk needs to scan the product barcodes one by one in the regular checkout method If a product does not have a barcode, such as bread and meals, the clerk needs to first visually confirm the items, and then input them into the POS checkout system one by one Based on actual measurements at a chain bakery, it takes 22 seconds for an experienced clerk from "visual recognition" to "entering product information of a plate of 6 items into the checkout system" New clerks may need even more time In addition, according to a Japanese bakery operator, it takes 1 to 2 months to train employees to become familiar with products Now with AI image recognition technology, store clerks let AI handle the "product recognition" step, and AI will play the role of copilot, quickly identifying items within 1 second, speeding up checkout to save 50 of checkout time, and optimizing customers'shopping experience The time cost of training employees to identify bread can also be effectively shortened Even for products with barcodes, AI can quickly identify multiple items in one second, which is more efficient than scanning barcodes one by one The self-checkout system "assisted" by AI image recognition allows consumers to successfully complete shopping without the help of store clerks, eliminating the trouble of swiping barcodes or searching for items on the screen, which improves the shopping experience In a time when store clerks are hard to hire due to labor shortage, this also helps stores reduce operating costs AI quickly identifies multiple checkout items in just one second Source of image Viscovery Recently, startups dedicated to developing AI image recognition checkout solutions have emerged in various countries The most lightweight solution currently known is in Taiwan It can be immediately used by installing a Viscovery lens and a tablet installed with Viscovery AI image recognition software at the checkout counter to connect to the store's existing POS checkout system There are various integration methods, including plug-and-play and API solutions integrated with the store's POS system Viscovery AI image recognition system can be painlessly integrated with the store's existing POS system Source of image Viscovery Example of AI image recognition checkout Currently, the Viscovery AI image recognition system is being used in bakery chains in Taiwan, Chinese noodle shops in Singapore, micromarkets in department stores in Sendai, Japan, and Japanese bakeries and cake shops Over 7 million transactions were completed through this AI system, which identified more than 40 million items These use cases demonstrate the extensive application of the Viscovery AI image recognition system in the retail industry In the future, the company will continue to explore the various possibilities of using Vision AI in retail and catering nbsp The Viscovery AI image recognition system is already being used in bakeries, cake shops, restaurants, and convenience stores in Japan, Singapore, and Taiwan Source of image Viscovery