:::

【2021 Solutions】 InfuseAI focuses on building the PrimeHub MLOps software platform to lower the threshold for enterprises to introduce AI technology

In light of the rapid development of Artificial Intelligence (AI), InfuseAI has created PrimeHub, a one-stop AI deployment platform software, with the aim of lowering the barriers for AI adoption and assisting enterprises in a successful transformation. InfuseAI Inc., founded in 2018 by senior developers from the Taiwanese open-source community, g0v's Chia-liang Kao, and KKTIX founder Liang-Bin Hsueh, specializes in PrimeHub, the AI deployment platform software.

InfuseAI's COO Liang-Bin Hsueh shares insights on MLOps implementation and application

▲InfuseAI's COO Liang-Bin Hsueh shares insights on MLOps implementation and application

InfuseAI's COO, Liang-Bin Hsueh mentioned that AI project development is an iterative process. During the AI product lifecycle from data collection, development, model deployment to monitoring etc., it contains many technical debts that prolong the project cycle, increase costs, and significantly reduce the benefits. Furthermore, according to statistics, data scientists spend 65% of their work hours on tasks other than model development. The realization of AI not only involves development but also includes maintenance and cross-team collaboration issues. All these factors significantly slow down the speed of AI development to match the company's needs.

Liang-Bin Hsueh points out that the issues arising from the inability of AI development speed to keep up with enterprise needs can be specifically illustrated in the following two aspects:

• Speed of AI model development: In the past, it took 1 to 2 years to complete an AI project. However, as enterprises adopt AI, the project and AI model numbers multiply.

• Operational issues after AI model deployment: The lifecycle of an AI model begins after deployment. As data accumulates, AI models can be retrained to enhance performance. However, as the number of models grows, operational issues and computational resource bottlenecks emerge.

Hence, the MLOps platform—PrimeHub, developed by InfuseAI, encompasses the processes from AI model development, training management, to operational deployment and monitoring, offering a one-stop platform service through a smooth automated AI workflow that enables true enterprise AI implementation.

MLOps ecosystem

In other words, the InfuseAI team continuously adds to PrimeHub Apps by integrating third-party application services, actively collaborating with more manufacturers to seamlessly integrate AI models into PrimeHub, and eagerly anticipates cooperation with more businesses focused on AI technology and SI partners to inspire more applications on the MLOps platform and further promote large-scale AI implementation. Since its formation over three years ago, InfuseAI's clients include Taiwan AI Academy, E.SUN Financial Holdings, Sinopac Financial Holdings, National Taiwan University Hospital, and Chi Mei Hospital. Among these, InfuseAI works closely with Taiwan AI Academy to address various academic needs. Teaching assistants at each branch only need to operate simply in the PrimeHub platform, where all management tasks are automatically completed. Students in PrimeHub’s self-service platform establish a uniform pre-configured environment, allowing multiple deep learning calculations simultaneously, isolated by containers without interference. Additionally, assistants can decide on the data to load based on the course progress, automatically loading course files and datasets when students launch the environment.

Yushan Financial Holdings began intensive integration into AI development in 2018, acquiring GPU computing resources. They discovered the need for robust infrastructure to speed up operations amidst numerous individuals and projects concurrently. They sought a management platform to assist with computing resource management and data authorization.

Liang-Bin Hsueh states that the PrimeHub platform aims to help enterprises scale AI development, reducing model deployment time from days to hours, further facilitated by APIs and APPs to automate and optimize workflows. PrimeHub operates on a yearly subscription basis, promising ongoing optimization of the platform environment and the flexibility to offer customized services to different customers. Currently, it offers three solutions: PrimeHub Enterprise Edition, PrimeHub Deploy—a lightweight model deployment management plan, and PrimeHub Community Edition, allowing users to choose according to their needs.

(This article is a curated selection from the "AI Engineering Online Meetup")

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

Recommend Cases

【解決案例】神經元科技 動鑑觀顫 以眼為始 最佳智慧醫療輔助者
Neuron Technology, moving and observing, starting with the eyes, the best smart medical assistant

Looking around the world and in Taiwan, the number of patients suffering from vertigo is increasing year by year When faced with vertigo testing, most people will choose to have it diagnosed by a doctor or hospitalized with neuromedical imaging tools such as MRI or CT for interpretation However, this may lead to misdiagnosis of vertigo, overdiagnosis of patients, waste of medical resources, etc, and cannot be effectively traced back The actual condition of the patient's dizziness In order to solve the long-term clinical pain point of vertigo, Neurobit has developed a smart wearable medical device - NeuroSpeed smart glasses and an AI assistance system It provides an AI smart solution for rapid screening of vertigo, assisting doctors in quickly screening more than 80 of brain diseases Efficient detection power with accuracy and sensitivity of over 90 Neuron Technology was founded in 2016 The co-founders include Yang Juncheng, Chen Weicheng, Huang Jinxun and Wang Jingfu The four of them were studying at the National Taiwan University Medical Engineering Institute and the Creative Entrepreneurship Program to jointly offer the "Biomedical Innovation and Commercialization" course Next, let’s embark on the entrepreneurial journey together The core team members of Neuron Technology are mainly composed of cross-field experts in medical engineering, information technology, neurology, patents, business, etc The common entrepreneurial concept is to continuously explore and explore unsolved pain points in medical clinical practice Seeing that patients were suffering from vertigo, Neuron Technology invented and independently developed an efficient and practical wearable smart medical device, NeuroSpeed Smart Glasses, which combines hardware AI-assisted decision-making system software to provide an AI smart solution for rapid dizziness screening Neuron Technology CEO Yang Juncheng said that the proportion of people in Taiwan and the world who suffer from vertigo is very high In Taiwan, as many as 3 of the people about 700,000 suffer from vertigo, while in the United States, about 1,000 people suffer from vertigo Tens of thousands of people go to see a doctor due to vertigo From the data, it can be seen that the vertigo medical care market is quite huge Regarding rapid screening for vertigo diagnosis in the clinic, current rapid screening tools in hospitals mainly use neurological examinations, and most of them are processed through questionnaire-type scales The diagnosis will rely too much on the patient's chief complaint and the doctor's experience judgment, and lacks quantitative indicators, resulting in Misdiagnosis of vertigo and loss of golden treatment time For patients with vertigo who are admitted to the emergency department, medical diagnosis in the past relied heavily on neuromedical imaging tools such as computed tomography CT or magnetic resonance imaging MRI After observation, it was found that the equipment was insufficient In the United States, the number of patients admitted to the emergency department Less than 3 of the population can receive immediate MRI diagnosis, and less than 50 can receive CT diagnosis However, excessive use of CT or MRI diagnosis may lead to overdiagnosis and treatment of patients, and even waste of medical resources In order to help vertigo patients accurately and efficiently screen and triage and reduce over-diagnosis in clinics and emergency rooms, NeuroSpeed Smart Glasses, a wearable smart medical device developed by Neuron Technology, is believed to have a great impact on the overall application of medical resources and patient diagnosis and treatment It is very helpful Yang Juncheng said that if it is not necessary to use MRI to track the diagnosis, this device can be used as a rapid screening test before the existing MRI neuroimaging tools and after the patient is admitted to the hospital, reducing over-diagnosis of patients NeuroSpeed's examination method is to quickly screen for nystagmus to identify patients with potential brain diseases The examination method of this technology is to observe the trembling state of the eyes and a series of eye movements when the subject is staring at a stationary object, so as to judge Brain risks for appropriate diagnosis and treatment NeuroSPeed smart glasses can crack brain and eye diseases, clinical trial sensitivity reaches 90 With the active research and development of Neuron Technology, two generations of NeuroSpeed smart glasses have been successfully launched The first generation NeuroSpeed mainly provides doctors with real-time observation of high-quality eye images and eye movement response-related examination data reference, and does not provide auxiliary diagnosis services The second generation of NeuroSpeed allows physicians to record, track, store and view relevant eye movement analysis and pupil area identification, thereby effectively diagnosing diseases of the central, meridian and peripheral nervous systems Because the second-generation NeuroSpeed product has the function of assisting doctors in analysis and diagnosis, it has been classified as a Class II medical device in the United States and must be certified by the US FDA before it can be sold externally At this stage, this product has completed pre-submission and will be submitted for review after the biocompatibility test is completed It is expected that the second generation NeuroSpeed will successfully obtain US FDA certification next year In terms of actual verification fields, NeuroSpeed has currently cooperated with New Taipei Shuanghe Hospital and Taichung Tzu Chi Hospital for clinical verification Its clinical results have confirmed the classification judgment of central nervous system diseases such as stroke and other vestibular system diseases With the addition of NeuroSpeed Smart glasses are designed to be lightweight and easy to carry, detect quickly and easily, and have the functional characteristics of automatic interpretation They can effectively assist doctors in quickly screening for potential brain disease risks in patients with vertigo Field verification with foreign medical centers mainly involves vertigo patients admitted to the emergency department Physicians can use NeuroSpeed smart devices to interpret the phenomenon of stroke diagnosis Currently, the accuracy of clinical trials in the above-mentioned cooperative hospitals reaches 80 and the sensitivity reaches 90 Neuron Technology develops smart wearable medical device NeuroSpeed smart glasses In addition to becoming the best smart medical assistant for doctors to quickly screen brain disease risks in the clinic and emergency department, NeuroSpeed smart glasses also assist doctors in the detection and assessment of strabismus risk They are jointly developed with the Tri-Service General Hospital through neuron technology The "Portable Strabismus and Abnormal Eye Movement Diagnosis System" is implemented in the Tri-Service General Hospital, providing doctors with a complete preoperative assessment of strabismus and postoperative tracking basis This excellent RD technology achievement was recognized by the 17th National Innovation Award-Clinical Innovation Award in 2020 We know that many diseases can cause congenital or acquired eye movement abnormalities, and even cause symptoms of strabismus or nystagmus Among them, as for strabismus, according to expert physician consultation surveys, the global prevalence of strabismus among school-age children is 2 5, while the prevalence of strabismus among school-age children in Taiwan is about 1 to 4 Yang Juncheng said that clinical diagnosis of the symptoms of these diseases is mainly divided into two categories deviation measurement of eye position and eye movement test It has been observed that there is currently a lack of objective quantitative methods for diagnosing eyeball position deviation, and eyeball deviation measuring instruments are used to check subjectively For Neuron Technology, the current NeuroSpeed instrument and strabismus and abnormal eye movement diagnosis system software can integrate objective and clear eyeball images, eye muscle function and eye deviation angle data as a rapid test that meets clinical diagnostic criteria Service, the data of these tests can be used as the basis for doctors’ complete preoperative assessment and postoperative tracking of strabismus Therefore, this system can provide objective and effective quantitative analysis, which can greatly reduce the medical risks of patients during surgery In terms of business model, Neuron Technology mainly cooperates with hospitals to sell research-based products It is believed that the company expects to have significant revenue performance after passing the FDA next year At present, Neuron Technology will use the country's first wearable smart medical device NeuroSpeed smart glasses and AI assistance system to provide software and hardware integrated solutions to assist doctors in quickly screening brain diseases and eye diseases, extending patients' golden treatment time and reducing social risks health care costs to create a triple win NeuroSpeed smart glasses assist doctors in quickly screening brain and eye diseases Looking at the world, the pioneer of smart medical equipment, a new paradigm of biomedical software and hardware system integration In the early days of Neuron Technology’s entrepreneurship, technology research and development and business expansion faced many challenges Yang Juncheng expressed his sentiments that since the NeuroSpeed product is the first medical smart glasses developed in China, both in terms of obtaining hardware raw materials and developing technical specifications, The challenges associated with building and integrating related systems are higher than those for consumer electronics Additionally, bringing together cross-disciplinary talent and integrating it into a core team is also an entrepreneurial challenge In terms of business development, although the company has accumulated clinical verification results and published related outputs, the establishment of market channels still requires continuous efforts to expand It is hoped that it can continue to rely on the assistance of legal persons such as the Capital Strategy Council to actively create more diversified market channels and connect potential Cooperating manufacturers In order to accelerate international expansion and market operation exchanges, the company will conduct a new round of fundraising planning at this stage to meet development Since its establishment, Neuron Technology has taken long-term development and stable profits as its operating goals Yang Juncheng said that in addition to focusing on existing professional fields, we also develop and apply value-added products through the company's software development and software development The accumulated RD energy of hard integration is expanded to other medical disciplines and fields Yang Juncheng hopes that Neuron Technology can become a pioneer in smart medical equipment, with research and development oriented to meet the needs of users and patients, and also hopes to become a model for the integration of biomedical software and hardware systems For the medium and long-term layout, the company will promote industry-university exchanges and cultivate biomedical talents integrate upstream and downstream industry resources to create smart medical solutions and platforms and actively explore foreign sales markets such as the United States and Europe as a basis for sustainable development of the company Vision 「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.
Understand customers better than they do themselves, business opportunities created by the pandemic doubles the performance of Spingence Technology

"After the COVID-19 pandemic, even though we could not expand business overseas, the pandemic increased demand for AI automated optical inspection AOI and drove another wave of rapid growth in performance," said Jesse Chen, Founder and President of Spingence Technology The company started early in AOI and accumulated a wealth of experience, resulting in performance in 2021 growing multiple times compared with the previous year Spingence Technology's AI technology has great potential, and its fundraising plan successfully attracted strategic investors, such as the leading industrial computer company Founded at the end of 2015, Spingence Technology started out as a developer of automation software to help customers lower the barrier to automation In just two years from 2017 to 2018, AI technology quickly took off At that time, Jesse Chen observed that product defect detection caused headaches for companies in the process of automating manufacturing The traditional method of manual visual inspection was time-consuming and laborious Moreover, human error easily occurred and AI deployment could not be customized, resulting in high investment costs for enterprises Seizing the AOI market, successful application of AI optical defect detection "The gap in AOI is defect detection Since defects cannot be clearly defined, conforming products were often labeled as non-conforming products to meet the customer demand of not releasing defective products As a result, the overkill rate is often very high, which not only increases the cost of the manufacturing plant, but also lead to a waste of resources" In addition, there is strong demand for process automation due to the shortage of labor in the production line, which created a rare business opportunity for Jesse Chen AI optical defect detection has become the focus of Spingence Technology's AI applications After accumulating nearly 8 years of experience in production line automation, Spingence Technology target customers in three major industries, namely passive components, connectors, and semiconductors, and established a huge database for the production line product defect data it collected Since Spingence Technology started early, it enjoys the advantages of a first mover and continues to optimize database data Its wealth of experience enables it to quickly determine customer needs and propose solutions Spingence Technology uses image recognition and deep learning technologies to develop AOI algorithms, and developed AI defect detection solutions with an accuracy greater than 99 In terms of specific benefits, the overkill rate in the semiconductor industry can be significantly reduced from an average of 5-8 to less than 3, and the overkill rate of passive components can also be reduced from 5 to 12, which can save customers nearly NT300 million in expenses This not only reduces the waste of human resources, but also makes Spingence Technology a good helper for customers' sustainable intelligent manufacturing The outbreak of COVID-19 at the end of 2019 has changed the way people live and work around the world During the pandemic, demand on automation in the manufacturing industry significantly increased due to the need to avoid frequent contact between people Spingence Technology has a good reputation in the industry, so business kept coming in and doubled the number of customers and revenue The secret to winning Follow the footsteps of customers and always think more than customers In addition to Taiwan, Spingence Technology has also extended its reach to markets such as China and Vietnam, where Taiwanese businessmen gather The team in China is expected to increase from 3-4 people to 10 people, and Vietnam also hopes to have a team of 5 people to provide closer services to customers "Spingence Technology always follows the footsteps of its customers We are wherever our customers are" Jesse Chen went on to say that although the pandemic has brought another wave of demand for AI automatic detection, due to the impact of border controls, it was hard to go on business trips Meanwhile, overseas markets on-site personnel to provide nearby services, including consultation and diagnosis, AI implementation, and calibration, all of which require dedicated personnel Spingence's automation software development platform integrates modularized visual inspection, motion control, IO control, and AI analysis functions Coupled with the AI training software "AINavi", one heat sink inspection machine can complete multiple inspections and continue to train the AI deep learning model, which will further lower the miss rate of the machine and improve its inspection quality The integration of AINavi and automation is already sufficient to meet most functional requirements of customers, and then a small number of customized services are provided for different industries and customer needs This allows the company to quickly provide solutions needed by different customers Popular among strategic investors, the company has attracted investments from industrial computer companies and venture capital Spingence Technology has made a mark in the field of AIAOI inspection in the manufacturing industry, and has gain the favor of strategic investors In 2019, a leading industrial computer company made a strategic investment in Spingence Technology and integrated the AINavi deep learning visual inspection tool with the company's complete hardware platform The cooperation between the two parties provides more complete service solutions to customers, creating an example of win-win through "big large company leading small start-up" In February 2022, Spingence Technology reported another good news in its fundraising and received an investment from venture capital, all of its original major shareholders also increased their investment, and nbspSpingence Technology successfully completed the Pre-A round of funding The company will continue to expand its business in the short term, and establish a foothold in China and Vietnam In the future, it plans to build expert systems for passive components, connectors, and semiconductor industries, deepening its domain knowledge and data analysis to create greater value for customers Spingence Technology prides itself as a management consulting company for the manufacturing industry, that is, it provides customers with the most appropriate and valuable consulting services through consulting services, diagnosis, introduction, system launch, personnel training, model optimization, and AI model management

【解決方案】五年磨一劍 太奇雲端專注影像辨識 獲智慧城市創新應用獎
After five years of hard work in image recognition, Touch Cloud won the Smart City Innovation Application Award

During the opening ceremony of the 2022 Smart City Summit amp Expo co-organized by the National Development Council, Ministry of Foreign Affairs, Taipei City Government, Taoyuan City Government, Kaohsiung City Government, and Taipei Computer Association on March 22, 2022, founder and president of Touch Cloud Cheng-Hsun Li went on stage to receive the "2022 Smart City Innovation Application Award_Smart Security Award" This award is hard-won because after five years of hard work, Touch Cloud brought its core technology of AI image analysis back to Taiwan after gaining popularity in overseas markets, becoming an important smart city solutions provider "Since the company was established, we have positioned ourselves to make AI products that are stable, easy to use, and truly needed by the market" The company's positioning is very clear, this step took Touch Cloud five years It was not until the beginning of 2020 that Touch Cloud launched its first intrusion detection product In less than two years, Touch Cloud has now launched seven products, with each product receiving excellent reviews from the market Specializing in the field of smart cities, Touch Cloud successfully entered the Asian market Founded in February 2016, Touch Cloud originally focused on AOI defect detection However, the definition of defects is different for every customer As a result, applications in this field are mostly project development and could not be commercialized The company later directed its efforts to AI Application Box, a product that integrates software and hardware, and focused on smart cities, including transportation, industrial safety, and security related markets In view of the high acceptance and price advantage of AI in overseas markets, Cheng-Hsun Li targeted the overseas market in the early stages of the company's development, and has produced many successful results in several major countries, including Hong Kong, Singapore, and Thailand Among them, Touch Cloud successfully entered the Hong Kong market with its debut in the Hong Kong International Airport project In 2018, Hong Kong International Airport planned to monitor ships around airline oil storage platforms to prevent ships from colliding with the oil storage platform However, the oil storage platform is located in the middle of the sea and lacks electricity and Internet connection, so it could only set up a camera on shore 3 km away from the platform In addition to the long distance and poor line of sight, monitoring was also affected by weather conditions, such as clouds and fog, and was ineffective Touch Cloud customized an AI algorithm specifically for the project, and continuously collected image data for model training It took half a year of discussions with the customer and system corrections to finally achieve an accuracy rate of 98, successfully completing the goal Touch Cloud won the 2022 Smart City Innovation Application Award_Smart Security Award Minister of Economic Affairs Mei-Hua Wang on the left, Touch Cloud founder and president Cheng-Hsun Li on the right Touch Cloud products have three advantages that provide customers with a better AI user experience 1 Plug and play 2 Provides flexible services, in addition to standard products, customized services meet customers' needs for quick launch and reliable problem solving 3 High adaptability and ability to achieve extremely high recognition accuracy in different environments Cheng-Hsun Li believes that the design of AI products should be as simple as using a "home appliance" and have a clear purpose Therefore, Touch Cloud built an AI Application Box to design products from the perspective of scene applications, focusing on image analysis related to "people" and "vehicles" At present, a total of seven products have been developed, including KekkAI personnel intrusion detection, card swiping and tailgating detection, KekkAI-H construction site safety detection, Abaci-P people flow and head count, Abaci-V traffic flow and vehicle count, Greygoose personnel intrusion detection and people flow, GotchA cross-camera person tracking search, and AISense license plate recognition Cross-camera tracking product GotchA won the Smart City Innovation Application Award GotchA is the product that won the Smart City Innovation Application Award This cross-camera tracking product is unique in the market Its system uses AI technology to search and track people by analyzing their features at the time Users can use image search to search for the path of a specific person in multiple cameras they can also use multiple appearance features to find suitable target groups Using GotchA, you can actively search for lost people, understand the shopping behavior of VIP customers in the mall, and even manage the footprints of visitors in the building For example, it is not uncommon for children to get lost in hypermarkets GotchA can access footage from the entrance camera for an image search, and compare the walking path of the child to quickly find the missing child Using GotchA in shopping malls and amusement parks, where people come and go, to find lost people It not only significantly reduces the time spent by 90, but also transforms the traditional approach of the service counter making an announcement to the lost person into an active search based on image analysis, thereby improving service levels and accuracy Touch Cloud's excellent AI image analysis technology is favored by large enterprises Cheng-Hsun Li said that his experience abroad has taught him to help customers find AI usage scenarios For example, a T-Bar large advertising billboard operator in Thailand wants to attract business, and advertisers want to understand the flow of people to evaluate advertising effectiveness The T-Bar company has more than 1,000 cameras and can determine the type of vehicle from the images and then calculate the number of people, which can be converted to advertising effectiveness After Touch Cloud used the data for training, it can even identify special vehicle types in various countries For example, the nbsp"pickup truck" is a special vehicle type in Thailand Model training is carried out through Open Data, scenario data, and Touch Cloud's self-built database, and can accurately identify traffic flow and vehicle type, which is used to estimate the number of people reached by the advertisement Rapid business development after returning to the Taiwan market from overseas Touch Cloud currently accounts for 70 of the overseas market, with customers in Thailand, Hong Kong, Singapore, Malaysia, Japan, South Korea, and the Philippines In the future, it will also develop into other Asian markets, such as Vietnam In addition to the Asian market, Touch Cloud has not forgotten to serve Taiwanese customers, and began to actively recruit partners to jointly develop the Taiwan market in June 2021 So far, the number of customers has quickly accumulated to more than a hundred Touch Cloud's excellent AI image analysis technology is favored by many companies, and most investors are strategic partners, working together to create synergies and allow Touch Cloud's technology to be quickly applied Cheng-Hsun Li said that it often takes more than five years for an AI startup to get its operations on track, and it takes at least three years to cultivate good AI talents Results are gradually emerging thanks to the considerable flexibility and space given by strategic investors to the company Touch Cloud products are already being put into practical use in major Asian countries such as Hong Kong, Singapore, Thailand, South Korea, and Japan, and are expected to enter more Asian countries within two years, driving more advanced image analysis applications The company hopes to complete representative cases in Asian markets every year, and move towards an IPO as a company featuring AI products