AI AOI Manufacturing Case ft. NVIDIA & Spingence, InnoTalks EP25 (2/3)

So in the next section, we are glad to invite a key independent software partner, Spinjens, to introduce the AI AOI application in manufacturing. Let's welcome the CEO of Spinjence, Jesse Chen. Hi, Jesse. It's so great to see you. Please introduce yourself and SpinjIns to our audience here. Sure. Thanks, Caroline. Hello. I'm Jesse, CEO of Spinjens. Spinjens is the leading AI solution provider, especially for manufacturers. We believe the latest technology can make a massive difference to the manufacturing industry, and our mission is to serve as the bridge between innovative technology and our manufacturing customers while guiding them toward a transformative future. As we all know, the AI stand at the forefront of tech of technology and cosmetic inspection applications are one of the connecting points between new tech and the manufacturers Since our expertise is in AI and in cosmetic inspection, we have built strong partnership with the industry leaders such as NVIDIA and Advantech to provide the best AI solutions and services to our customers. Based on the power of NVIDIA GPU technology alongside Advantech AI computing power, Combined with our core AI inspection solution, we have helped many manufacturing customers adopt AI into production line in various industries. So Jesse, SpinGits has many years of experience in AI and cosmetic inspection. Would you please share what your solution can serve, what problems it solves, and what are the benefits that will bring to industrial manufacturers? Okay. We deliver end to end solutions based on AI technologies, including machine vision, motion control, AI detection, and even automation platform, among others. Through our solutions, we can effectively help our customers lower the threshold for AI introduction and solve the problems they face on the on the production lines. Moreover, we can help customers effectively achieve the goal of digital transformation. One of our solution is the AI cosmetic inspection platform, AI Navy, which is made for many with manufacturers. It is one stop platform, which has mature software tools and a comprehensive implementation process that can help customers improve their autonomy in AI applications. We also provide complete educational training and consulting services to assist with model training. From initial data labeling to deployment on the production lines, NAB simplifies the entire process with user friendly guidelines, enabling customer to operate independently. Today, the hundreds of speechless AI assisted are operational in worldwide, conducting tens of thousands of inferences every second around the clock. Thank you so much, Jesse. It really seems like there's close cooperation between Spinjens, NVIDIA, and Advantech to provide solutions for industrial manufacturers. Could you explain a little more about each role and highlight the key ability of this strong ecosystem? That's a good question. BingJ's AI solutions, which are recognized by NVIDIA, have successfully been implemented in a variety of fields. Additionally, our compatible solutions provide users with greater flexibilities. The combination of NVIDIA, Advantech, and ourselves makes a strong team. Advantech provide powerful edge computing hardware. NVIDIA provide a bounded GPU computing resources, which are which can be paired with our software to bring further benefits to customers. We have helped many manufacturing customer adopt AI into production lines in many industries such as semiconductors, basic components, connectors, and more. In addition, I want to share some case studies of our AI solution. The first case is about the inductors in passive components. The item, to be inspected were semi finished inductors, known as CHOCS. The customer had been looking for a solution to inspect wire joints on the chalk for many years. Due to the characteristic of these wire joints, the traditional rule based solutions does not work. They require multi node a multi direction inspection solution up to that point. They have not been able to find a suitable way to accomplish that. That is until they encounter spin jets. We assisted them from the initial configuration of the optical architecture to the setup of the inspection system and then to the training and the adjustment AI, which lead to scaling up the the solutions. As mentioned above, for the particular characteristic of the wire joints, in order to complete examine each side of the chop, we utilize four industrial cameras to capture four images. We configured the MIC seven seventy with two edge computing computers to control the cameras we then successfully enabled factory inspection with one hundred percent air inspection, which also reduced labor cost from one hundred and fifty people to around less than twenty people. The second case study deals with the semiconductors. In the past, the inspection of semiconductor flannels was done by manual inspection of a wafer image, which made it difficult to define the defects in the standardized way. It caused two problems the first is time consuming and the second is inefficiency improving the production lines We successfully reduce the necessary inspection time while also improving the product qualities. In addition, the inspection data can also be digitalized and uploaded to the production repro on a regular basis. In this case, the customer generates the thickness of a wafer into images using specialized software called ADE. Then we used two AI algorithms to identify the location of six types of defects into those images. Since larger models were used, the customer wanted to upgrade the equipment to a higher level. We configured Skye six thousand four hundred server with the Tesla l four t s GPU cards to achieve the upgrade. The last is the case more relevant to our lives Among the types of among the types of manufacturers, the electronic components manufacturing industry is particularly competitive. Products in this industry are often rapidly iterated and produce, in high balance. Therefore, it requires high speed inspection, and the inspection standard are also very stringent. In this case, we are detecting more than type ten types of defects in type c connectors for mobile devices such as scratches and third particles third particles and more. We use industrial cameras with the server and the graphic cards, t x six thousand eight app, for the hardware, to a company that we have customized the software framework that includes not only the preprocessing functions of imaging position and cropping, but also the adjustment of threshold and determination of resultant data. With our AI solution, the inspection can be zero misdetection around ten air models at the same time. Thank you for sharing these use cases and how manufacturers have deployed this technology efficiently, Jesse. Can you explain how the AI Navi software works? AI Navi is divided into two functions, training and inference. The inference function usually works in conjunction with the a o I machine. Depending on the functionality and case demands, different hardware gpu specifications are required. We have shown these a few required hardware components that are better suited to the manufacturing industry, also including NVIDIA Quadro series, RTX six thousand ADA, a four thousand five hundred, and Tesla series NVIDIA L40s graphic card represented by Advantech. Other than AI cosmetic inspection, SpringGen also has developed the lightweight AI solution, iCAP, for factories. This all in one solution from image capture to deployment, which can be accomplished in one camera. It is necessary to identify attacks, on product that at the packing stage on the production line. Some factories are accustomed to checking to to checking text by manual inspection, but after installing the IKM with the OCR AI model it can be complete completely replace human labor and save a lot of working time. To cope with the diverse needs of different industries, SpinGen provides a complete and scalable solutions. Customer can choose the appropriate solutions configuration according to their need with NVIDIA, Advantech, and Spinjens. We can be confident that the super solution can be tailored for the most, of production challenges in the manufacturing industry. Thank you so much, Jesse. It's a pleasure having you with us today.