Yokoyama Shokai Co.,Ltd Product Lineup

Previous false detections by inspection machines: - False detections due to changes in angle or orientation - False detections due to irregular shapes - False detections due to color changes - False detections due to mixing of multiple types Therefore, - Aligning items to make them easier for the inspection machine to recognize - Inability to automate what was previously done visually - Inspection machines were purchased but cannot be used 1. Factories and Manufacturing - AI counting on production lines - Automatic measurement of part quantities - Automation and labor reduction from visual inspections 2. Logistics - Product checks with AI counting - AI counting for palletized goods - Remote monitoring and labor reduction 3. Food Industry - AI counting for quantity checks - Replacement of ambiguous weight measurements - Prevention of product absence and mistake-proofing - Prevention of human errors ★ Image testing can be done free of charge, so please feel free to contact us. ★ We can also provide suggestions based on the lighting environment if you send us your work. ★ Automation and equipment inquiries are welcome. ★ We also assist with solving other inspection-related issues.

Here is an example of inspecting the number of transparent zip bags. "Deep Counter" is an AI software that uses Deep Learning technology to achieve high-precision counting of unspecified items and objects placed in various orientations, which was difficult to accomplish with traditional rule-based image recognition technology. It counts quickly and accurately using Rist's unique AI algorithm, contributing to the reduction of manual counting tasks done by "visual inspection" on-site. 1. Improved accuracy: Prevents counting errors and oversights from manual work. 2. Time savings: Speeds up the process with cameras, improving work efficiency. 3. Alleviation of labor shortages: Reduces workload and addresses labor shortages. 4. Cost reduction: Cuts labor costs, reduces losses due to errors, and minimizes losses. 5. Data utilization: Enables real-time recording and monitoring, and can be used for data analysis.

Count management, count checking, count operations, count inspection, count examination Automating counting tasks through "visual inspection" with 'Deep Counter' Free verification available! Introduction of case studies! 1. Factories and Manufacturing - AI counting on production lines - Automatic measurement of parts quantity - Automation and labor reduction from visual inspection 2. Logistics - Product checking with AI counting - Pallet cargo AI counting - Remote monitoring and labor reduction 3. Food - Count checking with AI counting - Replacement of ambiguous weight measurement - Prevention of product absence and mistake-proofing - Prevention of human error Automation of counting tasks that were previously impossible is now possible! - It's okay if items are overlapping - It's okay if visibility changes with angles - Counting is possible with various shapes - Counting is possible with multiple objects ★ Easy teaching with about 10 training images!

AI image processing software for manufacturing sites "gLupe" - No knowledge of image processing is required. - Easy learning method that anyone can use. - It's okay if there are few defective images. - New defects can easily improve accuracy with additional learning. - Preliminary tests are free. No PoC required. Can start on a low budget. - One-time purchase for peace of mind. No running costs. Of course, there are areas where AI image recognition may struggle, but if there are things that traditional images couldn't do, things that didn't work well with other companies' inspection machines, or if you simply want to automate from manual inspection, please feel free to consult us. Let's TRY first!! - Detection of scratches and cracks (surface defects on metals, plastics, glass, etc.) - Detection of dirt and stains (defects in food, packaging, printed materials, etc.) - Foreign object contamination (in food, pharmaceuticals, cosmetics manufacturing lines) - Product size (detection of protrusions and chips) - Detection of missing or deformed parts (such as plastic and metal components) - Detection of bending and warping (such as circuit boards, films, panels) - Color unevenness and discoloration (such as painting and food cooking levels) - Classification of non-standard products (sorting good and defective products) - Product identification and classification (confirmation of mixed varieties, etc.)

In the manufacturing industry, "inspection operations" are essential for ensuring product quality. It is a crucial task to prevent the accidental shipment of defective products, but it does not directly contribute to profits. However, once defective products are shipped, it can lead to trust issues for the company. We also solve the challenges of transitioning from manual visual inspections to automated inspections using cameras! > We will implement it with minimal changes to current procedures and work rules. > Inspections are conducted without stopping, in line with the speed of the operators. > We distinguish not only the items but also the people involved. > Count before stacking products. Reduction of inspectors Reduction of human errors Increased work efficiency Ensuring traceability We will gather specific data on the challenges on-site and detailed opinions from employees. By clarifying the issues, we will propose the necessity and utilization of AI, as well as the effects of its implementation. We will also assist in modifying existing systems to integrate with AI. ★ Unlike conventional Deep Learning, we are developing a unique reinforcement learning AI that incorporates the engineering technology of "self-growing" biomimicry. Please feel free to consult us about inspection themes you may have previously given up on!

Free tests are being conducted at any time! We will assist you with imaging through work borrowing! We will propose the optimal imaging environment! Challenges of AI Implementation 1. Large number of samples Typically, a large number of samples are required for AI inspection implementation. → With 'gLupe', you can start testing with just one defective sample at a minimum. You can use multiple defects within a single image for learning and evaluation. 2. Unclear parameters and need for specialized knowledge We often hear feedback like, "I purchased AI inspection software, but it's difficult to use and I'm not achieving good accuracy." → With 'gLupe', anyone can use it with simple operations. Just roughly mark the defective areas and other locations. No complex parameter settings are needed. Even without specialized knowledge, anyone can build an inspection AI! 3. Time and cost involved Not only the initial investment but also the operation requires a tremendous amount of time and cost. → With 'gLupe', in-house production is easy! While manufacturer support is robust, the majority of our customers create the software themselves.

The OPPX series is an LED lighting controller that supports 12V and 24V inputs, allowing connection not only to our own lighting products but also to those from other manufacturers. It accommodates three dimming methods: PWM, constant voltage, and strobe overdrive, enabling the unification of controllers even when using different types of lighting, thus achieving reduced wiring and space-saving. ◆ Monitoring Function It measures the brightness of the lighting using a built-in photodiode and allows reference to brightness and temperature measurements on the controller. ◆ Feedback Control If brightness decreases due to voltage drop from extension cables or degradation of LEDs, the output voltage is automatically corrected to maintain the brightness level from the time of factory shipment. ◆ Identification of Individual Lights Data such as the cumulative operating time, which varies for each unit, can be referenced and used for predictive maintenance. ◆ Monitor Brightness Alarm An alarm output for brightness reduction can be triggered against a pre-set threshold. ◆ Absolute Brightness Dimming By setting the absolute brightness values to be the same among the lights, it is possible to achieve uniform brightness without individual variation.

Original Algorithm (ABHB) ▶ Inspections that were not possible with conventional image processing cameras or AI deep learning can now be performed!! ▶ No need for cumbersome program settings for each variety!! ▶ Quantitative color inspection is also possible!! 【Examples】 ■ Ingredient counting in food manufacturing processes ■ Detection of fine scratches and dirt on automotive parts ■ Motion analysis of small fish models ■ Verification of the presence of semiconductor-related components and solder inspection ■ Equipment inspection for railway companies ■ Color inspection of wall materials, flooring materials, etc. ■ Validation in the chemical industry, etc. Please consult us about things that other companies could not do! Please consult us about things you have given up on until now! 【Implementation Process】 ■ Hearing about the issues and image verification ■ Appropriate proposals for actual operation ■ Support for implementation on the production line

An algorithm called "ABHB" that detects defective products through image recognition, distinguishing defects in metal processing products such as casting defects from "contaminants" and "other defects." With "PhotoNavi - Visual-less," it is possible to detect areas that meet specific conditions within images and automatically identify similar colors, shapes, and patterns instantaneously, using an algorithm that closely resembles the sensory judgment made by humans. Inspection of casting defects Inspection of edges ◆ Works with a small number of sample images ◆ Usable on a general PC ◆ High-speed recognition depending on the software language ◆ Recognition accuracy of over 99% 　* Proven effective in inspections at factory line speeds of up to 40 meters per minute.

ABHB (Algorithm Based on Human Brain) Quantifies RGB values into hue, saturation, and brightness that are closer to human visual perception in a dual-cone-shaped 3D color space. ▼▼ Solving challenges that even color measurement devices struggle with!! ▼▼▼ Addressing issues with visual color inspection! - Inspection of materials used over a wide range - Applications in fields where color deviation is not permitted - Detection of subtle color changes - Detection of various color unevenness - Transitioning from sampling inspection to full inspection - For work shapes that cannot be measured with instruments Video and mobile inspections will also be possible!!

Target product: i-PRO mini Application: State Change Detection Application This AI edge camera can be utilized not only as a surveillance camera but also as a sensor. Detection that was previously difficult may now be implemented at a low cost. Although this is an experiment conducted on a desk, we hope you can envision the actual usage in the field. You can test whether detection is possible or not, so please feel free to consult with us. We look forward to new ways of using it.

Our company is developing a business themed around "small starts," which involves automating part of the production process with collaborative robots. While utilizing robots is a response to labor shortages, the public's perception of robots as "working 24 hours" and "capable of anything" often leads to the desire to have robots perform every task in the workflow. However, automating with this mindset can lead to issues such as: - Enormous costs that do not justify the return on investment, - Lack of installation space. These problems can result in a situation where automation does not progress. It is important to clearly define the purpose of automation as either "labor saving," "quantification," or "reduction of labor," and to start with simple processes. Therefore, we propose our original product, the Techman 360 AI visual inspection system, which allows for affordable automation installation with teaching-less and convenient setup. We aim to achieve "quantification," "quality improvement," and "labor saving" in the inspection process.