Apply the Classification and Segmentation technology of Solomon SolVision AI image platform to identify defect features. First, use the Classification tool to judge whether the wafer has too many defects and eliminate the defective products that cannot be repaired. Then, use image processing technology to segment the wafer image, and use the Segmentation tool to detect the defects in the image, record their features, coordinates, area and other information, which greatly improves the efficiency of subsequent repairs.

1. Line Scan high-speed detection + AI defect classification
2. Support for front and back side appearance defect detection of Silicon/Glass Wafer
3. Applicable to CIS/IQC/OQC
4. Can be equipped with “8/12” EFEM, supports SECS GEM200/300

1. High-speed automatic defect photography based on KLARF file for分区/分Die/Defect size/Defect type
2. Provides AI real-time defect detection; instant detection and classification, processing speed up to 50 FPS or more
3. Can be equipped with “8/12” EFEM, supports SECS GEM200/300
4. Defect photography and detection of wafer/packaging process QA inspection stations/CP & FT

1. AI real-time defect detection and classification, real-time detection and classification, processing speed can reach more than 50 FPS
2. Can be equipped with “8/12” EFEM, supports SECS GEM200/300
3. Min defect size ≧ 0.3µm
4. Can be equipped with a yield management system

Implementation of Cognex Deep Learning proves effective in solving applications of this nature. Component positioning tools easily handle complex vaccine assembly tests, e.g., items placed in varied directions, overlapping, missing, or containing different combinations of SKUs. After training a deep learning system from multiple angles to recognize assembled components and discriminate between existing and newly introduced ones (even visually alike counterparts), successful identification becomes effortless.

Cognex Deep Learning presents a highly trustworthy medical patch testing solution, hinging upon measurable medication dosages against specified locations. Training occurs through mounting patches above several accepted droplet sizes and shapes against rejected liquid forms, creating classes outlining permissible drops' shape and magnitude. Ultimately, any remaining inconsistencies fall into the removal category.

Solomon combines machine vision and artificial intelligence to create AI learning modules using the Classification function of SolVision AI image platform to judge the precipitation conditions from the image features in the database. Through deep learning technology, it can identify the precipitation conditions of liquids of different colors and accurately distinguish 7 different precipitation patterns, thereby judging the quality of the contents.

1. Resolution and defect detection capability range can reach 1.5um~5um
2. Zero dead angle detection area, can achieve 100% full panel coverage
3. Excellent autonomous image detection technology can support all sizes and different touch image designs
4. Intelligent defect classification function
5. Provide professional and customized services

Cognex Deep Learning can simplify and automate this identification process. The classification tool can learn to identify and classify different tire tread patterns from a set of training images. It then groups tires with rims according to their category, ensuring that the correct components are selected and installed on the vehicle.

1. Full Chinese operation interface
2. Applicable to various shapes of tablets
3. Statistical functions (total inspected/qualified/defective quantities)
4. Defective image storage and classification
5. Ability to adjust settings during inspection
6. Complete rejection mechanism planning (low/medium/high speed, contact/non-contact type)

1. Cognex Deep Learning is ideal for the complex task of finding and picking good nuts that are oriented correctly and placing them on chocolate.
2. The classification tool is trained on a set of images of good and bad nuts. The tool can quickly classify nuts into acceptable and unacceptable categories.
3. The part location tool is trained on a set of images of good nuts that are oriented correctly. It can then locate the nuts as they pass by, regardless of the number of nuts present at the same time.
4. Combining these tools ensures that good nuts are identified and located so that they can be picked and placed on the chocolate by a robotic arm. Manual inspection or other forms of machine vision cannot achieve the same speed and accuracy for this task.

Cognex Deep Learning can solve the orientation problem of liquid filling bags and other flexible containers during packaging and transportation. It can be trained on a set of images of possible containers with different orientation labels. The classification tool can accurately determine the orientation of each bag seen and provide the data to the picking and placing robots, which then pick and place them in the correct orientation in the secondary packaging containers.

Four AVT Prosilica GC1290C cameras are used to inspect raisins and sort them by appearance and size before packing them in boxes.

Cognex Deep Learning can rapidly and easily tackle connector placement location detection challenges thanks to the Assembly Validation Tool. Through training with a set of properly functioning connection and contact images, it becomes proficient in understanding all correct installation changes—even when faced with glare and complex backgrounds post-training. Consequently, the validation tool accepts all suitable components across the entire line, while simultaneously eliminating ones falling outside parameter limits.

Solomon combines machine vision and artificial intelligence to use the Feature Detection tool of the SolVision AI image platform to define the characteristics of the assembly positions of each component in the PCBA layout, and train the AI model with the defined image samples. Through the trained AI model, it can instantly detect abnormal conditions and locations such as missing components or assembly errors.

The system is able to first take pictures of the parts, and then create a model (Golden sample) based on them. Then, it can use this as a reference to identify and align the same parts, and at the same time perform dimensional measurements. The parts that are inspected and measured can be used for assembly; the abnormal ones are first sent back to the parts storage area.

Cognex PatMax technology and color tools can identify designs and stamps, regardless of changes in direction, angle, lighting, and other factors that can affect the appearance of items on the production line. PatMax uses a set of boundary curves to learn the geometric shape of an object and then searches for similar shapes in the image, without relying on specific grayscale values. Damaged or broken products are removed before being packaged or shipped, avoiding costly returns and protecting brand reputation.

Cognex Deep Learning can learn from a set of images of installed spring clamp pliers that are qualified and unqualified. The classification tool can quickly determine whether the clamp is working properly or identify potential problems that need to be corrected before further assembly of the vehicle can proceed.

Solomon combines machine vision and artificial intelligence to use the Segmentation technology of the SolVision AI image platform to locate and annotate scratches, dirt and other abnormalities and defects on the aluminum substrate in the image sample. Through AI deep learning, it can automatically and instantly detect and locate various defects on the aluminum substrate, greatly improving the production efficiency of the production line.

In addition to judging whether the packaging is sealed, in order to find the root cause of the problem, it is necessary to further confirm the type and cause of the incomplete seal. However, because the different types of sealing defects are very different, and the surface of the object is highly reflective, it is not easy to find and classify the defects with either the naked eye or automatic optical inspection (AOI).Combining machine vision and artificial intelligence, Solomon uses the Classification tool of SolVision to define the state of good seal from the image and compare it with multiple defects, including incomplete bottom seal, no seal on both bottom and side, and incomplete seal on both bottom and side. It can instantly detect incompletely sealed packages and classify the defects.

Deep learning streamlines tasks involving automatic positioning, identification, and classification within a single image's multiple characteristics. Depending on various item dimensions, shapes, and surface attributes, the system discerns and groups distinctive features accordingly. Users can train assembly positioning and verification tools to find desired items. Afterward, the picture gets segmented into separate sections, allowing the tool to assess the presence of the required item and validate its kind, regardless of orientation and lighting settings. Furthermore, deep learning finds and identifies flyers inside boxes, averting recalls and assuring patient safety.

1. Resolution and defect detection capability range can reach 3um~5um
2. Detection area can cover both the mask area and the frame area at the same time
3. Inspection technology can support mask products with different image designs and any shape
4. Intelligent defect classification function

Cognex 2D vision systems with edge learning perform pass/fail inspection and reject defective items or packages when detected. The embedded edge learning tool can be trained with just a few images that classify packages as pass (all items present) and fail (one or more items missing). Even under shrink wrap, it can confirm that all bottles and products are present, helping food and beverage manufacturers prevent operational errors and maintain customer satisfaction.

1. High-speed detection + AI defect classification
2. Detection items: Open/short circuits, foreign objects, dirt, scratches in the display area and peripheral Fan-Out area
3. Can support different sizes of substrates/panels according to customer needs

Metal case scratches are very fine. Under normal light, it is difficult for personnel to detect defects visually because the metal material is easy to reflect light, which can easily lead to poor appearance quality problems.By using SolVision's Segmentation technology, a defect defect database is established for the appearance shape of defects, and specific defects are classified, such as obvious defects, fine defects and extremely fine defects. Deep learning is used to identify obvious defects and ignore acceptable minor defects. , Effectively improve the detection accuracy and speed, and ensure that the finished products on the production line can enter the assembly process without any defects.

With simple drag-and-drop modules, you can easily create flowcharts to guide the arm for sorting and packaging.

Using the blob counting tool, the In-Sight 2000 sensor can detect the presence or absence of relevant liquids. When the bottle is underfilled, there will be too few blobs displayed, and when the bottle is overfilled, there will be too many. Bottles that fail the inspection will trigger the rejection function.

Traditionally, rule-based machine vision systems used with automated optical inspection (AOI) systems do not perform well. Detecting potential defects (NG) through deep learning can enhance the reliability of the inspection process. The AOI machine uses Cognex Deep Learning tools to identify potential NG situations and provide those images to the system. The defect detection tool can dynamically capture regions of interest, while the classification tool can categorize different types of defects, distinguishing between defective and acceptable wire bonds. Categorizing defects not only helps identify process issues to avoid costly rework downstream, but can also successfully identify defects at the micron level, improving IC chip yield and lifetime performance.

By using the Segmentation technology of SolVision AI image platform, AI models are trained for the image information of the name, concentration, and capacity on the IV bag body. The image features are learned to quickly identify and classify various types of infusion products.

1. Full Chinese operation interface
2. Quick adjustment page for product fine-tuning
3. Statistical functions (total inspected/qualified/defective quantities)
4. Defective image storage and classification
5. Ability to adjust settings during inspection
6. Complete rejection mechanism planning (low/medium/high speed, contact/non-contact type)
7. Equipment can process up to 1200 pcs/minute at maximum speed

Cognex Deep Learning can easily confirm the presence and absence of straws without damage. It can be trained on a small set of images of undamaged straws, as well as a set of images of various unacceptable missing, damaged, or misplaced straws. The classification tool learns to ignore any background and classifies all images into acceptable or unacceptable states. Since there is no need to identify and define specific defects, the classification process is very fast. It only needs to judge acceptable/unacceptable.

Cognex Deep Learning tools can help verify the difference between OK and NG probe marks, making probe mark detection easier and faster. The software is trained on a wide variety of images, including images showing correct probe marks and images showing unacceptable probe marks. Unacceptable marks can then be classified as "pressure related" or "off center".

1. AI real-time detection on the production line, fast, accurate and time-saving
2. Chips in the tray can be detected for misplacement/tilting/dropping/stacking

1. Can provide images of different magnifications
2. Composite camera head design (supports 1-4 Review Head groups)
3. Can correspond to defect data produced by different AOI equipment
4. Intelligent defect classification function

1. Matrox Imaging's software, hardware components and deep learning systems use deep learning technology to expand the field of image processing. Through systematic construction, it greatly saves cost, time and manpower, and at the same time greatly improves product quality.
2. By collecting more than 8,000 images of parts, manually marking and classifying them as OK and NG, the team used the interactive environment of MIL CoPilot to train and build models, and finally imported the models into Matrox DAX for automatic classification of new images.

Cognex Deep Learning is trained on a series of different images with different angles and rotation directions, and the classification tool can reliably classify painted colors. Then, when making a selection, Cognex Deep Learning will examine the entire image and appropriately weight the various changes, reflections, refractions, granularity and chromaticity within the image to select the best match.

Cognex Deep Learning tools provide an easier way to learn and classify chipping and burr marks, as well as distinguish them from normal cutting marks after the cutting process. The software is easy to train and can identify all chipping and burrs, classify them as acceptable or unacceptable, and ignore normal marks within the tolerance range.

1. The component location tool can be trained on a small set of images of pizzas with the desired number of pepperoni slices or other toppings. It can then identify, locate, and count pepperoni slices, even if they are touching or overlapping, while ignoring variations in cheese and sauce underneath. If the number of toppings is not within the specified limits, the pizza will be rejected.
2. The defect detection tool can be trained on a set of images of acceptable pizzas, even with multiple different toppings. It will then detect any physical contaminants, including toppings that are not suitable for this type of pizza, while ignoring variations in cheese and sauce underneath.
3. The classification tool can be trained on a set of images of all types of pizzas produced in the factory, with labels. It can then distinguish between different variations and classify them so that they can be put into the appropriate packaging and accurately entered into inventory. These tools can be used all together to help manufacturers automate processed food inspection, ensuring that only the highest quality products leave the factory and end up on customers' kitchen tables.

By using the Segmentation technology of SolVision AI image platform, AI models are trained with image samples of bottle cap text and barcodes, and optical character recognition (OCR) is performed. This can accurately identify product information on the outer packaging in the high-speed beverage production line. In addition to detecting products with poor printing, it also greatly enhances the efficiency of traceability management and record retention on the production line.

1. Full Chinese operation interface
2. Quick adjustment page for product fine-tuning
3. Statistical functions (total inspected/qualified/defective quantities)
4. Defective image storage and classification
5. Ability to adjust settings during inspection
6. Complete rejection mechanism planning (low/medium/high speed, contact/non-contact type)
7. Equipment can process up to 1200 pcs/minute at maximum speed

Cognex Deep Learning is ideal for sorting coffee beans because it can quickly and easily identify different colors, sizes, and characteristics of coffee beans. It can be trained on a set of images of each type of coffee bean that is received in the warehouse. Even if different types of beans look very similar, the classification tool can distinguish between different types, while accepting natural variations within each type. The defect detection tool can identify any physical contaminants in a batch of coffee beans before they are sent for blending.

1. Resolution and defect detection capability range can reach 3um~10um
2. Can support LTPS products, and can support peripheral line inspection for COA products
3. Area inspection, and also supports BM area inspection function
4. Excellent autonomous image detection technology can support all sizes, different image designs and any shape of panel products
5. Intelligent defect classification function

1. Built-in special lens and light source
2. Full Chinese operation interface
3. Quick adjustment page for product fine-tuning
4. Statistical functions (total inspected/qualified/defective quantities)
5. Defective image storage and classification
6. Ability to adjust settings during inspection
7. Complete rejection mechanism planning (low/medium/high speed, contact/non-contact type)
8. The equipment can handle up to 1000pcs/minute at the fastest

Using defect detection tools, engineers train the software in supervised mode with a collection of images labeled according to whether ceramic capacitors or electrolytic capacitors belong to the category "pass." During operation, the model captures and differentiates both kinds as belonging to the same type. Subsequently, the classification tool learns each unique capacitor property and accommodates intratype variation. Even if they look similar visually, color and label differences distinguish varying electrolytic capacitors effectively. Meanwhile, Cognex Deep Learning accurately classifies and distinguishes individual capacitors within singular images throughout runtime based on patterns learned during development.

1. Fully Chinese one-to-many operation interface
2. One-to-many screen micro-adjustment quick page
3. Independent statistical functions for each screen (total inspected/qualified/defective quantities)
4. Independent defective image storage and classification for each screen
5. Ability to adjust settings during inspection
6. Complete rejection mechanism planning (low/medium/high speed, contact/non-contact type)
7. Equipment can process up to 800 pcs/minute at maximum speed
8. Separate independent mechanisms for each station

1. Resolution and defect detection capability range can reach 10um~100um
2. Can support pre-cut panel size, post-cut “12~75” panel size and shaped products
3. Detection area can cover both the inner area and the glass cutting edge at the same time
4. Defect classification function

Using Solomon's core AI machine vision system SolVision, an AI model can quickly and accurately identify the direction and angle of each artichoke, using its stem as a reference point, regardless of the size of the stem itself. SolVision also allows AI detection results to be exported through various communication protocols.

1. Cognex AI vision systems and software are effective solutions for meat quality inspection. They can be trained on a set of images of each labeled cut of meat or poultry. The classification tool then accepts a range of natural variations while accurately classifying each cut, ensuring it receives the proper grade so that the product can be sold at the highest reasonable price.
2. Regardless of the source, whether from damaged packaging, machinery, or other sources, the defect detection tool quickly finds any physical contaminants and flags them before the product is shipped. This ensures that all shipped cuts are correctly classified and priced, leading to increased customer satisfaction and revenue.

Laser welding is divided into two categories: laser heat conduction welding and laser deep penetration welding according to the power density. It has different weld seam characteristics. Due to the different welding positions and styles of the products, it is impossible to use traditional optical detection to distinguish the annular distribution of the weld seam, the appearance of the finished product such as no welding, etc., so the production line is finally inspected manually, which often causes the welding quality to be uneven.By using the Classification and Segmentation technology of Solomon SolVision AI image platform, AI models can be trained with weld feature images to identify welding power and no welding defects, and the number and distribution of fish scale patterns of weld seams can be accurately detected through deep learning.

1. Cognex Deep Learning can be trained on the full range of possible engine types and component configurations. The distance between the components and the center distance can cause the wide-field lens to present the engine components at different angles. No matter what angle the components are presented at, the classification tool can learn to identify each component.
2. In order to accurately perform engine assembly verification, the classification tool must learn the engine type and the required locations of different oil filters, wires, hoses and other components, and immediately mark any engines with missing or incorrectly installed components. This allows errors to be identified before the engines are ready to be installed in vehicles.