Wafer box-packed shipment detection equipment, method and application

Abstract

The invention discloses wafer box shipment detection equipment and method and application. The device comprises a positioning module, a reflection module, an image acquisition module and an image processing module. The reflection module comprises a long-strip-shaped reflection mirror and a plurality of positioning marks, the length direction of the long-strip-shaped reflection mirror is parallel to the axial direction of the wafer, and the positioning marks are in one-to-one correspondence with the containing positions of the wafer; the image acquisition module at least shoots a reflection image of the wafer in the reflector and the positioning mark to form a detection image; the image processing module comprises an image recognition model and is used for recognizing whether the actual containing position of the wafer is matched with the target containing position or not. The inspection image of the wafer placed in the wafer box is obtained through specular reflection imaging, stray interference caused by perpendicular incidence of lamplight is avoided, meanwhile, the positioning mark arranged on the reflection face is used for assisting the image recognition model to recognize the wafer graph, and the recognition speed and accuracy are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wafer packaging and testing, and particularly to a wafer boxed shipping detection device, method and application. Background Art

[0002] In the process of wafer testing and production, there are scenarios of batching, where some wafers are separated from the mother batch into sub-batches for individual shipment. During batching, problems such as incorrect wafer quantities or incorrect extraction of wafers with different lot numbers may occur, and currently, effective interception cannot be achieved through the management system. In addition, another problem faced in the industry currently is that after the completion of wafer testing and production and before shipment, it is necessary to place drying packets and then evacuate the air before shipment, which may result in problems such as incorrect quantities of drying packets or incorrect use of vacuum bags, and may also lead to certain quality accidents.

[0003] In response to the above situations, as Figure 1 shown, the common practice in the industry currently is that after the operator of the shipping packaging process enters the station in the MES (manufacturing execution system) account material system (hereinafter referred to as MES), they manually visually inspect the quantity, lot number, and packaging materials of the products according to the document specifications to determine whether they are consistent with the documents. If they are consistent, the MES system will perform out-of-station operations; if they are inconsistent, the operation will be detained and stopped. The judgment of the consistency of the shipping quantity, lot number, and packaging materials with the specifications completely relies on manual visual inspection, which has low efficiency and low accuracy.

[0004] Moreover, for wafer shipping, since the surface of the wafer box is reflective and the lighting in the wafer factory is relatively bright, a lot of stray interference will be generated, resulting in deviations in the detection of the wafer position inside the wafer box whether by manual recognition or using intelligent recognition methods, which may cause production accidents. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the purpose of the present invention is to provide a wafer boxed shipping detection device, method and application.

[0006] To achieve the foregoing invention purpose, the technical solutions adopted by the present invention include: In the first aspect, the present invention provides a wafer boxed shipping detection device for detecting whether the contents in the wafer box are correctly placed. The contents include wafers. The wafer boxed shipping detection device includes: a positioning module, a reflection module, an image acquisition module, and an image processing module; The positioning module is used to position the posture of the wafer box; the reflection module includes a long strip-shaped mirror disposed beside the wafer box when the wafer box is positioned and a plurality of positioning marks disposed on the reflection surface of the long strip-shaped mirror. The length direction of the long strip-shaped mirror is parallel to the axial direction of the wafer, and the positioning marks correspond one-to-one with the accommodating positions of the wafers; The image acquisition module captures at least the reflected image of the wafer in the mirror and the positioning mark to form an inspection image; the image processing module includes an image recognition model, and the image recognition model is used to identify whether the actual accommodation position of the wafer matches the target accommodation position based on the inspection image.

[0007] In a second aspect, the present invention further provides a wafer boxed shipment detection method based on the above wafer boxed shipment detection device, which includes: Providing a plurality of training wafer boxes, in which a number of wafers are arranged in a training layout state, and using the training layout state as the ground truth label; Placing the training wafer box on the positioning module and obtaining a training image by using the image acquisition module; Using the training image and the ground truth label as a training set to update the parameters of the initial image recognition model to obtain an image recognition model; Placing the wafer box to be detected on the positioning module and obtaining an inspection image; Using the image recognition model to identify whether the actual accommodation position of the wafer matches the target accommodation position based on the inspection image.

[0008] In a third aspect, the present invention further provides an application of the above wafer boxed shipment detection method in the field of wafer packaging detection.

[0009] Based on the above technical solutions, compared with the prior art, the beneficial effects of the present invention at least include: The shipment detection device and method provided by the present invention use specular reflection imaging to obtain the inspection image of the wafer placed in the wafer box, avoiding the stray interference caused by direct light, and at the same time using the positioning mark set on the reflecting surface to assist the image recognition model to identify the wafer pattern, significantly improving the recognition speed and accuracy, replacing manual visual inspection for shipment detection, and ultimately being able to significantly improve the accuracy and efficiency of wafer boxed shipment detection, with broad application prospects.

[0010] The above description is only an overview of the technical solutions of the present invention. In order to enable those skilled in the art to understand the technical means of the present application more clearly and to implement it according to the content of the specification, the following takes the preferred embodiments of the present invention and combines with detailed drawings to illustrate as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 is a schematic flow chart of manual visual inspection for shipment detection provided by the background art of the present invention; Figure 2 is a schematic structural diagram of a wafer boxed shipment detection device provided by a typical embodiment of the present invention; Figure 3It is an exemplary inspection image obtained by the wafer cassette shipping detection device provided by a typical embodiment of the present invention; Figure 4 It is a schematic flowchart of the wafer cassette shipping detection method provided by a typical embodiment of the present invention. Detailed implementation manners

[0012] In view of the deficiencies in the prior art, the inventors of this case have, through long-term research and a large number of practices, been able to propose the technical solution of the present invention. The following will further explain the technical solution, its implementation process, principles, etc.

[0013] In the following description, many specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited by the specific embodiments disclosed below.

[0014] Moreover, relative terms such as "first" and "second" are only used to distinguish one component or method step with the same name from another, and do not necessarily require or imply any actual relationship or order between these components or method steps.

[0015] As Figure 2 shown, an embodiment of the present invention provides a wafer cassette shipping detection device for detecting whether the contents in the wafer cassette are correctly placed. The contents include wafers. The wafer cassette shipping detection device includes: a positioning module, a reflection module, an image acquisition module, and an image processing module; the positioning module is used to position the posture of the wafer cassette; the reflection module includes a long strip-shaped reflecting mirror disposed beside the wafer cassette when the wafer cassette is positioned, and a plurality of positioning marks disposed on the reflecting surface of the long strip-shaped reflecting mirror. The length direction of the long strip-shaped reflecting mirror is parallel to the axial direction of the wafer, and the positioning marks correspond one-to-one to the accommodation positions of the wafers; the image acquisition module at least captures the reflection image of the wafer in the reflecting mirror and the positioning marks to form an inspection image; the image processing module includes an image recognition model, and the image recognition model is used to identify whether the actual accommodation position of the wafer matches the target accommodation position based on the inspection image.

[0016] As a typical example, continue to refer to Figure 2 shown, the positioning module may include a limiting structure disposed under the wafer cassette, usually some protrusions corresponding to the bottom slots of the wafer cassette, which are generally common structures in the semiconductor field, but are not limited thereto. Any mechanical structure capable of positioning the wafer cassette can be used; the reflection module in this example is Figure 2The mirror in it is strip-shaped and installed near the back of the wafer cassette (i.e., the side facing away from the opening of the wafer cassette), with the mirror surface facing the wafer cassette. Of course, it can be replaced with other components with reflection functions as shown below; the image acquisition module is Figure 2 the industrial camera used in Figure 2 . However, even if it is replaced with an ordinary electronic camera, corresponding images can still be obtained, but its stability is not as good as that of a professional industrial camera. The present invention does not specifically limit the specific selection of these modules.

[0017] When solving the problems of low efficiency and low accuracy of manual visual inspection, the inventor of the present invention first considered a technical solution of directly taking pictures of the wafers inside the transparent wafer cassette through the wafer cassette to perform AI image recognition, so as to determine whether the wafers in the wafer cassette are correctly placed. However, it was found that there are still some insurmountable problems in this implementation method during specific applications. After analysis, the main reason should be that the surface of the wafer cassette has a certain degree of reflectivity, and the lights in the semiconductor factory are relatively bright, and the movement of personnel and equipment is also relatively frequent, which easily leads to light interference around the detection equipment, thus interfering with the accuracy of image recognition and resulting in misjudgment. To solve the above problems, the present invention first adopts the method of imaging with a strip-shaped reflector, so that the side edges of the wafers stacked vertically (the edges of the wafers are the most obvious in the image) are imaged in the image acquisition module through the reflector. The strip-shaped reflector limits the incident angle of light and largely avoids the interference brought by external light. In addition, the present invention also combines the method of reflection marks, uses positioning marks to strongly indicate the positions of the wafers in the image, and introduces additional prompt information for the training and recognition of the image recognition model, thereby significantly improving the training and recognition efficiency and accuracy of the image recognition model. Through various technical means, the wafer cassette shipping detection device and the corresponding method provided by the present invention achieve almost error-free detection of the in-position situation of the wafers in the wafer cassette, greatly improving the efficiency and accuracy of the wafer cassette shipping detection process.

[0018] In the technical solution of the present invention, the length of the strip-shaped mirror is generally equal to or greater than the height of the space for accommodating wafers in the wafer cassette, and its width can be set by itself, for example, 3-5 cm or a similar size is acceptable, as long as it can play the role of line-of-sight limitation and stray light avoidance. Its light reflection principle can adopt the coating reflection principle or the total reflection principle, and no excessive restrictions are imposed on this; at the same time, the image acquisition module can be a series of optoelectronic devices such as industrial cameras that can obtain images sufficient for recognition; in addition, for the image processing module and the image recognition model it contains, various existing pre-trained image recognition models can be used. Specifically, in the embodiment of the present invention, the image recognition model is obtained after training with a purchased image recognition software product, and different types of wafer cassette samples are used during the training process (for different types of wafer cassettes, the accommodation positions of the wafers and the corresponding positioning marks are different); the role of the positioning mark is to introduce a prominent prompt in the inspection image and enhance the sensitivity of the model to the edge pattern and color of the wafer. It can be numbers, graphics, letters, and their combinations. For example, it can be numbers marked in sequence, or arrows pointing to different slots. Preferably, it is a combination of alphanumeric and arrows, such as the arrangement of Slot1→, Slot2→... Slot25→.

[0019] In order to further improve the prompt intensity, in the preferred embodiment of the present invention, color differences are also used to make the prompt more prominent, that is, in some embodiments, the reflective surface is divided into a differently colored area and a primary colored area, both of which are rectangular. For a specific exemplary division, reference can be made to Figure 3 As shown, the dividing line between the differently colored area (such as the left half green area shown in Figure 3 and the primary colored area (such as the right half transparent area shown in Figure 3 is parallel to the length direction of the strip-shaped mirror. The color of the differently colored area is optional, and the color of the primary colored area remains the original color of the mirror surface, that is, a transparent color, generally colorless and transparent or bright silver. The color of the differently colored area is a contrast color to the background color of the wafer cassette. For example, usually the color of the wafer cassette is transparent, gray or brown, and the ambient light color is usually white or yellow. Therefore, the color of the differently colored area can usually be selected as a more distinct color, such as green, red, etc.; from the shooting perspective of the image acquisition module, the reflected image is at least imaged in the primary colored area; the positioning mark is set in the differently colored area, and the color of the positioning mark is a contrast color to the color of the differently colored area. For example, when the color of the differently colored area is green, the color of the positioning mark can be designed as red, black, etc., which form an obvious contrast with green.

[0020] Specifically, in some embodiments, the color of the differently colored area includes any one of green, blue, red, and yellow, and is not limited thereto. Usually, a more distinct color is selected to distinguish it from the background color.

[0021] In addition, an embodiment of the present invention further provides a technical solution for further eliminating the interference of stray light by using the difference of symmetric shooting. That is, in some embodiments, the image acquisition module includes a first camera and a second camera, the first camera and the second camera are symmetrically arranged on both sides with the long strip-shaped mirror as the central axis, and the shooting perspectives of the first camera and the second camera both point to the long strip-shaped mirror; when acquiring the inspection image, the first camera obtains a first image, and the second camera obtains a second image, and the difference image between the first image and the second image is used as the inspection image, and the difference image removes the different parts in the first image and the second image.

[0022] The above examples the implementation manner in which multiple cameras share a long strip-shaped mirror. In other embodiments, the number of long strip-shaped mirrors can also be multiple, and the multiple long strip-shaped mirrors are equally spaced, and one long strip-shaped mirror corresponds to one camera. For example, the perspective of the first camera points to one of the long strip-shaped mirrors to obtain a first image, and the perspective of the second camera points to another long strip-shaped mirror to obtain a second image, so as to obtain the inspection image, which can further improve the accuracy of shipping inspection.

[0023] In a specific implementation manner, taking an RGB color image as an example, the first image and the second image can be obtained first, and then the first image and the second image are adjusted by a image processing program such as scaling and rotation to align the positioning marks of the two, and then the pixel difference value between the two aligned images is calculated according to the RGB value of each pixel. When the pixel difference value is above a preset threshold (that is, the difference value of a certain area in the two pictures is too large), the pixel point is excluded, and only the partial pixel points with the pixel difference value less than the preset threshold are retained. In this way, the difference processing of the two symmetrically shot images can further eliminate the stray interference brought by external light and further improve the recognition accuracy.

[0024] Moreover, in response to another technical problem mentioned in the above background technology, that is, when performing boxed shipping packaging, it is often easy to use the wrong packaging material or there is an omission in the placement of accessories such as desiccants. Therefore, a more preferred embodiment of the present invention further proposes a technical solution for further detecting whether the shipping packaging is qualified by weighing. That is, in some embodiments, the wafer box is wrapped with a packaging body, the packaging body contains accessories, the accessories at least include desiccants, and the wafer box shipping detection device further includes a weighing module, and the weighing module is arranged at the positioning module for weighing the total mass of the wafer box, the packaging body and the accessories.

[0025] The ideal quality of the appendages, the ideal quality of the packaging materials, and the ideal quality of the wafer cassette and several wafers therein are stored in the account and material system. Therefore, by weighing, it is first possible to determine whether the total quality is correct, and at the same time, it is also possible to judge which item has an abnormality based on the magnitude of the difference between the actual quality and the total quality. For example, if the magnitude of the difference is exactly equal to the quality of one wafer or its multiple, it can be recognized that the number of wafers is abnormal; and if the difference value is very close to the quality of the desiccant, it can be considered that there is a problem with the placement of the desiccant, and so on.

[0026] Corresponding to the composition and structure of the above device, the second aspect of the embodiments of the present invention further provides a wafer cassette shipping detection method based on the wafer cassette shipping detection device provided in any of the above embodiments, which includes the following steps: Provide a plurality of training wafer cassettes, in which several wafers are arranged in a training layout state, and use the training layout state as the ground truth label; Place the training wafer cassette on the positioning module and obtain a training image using the image acquisition module; Use the training image and the ground truth label as a training set to update the parameters of the initial image recognition model to obtain an image recognition model; Place the wafer cassette to be detected on the positioning module and obtain an inspection image; Use the image recognition model to identify whether the actual placement position of the wafers in the inspection image matches the target placement position.

[0027] In some embodiments, the wafer cassette shipping detection method may specifically include the following process: Use the first camera and the second camera to obtain a first image and a second image with symmetric perspectives centered on the long strip mirror; Perform image difference processing on the first image and the second image to obtain a difference image, which removes the different parts in the first image and the second image; During training, use the difference image as the training image to update the parameters, and during detection, use the difference image as the inspection image for image recognition.

[0028] In addition, in combination with weighing measurement, in some embodiments, the wafer cassette shipping detection method may further include the following steps: Calculate the target quality based on the theoretical number of wafers in the wafer cassette, the theoretical quality of the wafer cassette, the theoretical quality of the packaging body, and the theoretical quality of the appendages; Weigh the actual total quality of the wafer cassette, the packaging body, and the appendages; Based on the matching between the actual total mass and the target mass, determine whether the number of wafers in the wafer cassette, the selection of the package, and the presence of accessories are normal.

[0029] For the shipping management of wafer cassettes, in some embodiments, the wafer cassette shipping detection method may further include the following steps: When any one of the actual accommodation position of the wafer, the number of wafers in the wafer cassette, the selection of the package, and the presence of accessories is abnormal, mark the wafer cassette to be detected as in a detained state.

[0030] As a typical example of the above technical solution, the flow of the specific wafer cassette shipping detection method can be seen in Figure 4 as shown, mainly including: 1. Implement the device AI recognition function: Obtain the number of wafer products and the Slot number and automatically compare them with the MES system, and store the comparison results; 2. Implement the device weighing function: The MES system calculates the mass difference data before and after packaging, outputs the mass of the packaging material and compares it with the set packaging material mass in the MES system, and stores the comparison results; 3. Implement the device automatic shutdown and alarm function: Automatically alarm after comparing with the MES system and getting FAIL (abnormal); 4. MES system control: The MES system grabs device comparison information at the in and out stations. If it is not Pass (passed), the MES system will forcibly detain the wafer cassette; 5. MES system backup function: Implement the function of backing up key data in the database of the MES system.

[0031] The third aspect of the embodiments of the present invention also provides an application of the wafer cassette shipping detection method provided in any of the above embodiments in the field of wafer packaging detection.

[0032] The technical solutions of the present invention will be further described in detail below through several embodiments in conjunction with the accompanying drawings. However, the selected embodiments are only used to illustrate the present invention and do not limit the scope of the present invention. Example 1

[0033] In this embodiment, a device is designed to detect the number of wafers, the Slot number, and the packaging material before shipping and compare them with the MES system. If it does not pass, it cannot be shipped, so as to control the problems of material mixing and inconsistent packaging requirements in the final shipping process of wafer testing.

[0034] The specific operation process of the equipment is: the product is placed on the equipment platform, the image recognition model of the equipment is used to identify the quantity and piece number of the product, the quantity and piece number of the product are output and compared with MES for consistency, if inconsistent, the device will trigger an alarm, and if consistent, the result will be automatically saved; the equipment weighing module is used to extract the physical mass before packaging (T1) and the mass after packaging (T2), and automatically output the packaging material mass (T2-T1), and compare it with the quality standards stored in the background of the equipment system for consistency, if inconsistent, the device will alarm, and if consistent, the result will be automatically saved.

[0035] The MES system retrieves the consistency comparison results stored in the equipment when entering and exiting the station. If the stored result is PASS, the station can exit normally in the MES system. If the stored result is not PASS, the station cannot exit and will be forcibly detained.

[0036] Extraction of wafer product quantity and wafer number: A product placement positioning module is added to the equipment platform to ensure that the product placement direction and angle are exactly the same each time. The equipment uses two high-resolution industrial cameras to ensure image clarity. The embedded AI computing module is used to establish an AI model to collect a large number of product images, covering different lighting conditions and backgrounds, and annotate the products in the image. The annotation content includes images of different slots from Slot 1 to Slot25 (the spacing between Slot 1 to Slot25 is exactly the same and gradually increases). A large number of annotations are used to strengthen the training model and optimize the recognition accuracy. The number of products is obtained through the number of slots identified, and the slot number is obtained through the number of slots identified. Then, the AI module is used to parse and output the number of products and the wafer number.

[0037] Specifically, a long strip of reflective mirror is placed near the back of the wafer box (i.e., the side facing away from the opening of the wafer box), facing the wafer box. Two high-precision cameras shoot the wafer placement inside the wafer box from the left and right sides of the wafer box through mirror reflection. The obtained image is differentially processed to retain some similar pixels and remove pixels with obvious differences (the specific method is to use the pixel RGB three-channel algorithm of the image recognition software to set a brightness threshold. If the brightness is higher than the pixel threshold, the pixel difference is considered to be too large). It is best to set the Slot ID number on the mirror and also set some triangles or arrow shapes to indicate the slot position of the wafer. The wafer on the imaging mirror corresponds one-to-one to the Slot ID mark, which helps to improve the AI image recognition capability of the recognition model.

[0038] In addition, one side of the mirror surface of the long strip reflector is painted green. The Slot ID number is set in the painted area. Other colors that are significantly different from the painted color are selected to increase the contrast between the wafer and the background color, further helping to improve AI image recognition capabilities.

[0039] In addition, the wafer box shipping detection system provided in this embodiment also measures mass data, theoretical number of wafers, and theoretical outer box mass through an electronic scale, calculates theoretical data, and performs weighted processing on the AI-recognized quantity to improve the AI accuracy.

[0040] In this embodiment, the above-mentioned equipment and method were used to conduct a six-month trial verification on the No. 1 shipping detection equipment in the production line, and it was found that the customer complaints regarding the number of wafers, Slot numbers, packaging materials, and internal desiccants in wafer box shipping dropped to zero. Embodiment 2

[0041] This embodiment is generally the same as Embodiment 1, and the main difference lies in: One of the two symmetric high-precision cameras is retained, and the images captured by this camera are used for training and shipping detection.

[0042] During the above six-month period, parallel verification was carried out on the No. 2 shipping detection equipment in the production line, and it was found that for the products shipped from this machine, there were still a small number of customer complaints regarding the number of wafers and Slot numbers, but significantly compared to the customer complaint frequency of manual visual inspection, it decreased significantly. Embodiment 3

[0043] This embodiment is generally the same as Embodiment 1, and the main difference lies in: One side of the long strip-shaped mirror was not painted.

[0044] During the above six-month period, parallel verification was carried out on the No. 3 shipping detection equipment in the production line, and it was found that for the products shipped from this machine, there were still a small number of customer complaints regarding the number of wafers and Slot numbers, but significantly compared to the customer complaint frequency of manual visual inspection, it decreased significantly. Comparative Example 1

[0045] This comparative example is generally the same as Embodiment 1, and the main difference is that instead of using a mirror, a single high-precision camera is directly used to capture the image of the wafers placed inside the wafer box for training and detection.

[0046] This implementation method was the technical solution adopted during the early R & D process of the present invention. The customer complaints regarding the number of wafers and Slot numbers were improved to a certain extent compared to manual visual inspection, but the improvement amplitude was limited, and the customer complaint occurrence frequency was about three times that of Embodiment 2 and 3.

[0047] Based on the above embodiments and comparative examples, it can be clearly seen that the shipping detection device and method provided by the embodiments of the present invention use specular reflection imaging to obtain inspection images of the placement of wafers in the wafer cassette, avoiding stray interference caused by direct light irradiation. At the same time, positioning marks provided on the reflecting surface are used to assist the image recognition model to identify wafer patterns, significantly improving the recognition speed and accuracy, replacing manual visual inspection for shipping detection, and ultimately being able to significantly improve the accuracy and efficiency of wafer cassette shipping detection, with broad application prospects.

[0048] It should be understood that the above embodiments are only used to illustrate the technical concept and characteristics of the present invention, and their purpose is to enable those familiar with this technology to understand the content of the present invention and implement it accordingly, and cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit and essence of the present invention should be covered within the protection scope of the present invention.

Claims

1. A wafer cassette shipping detection device for detecting whether the contents in the wafer cassette are correctly placed, the contents including wafers, characterized in that, The wafer cassette shipping detection device includes: a positioning module, a reflection module, an image acquisition module, and an image processing module; The positioning module is used to position the posture of the wafer cassette; the reflection module includes a strip-shaped mirror disposed beside the wafer cassette when the wafer cassette is positioned, and a plurality of positioning marks disposed on the reflection surface of the strip-shaped mirror. The length direction of the strip-shaped mirror is parallel to the axial direction of the wafer, and the positioning marks correspond one-to-one with the accommodation positions of the wafers; The image acquisition module at least captures the reflection image of the wafer in the mirror and the positioning marks to form an inspection image; the image processing module includes an image recognition model, and the image recognition model is used to identify whether the actual accommodation position of the wafer matches the target accommodation position based on the inspection image.

2. The wafer cassette shipping detection device according to claim 1, wherein The reflection surface is divided into a differently colored area and a primary colored area that are both rectangular. The boundary line between the differently colored area and the primary colored area is parallel to the length direction of the strip-shaped mirror, and the color of the differently colored area is a contrast color to the background color of the wafer cassette; Viewed from the shooting perspective of the image acquisition module, the reflection image is at least imaged in the primary colored area; the positioning marks are disposed in the differently colored area, and the color of the positioning marks is a contrast color to the color of the differently colored area.

3. The wafer cassette shipping detection device according to claim 2, wherein, The color of the differently colored area includes any one of green, blue, red, and yellow.

4. The wafer box shipping detection device according to claim 1 or 2, characterized in that, The image acquisition module includes a first camera and a second camera. The first camera and the second camera are symmetrically arranged on both sides with the strip-shaped mirror as the central axis, and the shooting perspectives of the first camera and the second camera both point to the strip-shaped mirror; When acquiring the inspection image, the first camera obtains a first image, and the second camera obtains a second image. The difference image between the first image and the second image is used as the inspection image, and the difference image removes the different parts in the first image and the second image.

5. The wafer cassette shipping detection device according to claim 1, characterized in that, The wafer cassette is wrapped with a packaging body, and accessories are accommodated in the packaging body. The accessories at least include a desiccant. The wafer cassette shipping detection device further includes a weighing module, and the weighing module is disposed at the positioning module for weighing the total mass of the wafer cassette, the packaging body, and the accessories.

6. A wafer cassette shipping detection method for the wafer cassette shipping detection device according to any one of claims 1-5, characterized in that, Including: Providing a plurality of training wafer cassettes, in which a number of wafers are arranged in a training arrangement state, and using the training arrangement state as a ground truth label; Placing the training wafer cassette on the positioning module and using the image acquisition module to obtain a training image; Using the training image and the ground truth label as a training set to update the parameters of the initial image recognition model to obtain an image recognition model; Placing the wafer cassette to be detected on the positioning module to obtain an inspection image; Using the image recognition model to identify whether the actual accommodation position of the wafer matches the target accommodation position based on the inspection image.

7. The wafer cassette shipping detection method according to claim 6, wherein Specifically including: Using the first camera and the second camera to obtain a first image and a second image from symmetric perspectives with the strip-shaped mirror as the axis; Perform image difference processing on the first image and the second image to obtain a difference image, which removes the different parts in the first image and the second image; During training, use the difference image as the training image to update parameters, and during detection, use the difference image as the test image for image recognition.

8. The wafer cassette shipping detection method according to claim 6, wherein, Further comprising: Calculate the target mass based on the theoretical number of wafers in the wafer cassette, the theoretical mass of the wafer cassette, the theoretical mass of the package, and the theoretical mass of the accessories; Weigh the actual total mass of the wafer cassette, the package, and the accessories; Based on the matching between the actual total mass and the target mass, determine whether the number of wafers in the wafer cassette, the selection of the package, and the presence of accessories are normal.

9. The wafer box shipping detection method according to claim 6, wherein Further comprising: When any one of the actual accommodation position of the wafer, the number of wafers in the wafer cassette, the selection of the package, and the presence of accessories is abnormal, mark the wafer cassette to be detected as in a retained state.

10. Application of the wafer cassette shipping detection method according to any one of claims 6-9 in the field of wafer package detection.

Images