Rubber product defect detection device and defect detection method thereof

Abstract

The invention discloses a rubber product defect detection device and a defect detection method thereof.The rubber product defect detection device comprises detection sorting equipment and a processor, the detection sorting equipment comprises a loading disc, a belt feeder and a workbench, the workbench comprises a glass disc and multiple sets of detection cameras, and a defect detection model trained in advance is integrated in the processor; a to-be-tested rubber product is placed through the loading tray; the rubber product to be detected is conveyed from the loading tray to the glass tray through the belt feeder; enabling the rubber product to be detected to sequentially pass through a plurality of groups of detection cameras through a glass disc; collecting three views of the rubber product to be detected through a plurality of groups of detection cameras; performing defect detection on the three views through a defect detection model to obtain a defect detection result; wherein the three views comprise a top view, a bottom view and a side view. The device can realize automatic defect detection of the rubber product, reduces the manual operation process, improves the production efficiency and the defect detection accuracy, and can be widely applied to the technical field of automatic detection.

Description

Technical Field

[0001] The present application relates to the field of automated detection technology, and in particular to a rubber product defect detection device and a defect detection method thereof. Background Art

[0002] In the field of quality inspection of traditional rubber products, defect determination mainly relies on manual operation. Although this method is basic, it has significant disadvantages. First, in the face of large-scale production of rubber products, manual inspection requires repeatedly turning over each product, which is not only labor-intensive, but also requires a large amount of human resources, thus greatly increasing production costs and resulting in low production efficiency. Secondly, long-term continuous visual inspection can easily lead to visual fatigue of operators, thus affecting the accuracy and reliability of rubber product inspection, increasing the risk of missed inspections and misjudgments, and ultimately may lead to customer complaints due to product quality issues. Summary of the invention

[0003] In order to solve the above technical problems, the object of the present invention is to provide a rubber product defect detection device and a defect detection method thereof with high efficiency and accuracy.

[0004] To achieve the above-mentioned purpose, one aspect of an embodiment of the present application provides a rubber product defect detection device, comprising a detection and sorting device and a processor, the detection and sorting device comprising a loading tray, a belt feeder and a workbench, the workbench comprising a glass tray and multiple groups of detection cameras, one end of the belt feeder is connected to the loading tray, and the other end of the belt feeder is connected to the glass tray, each group of the detection cameras is located above or below the glass tray, and each group of the detection cameras is also connected to the processor, and a pre-trained defect detection model is integrated in the processor, the loading tray is used to place the rubber product to be tested, the belt feeder is used to transport the rubber product to be tested from the loading tray to the glass tray, and the glass tray is used to make the rubber product to be tested pass through each group of the detection cameras in sequence, each group of the detection cameras is used to collect three views of the rubber product to be tested, and the defect detection model is used to perform defect detection on the three views to obtain defect detection results, wherein the three views include a top view, a bottom view and a side view.

[0005] In some embodiments, the detection and sorting equipment also includes a plurality of sorting boxes, the workbench also includes a plurality of air blowing ports, each of the air blowing ports is located on the glass plate, each of the sorting boxes is located below each of the air blowing ports, each of the air blowing ports is also connected to the processor, and the air blowing ports are used to blow the rubber product to be tested into the corresponding sorting box according to the defect detection result.

[0006] In some embodiments, the detection camera includes a first detection camera, a second detection camera and a third detection camera, the first detection camera is located above the glass plate, the second detection camera is located below the glass plate, and the third detection camera is placed on the glass plate. The first detection camera is used to capture the top view of the rubber product to be tested, the second detection camera is used to capture the bottom view of the rubber product to be tested, and the third detection camera is used to capture the side view of the rubber product to be tested.

[0007] In some embodiments, the belt feeder includes a lifting belt and a feeding belt, one end of the lifting belt is connected to the bottom of the loading tray, the other end of the lifting belt is connected to one end of the feeding belt, and the other end of the feeding belt is connected to the edge of the glass tray, the lifting belt is used to transport the rubber product to be tested from the loading tray to the feeding belt, and the feeding belt is used to transport the rubber product to be tested from the lifting belt to the glass tray.

[0008] In some embodiments, the blowing port includes a first blowing port, a second blowing port, and a third blowing port, the sorting box includes a qualified product box, a waste box, and a box to be processed, the qualified product box is located below the first blowing port, the waste box is located below the second blowing port, and the box to be processed is located below the third blowing port, and the first blowing port, the second blowing port, and the third blowing port are used to blow the rubber product to be tested into the qualified product box, the waste box, or the box to be processed according to the defect detection result.

[0009] To achieve the above object, another aspect of the embodiment of the present application provides a defect detection method of a rubber product defect detection device, comprising the following steps:

[0010] Place the rubber product to be tested on the loading tray;

[0011] The rubber product to be tested is conveyed from the loading tray to the glass tray by a belt feeder;

[0012] The rubber product to be tested is passed through multiple groups of detection cameras in sequence through the glass plate;

[0013] Collect three views of the rubber product to be tested by using multiple groups of the detection cameras, wherein the three views include a top view, a bottom view, and a side view;

[0014] Defect detection is performed on the three views using a defect detection model pre-trained in the processor to obtain a defect detection result.

[0015] In some embodiments, the defect detection method further includes the step of pre-training the defect detection model, wherein the pre-training of the defect detection model specifically includes:

[0016] Acquire a training sample set, wherein the training sample set includes a plurality of top view samples, bottom view samples, side view samples and corresponding detection labels of the rubber product to be tested;

[0017] Determine defect judgment rules, perform multi-stage training on a pre-constructed convolutional neural network according to the training sample set, the detection labels and the defect judgment rules, and obtain the defect detection model.

[0018] In some embodiments, the determining of the defect judgment rule includes performing multi-stage training on a pre-built convolutional neural network according to the training sample set, the detection label, and the defect judgment rule to obtain the defect detection model, specifically including:

[0019] Determining a defect determination rule, and then inputting the defect determination rule into the convolutional neural network;

[0020] Inputting the top view sample and the side view sample into the convolutional neural network for first-stage training, and then randomly selecting the top view sample and the side view sample to test the convolutional neural network after the first-stage training to obtain a first defect detection result;

[0021] The defect determination rule is updated according to the first defect detection result until the first defect detection result reaches a preset first test condition and the top view sample and the side view sample randomly input reach a preset first quantity threshold;

[0022] Inputting the top view sample and the side view sample at the critical point into the convolutional neural network for second-stage training to obtain a second defect detection result;

[0023] The defect determination rule is updated according to the second defect detection result and the detection label until the second defect detection result reaches a preset second test condition and the input top view sample and the side view sample at the critical point reach a preset second quantity threshold;

[0024] Determine the lettering features of the rubber product to be tested, import the lettering features into a preset shielding list, and then input the bottom view sample and the shielding list into the convolutional neural network for third-stage training to obtain a third defect detection result;

[0025] The defect judgment rule is updated according to the third defect detection result and the detection label until the third defect detection result reaches a preset third test condition and the input bottom view sample reaches a preset third quantity threshold, and the training is stopped to obtain the trained defect detection model.

[0026] In some embodiments, the performing defect detection on the three views by using a pre-trained defect detection model in the processor to obtain a defect detection result specifically includes:

[0027] Inputting the top view, the bottom view and the side view into the defect detection model, and then dividing the top view, the bottom view and the side view into regions to obtain a plurality of regions to be tested;

[0028] Performing defect detection on each of the test areas to obtain defect detection results;

[0029] Wherein, the defect detection includes at least one of glue deficiency detection, bubble detection, flower chipping detection, crack detection, inclusion detection, deformation detection or size detection.

[0030] In some embodiments, the defect detection method further includes the step of blowing the rubber product to be tested into a corresponding sorting box through a plurality of air ports according to the defect detection result, the air ports include a first air port, a second air port, and a third air port, the sorting box includes a qualified product box, a waste product box, and a to-be-processed box, and blowing the rubber product to be tested into the corresponding sorting box through the plurality of air ports according to the defect detection result specifically includes:

[0031] When the defect detection result shows that the rubber product to be tested is a qualified product, the rubber product to be tested is blown into the qualified product box through the first blowing port;

[0032] When the defect detection result shows that the rubber product to be tested is waste, the rubber product to be tested is blown into the waste box through the second blowing port;

[0033] When the defect detection result shows that the rubber product to be tested is a product to be processed, the rubber product to be tested is blown into the processing box through the third blowing port.

[0034] The beneficial effects of the present invention are as follows: the rubber product defect detection device and defect detection method of the present invention include a detection and sorting device and a processor, the detection and sorting device includes a loading tray, a belt feeder and a workbench, the workbench includes a glass tray and multiple groups of detection cameras, and a pre-trained defect detection model is integrated in the processor. The present invention automatically collects the top view, bottom view and side view of the rubber product to be tested through the detection and sorting device, and then performs defect detection on the collected top view, bottom view and side view through the pre-trained defect detection model, which can realize automatic defect detection of rubber products, reduce the process of manual operation, and thus improve production efficiency and the accuracy of defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solution in the embodiments of the present invention, the following introduction is made to the drawings required for use in the embodiments of the present invention. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solution of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 A structural block diagram of a rubber product defect detection device provided by an embodiment of the present invention;

[0037] Figure 2 A top view of a detection and sorting device provided in an embodiment of the present invention;

[0038] Figure 3 A front view of a detection and sorting device provided by an embodiment of the present invention;

[0039] Figure 4 A flow chart of the steps of a defect detection method of a rubber product defect detection device provided by an embodiment of the present invention;

[0040] Figure 5 A schematic diagram of a crack defect of a rubber product provided by an embodiment of the present invention;

[0041] Figure 6 A schematic diagram of inclusion defects of a rubber product provided by an embodiment of the present invention;

[0042] Figure 7 A schematic diagram of inclusion defects of a rubber product provided by an embodiment of the present invention;

[0043] Figure 8 A schematic diagram of the first stage training process provided by an embodiment of the present invention;

[0044] Fig. 9 A schematic diagram of the process flow of the second stage training provided by an embodiment of the present invention;

[0045] Fig.10 A schematic diagram of the bottom of a rubber product with cracks provided by an embodiment of the present invention;

[0046] Fig.11 A schematic diagram of the bottom of a qualified rubber product provided by an embodiment of the present invention;

[0047] Fig.12 A schematic diagram of the process flow of the third stage training provided by an embodiment of the present invention;

[0048] Fig.13 A schematic diagram of a method for dividing regions of a top view of a shock-absorbing pad provided by an embodiment of the present invention;

[0049] Fig.14A schematic diagram of another method for dividing regions of a top view of a shock-absorbing pad provided by an embodiment of the present invention;

[0050] Fig.15 A schematic diagram showing a comparison of a top view, a bottom view and a side view of a rubber shock-absorbing pad A and a rubber shock-absorbing pad B provided in an embodiment of the present invention;

[0051] FIG. 16( a ) is a schematic diagram of defect detection results of a top view of a shock-absorbing pad provided by an embodiment of the present invention;

[0052] FIG. 16( b ) is a schematic diagram of defect detection results of a top view of shock-absorbing pad B provided in an embodiment of the present invention;

[0053] FIG. 17( a ) is a schematic diagram of defect detection results of a bottom view of a shock-absorbing pad A provided by an embodiment of the present invention;

[0054] FIG. 17( b ) is a schematic diagram of defect detection results of a bottom view of a shock absorbing pad B provided by an embodiment of the present invention;

[0055] FIG. 18( a ) is a schematic diagram of defect detection results of a side view of a shock-absorbing pad A provided by an embodiment of the present invention;

[0056] FIG18( b ) is a schematic diagram of defect detection results of a side view of a shock absorbing pad B provided by an embodiment of the present invention;

[0057] Figure numerals: 1. workbench; 2. glass plate; 3. first air blowing port; 4. second air blowing port; 5. third air blowing port; 6. feeding belt; 7. lifting belt; 8. loading tray; 9. first inspection camera; 10. second inspection camera; 11. third inspection camera; 12. qualified product box; 13. waste box; 14. pending processing box. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are only examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the attached claims.

[0059] It is understood that the terms "first", "second", etc. used in this application can be used to describe various concepts in this article, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another concept. For example, without departing from the scope of the embodiment of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein can be interpreted as "at the time of" or "when" or "in response to determination".

[0060] The terms "at least one", "multiple", "each", "any", etc. used in this application, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0061] In the field of quality inspection of traditional rubber products, defect determination mainly relies on manual operation. Although this method is basic, it has significant disadvantages. First, in the face of large-scale production of rubber products, manual inspection requires repeatedly turning over each product, which is not only labor-intensive, but also requires a large amount of human resources, thus greatly increasing production costs and resulting in low production efficiency. Secondly, long-term continuous visual inspection can easily lead to visual fatigue of operators, thus affecting the accuracy and reliability of rubber product inspection, increasing the risk of missed inspections and misjudgments, and ultimately may lead to customer complaints due to product quality issues.

[0062] To this end, an embodiment of the present invention proposes a rubber product defect detection device, including a detection and sorting device and a processor, the detection and sorting device includes a loading tray, a belt feeder and a workbench, the workbench includes a glass tray and multiple groups of detection cameras, and a pre-trained defect detection model is integrated in the processor. The present invention automatically collects the top view, bottom view and side view of the rubber product to be tested through the detection and sorting device, and then performs defect detection on the collected top view, bottom view and side view through the pre-trained defect detection model, which can realize automatic defect detection of rubber products, reduce the process of manual operation, and thus improve production efficiency and the accuracy of defect detection.

[0063] Reference Figure 1 , Figure 1The present invention provides a structural block diagram of a rubber product defect detection device provided in an embodiment of the present invention. The present invention provides a rubber product defect detection device, including a detection and sorting device and a processor. The detection and sorting device includes a loading tray 8, a belt feeder and a workbench 1. The workbench 1 includes a glass tray 2 and multiple groups of detection cameras. One end of the belt feeder is connected to the loading tray 8, and the other end of the belt feeder is connected to the glass tray 2. Each group of detection cameras is located above or below the glass tray 2. Each group of detection cameras is also connected to the processor. A pre-trained defect detection model is integrated in the processor. The loading tray 8 is used to place the rubber product to be tested. The belt feeder is used to transport the rubber product to be tested from the loading tray 8 to the glass tray 2. The glass tray 2 is used to allow the rubber product to be tested to pass through each group of detection cameras in sequence. Each group of detection cameras is used to collect three views of the rubber product to be tested. The defect detection model is used to perform defect detection on the three views to obtain defect detection results, wherein the three views include a top view, a bottom view and a side view.

[0064] Specifically, the detection and sorting equipment is used to automatically collect the top view, bottom view and side view of the rubber product to be tested;

[0065] The processor, as the core component of the entire detection device, integrates a defect detection model trained based on a deep learning algorithm, which is used to perform various defect detections on the top view, bottom view, and side view of the rubber product to be tested to obtain defect detection results.

[0066] Reference Figure 2 and Figure 3 , Figure 2 To detect the top view of the sorting equipment, Figure 3 To detect the front view of the sorting equipment, further as an optional embodiment, the belt feeder includes a lifting belt 7 and a feeding belt 6, one end of the lifting belt 7 is connected to the bottom of the loading tray 8, the other end of the lifting belt 7 is connected to one end of the feeding belt 6, the other end of the feeding belt 6 is connected to the edge of the glass tray 2, the lifting belt 7 is used to transport the rubber product to be tested from the loading tray 8 to the feeding belt 6, and the feeding belt 6 is used to transport the rubber product to be tested from the lifting belt 7 to the glass tray 2.

[0067] Specifically, the detection and sorting equipment proposed in the embodiment of the present invention includes a loading tray 8, a belt feeder and a workbench 1, and the workbench 1 includes a rotatable glass tray 2 and multiple groups of detection cameras. A belt feeder is installed on one side of the glass tray 2, and the belt feeder includes a lifting belt 7 and a feeding belt 6. The lifting belt 7 connects the loading tray 8 and the feeding belt 6, and is used to transport the rubber product to be tested from the loading tray 8 to the feeding belt 6. The feeding belt 6 connects the lifting belt 7 and the glass tray 2, and is used to transport the rubber product to be tested from the lifting belt 7 to the glass tray 2.

[0068] Reference Figure 2and Figure 3 As an optional embodiment, the detection camera includes a first detection camera 9, a second detection camera 10 and a third detection camera 11. The first detection camera 9 is located above the glass plate 2, the second detection camera 10 is located below the glass plate 2, and the third detection camera 11 is placed on the glass plate 2. The first detection camera 9 is used to capture a top view of the rubber product to be tested, the second detection camera 10 is used to capture a bottom view of the rubber product to be tested, and the third detection camera 11 is used to capture a side view of the rubber product to be tested.

[0069] In some optional embodiments, the detection camera includes a first detection camera 9, a second detection camera 10 and a third detection camera 11. The first detection camera 9 is located above the glass plate 2 and is used to take photos of the rubber product to be tested from top to bottom, that is, a top view of the rubber product to be tested; the second detection camera 10 is located below the glass plate 2 and is used to take photos of the rubber product to be tested from bottom to top, that is, a bottom view of the rubber product to be tested; the third detection camera 11 is placed on the glass plate 2 and is used to take photos of the rubber product to be tested from the side, that is, a side view of the rubber product to be tested. Among them, the third detection camera 11 is provided with at least four cameras to take all-round photos of the side of the rubber product to be tested, and the detection cameras respectively collect at least 6 photos of the rubber product to be tested.

[0070] It should be noted that a single-angle photograph may lead to false alarms or missed reports of defects due to lighting, shadows, occlusions, etc. The embodiment of the present invention can provide more data points for the defect detection module by photographing the rubber product to be tested from different angles, displaying various parts and details of the product, thereby improving the accuracy and reliability of defect detection.

[0071] Furthermore, the embodiment of the present invention provides a rotatable and transparent glass plate 2, so that the rubber product to be tested passes through the first detection camera 9, the second detection camera 10 and the third detection camera 11 in sequence, and the top view, the bottom view and the side view of the rubber product to be tested are collected. During the process, there is no need for the operator to manually turn the product multiple times, which can reduce the investment in human resources and improve production efficiency.

[0072] Reference Figure 2 and Figure 3 In some optional embodiments, the detection and sorting equipment also includes a plurality of sorting boxes, and the workbench 1 also includes a plurality of air blowing ports, each of which is located on the glass plate 2, and each of the sorting boxes is located below each of the air blowing ports. Each of the air blowing ports is also connected to the processor, and the air blowing ports are used to blow the rubber products to be tested into the corresponding sorting boxes according to the defect detection results.

[0073] Reference Figure 2 and Figure 3In some optional embodiments, the air blowing port includes a first air blowing port 3, a second air blowing port 4 and a third air blowing port 5, the sorting box includes a qualified product box 12, a waste box 13 and a pending box 14, the qualified product box 12 is located below the first air blowing port 3, the waste box 13 is located below the second air blowing port 4, and the pending box 14 is located below the third air blowing port 5, and the first air blowing port 3, the second air blowing port 4 and the third air blowing port 5 are used to blow the rubber product to be tested into the qualified product box 12, the waste box 13 or the pending box 14 according to the defect detection result.

[0074] In summary, the working process of a rubber product defect detection device according to an embodiment of the present invention is as follows: when working, the lifting belt 7 brings the rubber product to be tested from the loading tray 8 to the feeding belt 6, and the feeding belt 6 brings the rubber product to be tested into the glass plate 2. When the glass plate 2 rotates, the rubber product to be tested passes through the first detection camera 9 in sequence, and the first detection camera 9 takes a top view of the rubber product to be tested from top to bottom; the rubber product to be tested then passes through the second detection camera 10, and the second detection camera 10 takes a top view of the rubber product from bottom to top; the rubber product then passes through the third detection camera 11, and the third detection camera 11 takes a side view of the rubber product from the side. Then, the defect detection model in the processor comprehensively determines whether the rubber product to be tested is a qualified product, a waste product, or a product to be processed through the top view, bottom view, and side view of the rubber product to be tested. The qualified product is blown into the qualified product box 12 through the first blowing port 3, the waste product is blown into the waste product box 13 through the second blowing port 4, and the product to be processed is blown into the to-be-processed box 14 through the third blowing port 5.

[0075] The above describes the structure and workflow of the rubber product defect detection device according to the embodiment of the present invention. It can be recognized that compared with the existing rubber product defect detection method, the embodiment of the present invention has the following advantages:

[0076] 1. By setting a rotatable and transparent glass plate, the rubber product to be tested is passed through the first detection camera, the second detection camera and the third detection camera in sequence, and the top view, bottom view and side view of the rubber product to be tested are collected. During the process, there is no need for the operator to manually turn the product many times, which can reduce the investment of human resources and improve production efficiency;

[0077] Second, by collecting top, bottom and side views of the rubber product to be tested from different angles, it can provide more data points for the defect detection module, showing the various parts and details of the product, thereby improving the accuracy and reliability of defect detection;

[0078] 3. The top view, bottom view and side view of the rubber product to be tested are automatically collected by the detection and sorting equipment, and then the collected top view, bottom view and side view are detected by the pre-trained defect detection model. This can realize automatic defect detection of rubber products and reduce the manual operation process, thereby improving production efficiency and the accuracy of defect detection.

[0079] Reference Figure 4 , Figure 4 The present invention is a flowchart of a defect detection method of a rubber product defect detection device. The present invention provides a defect detection method of a rubber product defect detection device, which is used to be performed by the above-mentioned rubber product defect detection device. The method includes steps S101 to S105:

[0080] S101, placing the rubber product to be tested on a loading tray;

[0081] S102, conveying the rubber product to be tested from the loading tray to the glass tray by a belt feeder;

[0082] S103, passing the rubber product to be tested through a glass plate through multiple groups of testing cameras in sequence;

[0083] S104, collecting three views of the rubber product to be tested by using multiple groups of detection cameras, where the three views include a top view, a bottom view, and a side view;

[0084] Furthermore, different areas of the rubber product have different defect detection effects when images are taken under different light source conditions. Therefore, in order to improve the accuracy and reliability of model defect detection, at least a red light source, a green light source, a blue light source and a combined light source are used to shoot the top view, bottom view and side view of the rubber product to be tested.

[0085] S105 , performing defect detection on the three views using a defect detection model pre-trained in the processor to obtain defect detection results.

[0086] As an optional implementation, the defect detection method further includes a step of pre-training a defect detection model. The step of pre-training a defect detection model can be further divided into the following steps A101 and A102:

[0087] A101, obtaining a training sample set, the training sample set including a plurality of top view samples, bottom view samples, side view samples and corresponding detection labels of the rubber products to be tested;

[0088] Specifically, a certain number of top view samples, bottom view samples, and side view samples of the rubber products to be tested can be collected from the production line or inventory, half of each sample is qualified products, and the other half is clearly marked as waste products, and each sample includes at least images taken by red light source, green light source, blue light source, and comprehensive light source, and then these samples are labeled to determine whether they are qualified products or waste products. Furthermore, the defect type of the waste product can also be labeled so that the model learns how to classify the defect type of the waste product. In addition to judging whether the rubber product to be tested is a qualified product or a waste product, the defect type corresponding to the waste product is output to improve the defect detection accuracy of the system.

[0089] A102. Determine defect judgment rules, perform multi-stage training on the pre-built convolutional neural network according to the training sample set, detection labels and defect judgment rules, and obtain a defect detection model.

[0090] As an optional implementation, the defect determination rule is determined, and the pre-built convolutional neural network is trained in multiple stages according to the training sample set, the detection label and the defect determination rule to obtain the defect detection model. This step can be further divided into the following steps A1021 to A1027:

[0091] A1021. Determine a defect determination rule, and then input the defect determination rule into a convolutional neural network;

[0092] In some optional embodiments, the defects of the rubber product to be tested include but are not limited to lack of glue (i.e., missing material on the surface or inside of the product), bubbles (i.e., cavities formed inside the product), chipping (i.e., surface material falling off or damaged), cracks (i.e., small cracks on the surface or inside of the product, such as Figure 5 As shown), inclusions (i.e. foreign matter mixed into the product, such as Figure 6 and Figure 7 There are many types of defects, such as deformation (i.e. the shape of the product deviates from the design standard) or size (i.e. the size of the product does not meet the specification requirements).

[0093] For quantifiable features (such as area and grayscale contrast), clear thresholds can be set as defect judgment criteria. For example, the area critical value of a certain area of ​​the rubber product to be tested is set to 500, and the grayscale contrast critical value is set to 40. When the total area of ​​the area exceeds 500 and the grayscale contrast exceeds 40, it is defined as a waste product;

[0094] For products with specific shape and size requirements, clear geometric parameter standards can be set, and algorithms can be designed to measure and compare these parameters. For example, the roundness critical value of the shock-absorbing pad is defined as 0.4, the upper limit of the inner diameter in a certain area is 9.9, the lower limit of the inner diameter is 9.3, the upper limit of the outer diameter is 27, and the lower limit of the outer diameter is 25. If the maximum and minimum values ​​of the roundness and diameter of the shock-absorbing pad are out of the specified range, it is defined as a scrap product;

[0095] For defects that are difficult to determine based on a single condition, multispectral imaging technology can be used. For example, the images of area A of the rubber product to be tested taken by a red light source are used for defect judgment, and the images of area B taken by a green light source are used for defect judgment. When the images under different light sources show different image features from qualified products, they are defined as waste products.

[0096] For rubber products to be tested that are difficult to identify defects, such as those containing residual rubber, they are defined as products to be processed.

[0097] A1022. Input the top view sample and the side view sample into the convolutional neural network for the first stage training, and then randomly select the top view sample and the side view sample to test the convolutional neural network after the first stage training to obtain a first defect detection result;

[0098] A1023. updating the defect determination rule according to the first defect detection result until the first defect detection result reaches a preset first test condition and the randomly input top view samples and side view samples reach a preset first quantity threshold;

[0099] Specifically, Figure 8 The figure shows a flow chart of the first stage of training. First, a large number of top view samples and side view samples of the rubber products to be tested (half of which are qualified products and half are defective products) are input to train the convolutional neural network. When the defect detection result output by the network is inconsistent with the detection label corresponding to the sample, the defect judgment rule of the network is adjusted, and the top view samples and side view samples are continuously input to train the convolutional neural network. When the defect detection result output by the network is consistent with the detection label corresponding to the sample, it is further determined whether the input number of the top view samples and the side view samples reaches a first preset number. When the first preset number is reached, random photos are additionally input to test the convolutional neural network. When the first defect detection result output by the network reaches the first test condition and the input random photos reach the first quantity threshold, it can be considered that the network has initially entered the practical stage and enters the next stage of training.

[0100] Those skilled in the art can understand that the first preset number, the first test condition and the first number threshold can be set according to actual needs. For example, the first preset number can be 1500 sheets, 1000 sheets, etc., the first test condition can be that the first defect detection result is consistent with the manual judgment standard, etc., and the first number threshold can be 100 sheets, 200 sheets, etc.

[0101] A1024, inputting the top view sample and the side view sample at the critical point into the convolutional neural network for second stage training to obtain a second defect detection result;

[0102] A1025. Update the defect determination rule according to the second defect detection result and the detection label until the second defect detection result reaches a preset second test condition and the input top view samples and side view samples at the critical point reach a preset second quantity threshold;

[0103] Specifically, Fig. 9 The figure shows a flow chart of the second stage of training, in which a large number of top view samples and side view samples of the rubber products to be tested at the critical point are input (i.e., in the manual inspection state, the defect points are not obvious, and inspector A judges them as qualified products, while inspector B judges them as scrap products), and the convolutional neural network is trained. When the second defect detection result output by the network is inconsistent with the detection label corresponding to the sample, the defect judgment rule of the network is adjusted, and the top view samples and side view samples at the critical point are continuously input to train the convolutional neural network, until the second defect detection result output by the network meets the second test condition, and the top view samples and side view samples input at the critical point reach the second quantity threshold, and then the next stage of training is entered.

[0104] Those skilled in the art can understand that the second test condition and the second quantity threshold can be set according to actual needs. For example, the second test condition can be that the second defect detection result is consistent with the corresponding detection label or the second defect detection result is consistent with the manual judgment standard, and the second quantity threshold can be 200 sheets, 100 sheets, etc.

[0105] A1026, determining the lettering features of the rubber product to be tested, importing the lettering features into a preset shielding list, and then inputting the bottom view sample and the shielding list into the convolutional neural network for the third stage training to obtain a third defect detection result;

[0106] A1027. Update the defect judgment rules according to the third defect detection result and the detection label until the third defect detection result reaches the preset third test condition and the input top view sample reaches the preset third quantity threshold, stop training, and obtain a trained defect detection model.

[0107] It should be noted that the bottom of rubber products usually has letters or numbers related to the product specifications, such as "w026", "P2" and "214". Among them, "w026" and "P2" are fixed numbers of the rubber product specifications, and "214" is the cavity number of the rubber product mold, ranging from 001 to 367. When determining defects in the model, these letters and numbers are easily confused with defects such as chipping and cracks. Fig.10 The bottom diagram of the rubber product with cracks is shown in FIG. Fig.11 The figure shows a schematic diagram of the bottom of a qualified rubber product. Therefore, the above-mentioned lettering and numbers need to be included in the model's shielding list to avoid the model listing the lettering or numbers as defects.

[0108] Specifically, Fig.12 The figure shows a flow chart of the third stage of training. First, the lettering features that need to be blocked are determined, such as the letters "W", "P", and numbers "1-9" as graphics, which are included in the pre-established blocking list in the model. Then, a large number of top-view samples with lettering features are input to train the convolutional neural network. When the third defect detection result output by the network is inconsistent with the detection label corresponding to the sample, the defect judgment rule of the network is adjusted until the third defect detection result output by the network meets the third test condition and the input top-view samples reach the preset third quantity threshold, thus obtaining a trained defect detection model.

[0109] Those skilled in the art can understand that the third preset number and the third test condition can be set according to actual needs. For example, the third preset number can be 500 sheets, 200 sheets, etc., and the third test condition can be that the third defect detection result is consistent with the corresponding detection label or the third defect detection result is consistent with the manual judgment standard, etc.

[0110] As an optional implementation, the step of performing defect detection on the three views by using a pre-trained defect detection model in the processor to obtain a defect detection result can be further divided into the following steps S1051 and S1052:

[0111] S1051, inputting the top view, the bottom view and the side view into the defect detection model, and then dividing the top view, the bottom view and the side view into regions to obtain a plurality of regions to be tested;

[0112] S1052, performing defect detection on each area to be tested to obtain defect detection results;

[0113] Among them, defect detection includes at least one of glue deficiency detection, bubble detection, flower chip detection, crack detection, inclusion detection, deformation detection or size detection.

[0114] Specifically, the three-view images of the rubber product to be tested can be divided into regions according to the actual testing requirements and goals. For example, for some key components or areas, they can be divided more finely for more stringent testing; while for non-key areas, the testing standards can be appropriately relaxed to reduce the amount of calculation. For example, taking the top view of a rubber shock-absorbing pad as an example, Fig.13 The figure shows a schematic diagram of a method for dividing the area of ​​the top view of the shock-absorbing pad, as shown in FIG. Fig.14 The figure shows another method of dividing the regions of the top view of the shock-absorbing pad. After the regions are divided, the model performs defect detection on each region to be tested according to the determined defect judgment rules, and the output defect detection results show whether the rubber product to be tested is a qualified product, a waste product, or a product to be processed.

[0115] As an optional embodiment, the defect detection method further includes the step of blowing the rubber product to be tested into a corresponding sorting box through a plurality of air ports according to the defect detection result, the air ports include a first air port, a second air port and a third air port, the sorting box includes a qualified product box, a waste product box and a to-be-processed box, and the step of blowing the rubber product to be tested into the corresponding sorting box through the plurality of air ports according to the defect detection result can be further divided into the following steps S1061 to S1063:

[0116] S1061. When the defect detection result shows that the rubber product to be tested is a qualified product, the rubber product to be tested is blown into a qualified product box through a first blowing port;

[0117] S1062, when the defect detection result shows that the rubber product to be tested is waste, the rubber product to be tested is blown into a waste box through the second blowing port;

[0118] S1063. When the defect detection result shows that the rubber product to be tested is a product to be processed, the rubber product to be tested is blown into a processing box through the third blowing port.

[0119] Take a rubber shock-absorbing pad as an example. Fig.15 The figure shows a schematic diagram comparing the top view, bottom view and side view of the rubber shock-absorbing pad A and the rubber shock-absorbing pad B. The following is a detailed description of the working process of the defect detection method of the rubber product defect detection device according to the embodiment of the present invention:

[0120] First, the shock-absorbing pad A and the shock-absorbing pad B are transported from the loading tray to the glass tray by a belt feeder;

[0121] Next, taking the example that the first detection camera includes one camera, the second detection camera includes one camera, and the third detection camera includes four cameras, the glass plate rotates, and the shock-absorbing pad A and the shock-absorbing pad B pass through the first detection camera, the second detection camera, and the third detection camera in sequence, and a top view, a bottom view, and four side views of the shock-absorbing pad A and the shock-absorbing pad B are acquired;

[0122] Furthermore, the collected photos are input into the defect detection model trained in the processor, and the model makes a comprehensive judgment on the six collected photos. As shown in Figure 16(a), it is a schematic diagram of the defect detection result of the top view of shock absorber pad A, and as shown in Figure 16(b), it is a schematic diagram of the defect detection result of the top view of shock absorber pad B. As shown in Figure 17(a), it is a schematic diagram of the defect detection result of the bottom view of shock absorber pad A, and as shown in Figure 17(b), it is a schematic diagram of the defect detection result of the bottom view of shock absorber pad B. As shown in Figure 18(a), it is a schematic diagram of the defect detection result of the side view of shock absorber pad A, and as shown in Figure 18(b), it is a schematic diagram of the defect detection result of the side view of shock absorber pad B. It can be seen that the model judges all three views of shock absorber pad A as qualified products, and all three views of shock absorber pad B as defective products.

[0123] Finally, the shock-absorbing pad A is blown into the qualified product box through the first blowing port, and the shock-absorbing pad B is blown into the waste box through the second blowing port.

[0124] The contents of the above-mentioned rubber product defect detection device embodiment are all applicable to the defect detection method embodiment of the rubber product defect detection device. The functions specifically implemented by the defect detection method embodiment of the rubber product defect detection device are the same as those of the above-mentioned rubber product defect detection device embodiment, and the beneficial effects achieved are also the same as those achieved by the above-mentioned rubber product defect detection device embodiment.

[0125] In some selectable embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the above-mentioned boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided by way of example, for the purpose of providing a more comprehensive understanding of technology. The disclosed method is not limited to the operation and logic flow presented herein. Selectable embodiments are expected, wherein the order of various operations is changed and the sub-operation of a part for which is described as a larger operation is performed independently.

[0126] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise specified to the contrary, one or more of the above-mentioned functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present invention. More specifically, in view of the properties, functions and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional skills of the engineer. Therefore, those skilled in the art can implement the present invention set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0127] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the above methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0128] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0129] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and editable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the above-mentioned program is printed, since the above-mentioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or processing in other suitable ways as necessary, and then stored in a computer memory.

[0130] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0131] In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0132] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0133] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A rubber product defect detection device, characterized in that: The invention comprises a detection and sorting device and a processor, wherein the detection and sorting device comprises a loading tray, a belt feeder and a workbench, wherein the workbench comprises a glass tray and a plurality of detection cameras, wherein one end of the belt feeder is connected to the loading tray, and the other end of the belt feeder is connected to the glass tray, wherein each group of the detection cameras is located above or below the glass tray, and each group of the detection cameras is also connected to the processor, wherein a pre-trained defect detection model is integrated in the processor, wherein the loading tray is used for placing the rubber product to be tested, the belt feeder is used for conveying the rubber product to be tested from the loading tray to the glass tray, and the glass tray is used for allowing the rubber product to be tested to pass through each group of the detection cameras in sequence, and each group of the detection cameras is used for collecting three views of the rubber product to be tested, and the defect detection model is used for performing defect detection on the three views to obtain defect detection results, wherein the three views comprise a top view, a bottom view and a side view.

2. A rubber product defect detection device according to claim 1, characterized in that: The detection and sorting equipment also includes a plurality of sorting boxes, and the workbench also includes a plurality of air blowing ports, each of which is located on the glass plate, and each of the sorting boxes is located below each of the air blowing ports. Each of the air blowing ports is also connected to the processor, and the air blowing ports are used to blow the rubber product to be tested into the corresponding sorting box according to the defect detection result.

3. A rubber product defect detection device according to claim 1, characterized in that: The detection camera includes a first detection camera, a second detection camera and a third detection camera, the first detection camera is located above the glass plate, the second detection camera is located below the glass plate, and the third detection camera is placed on the glass plate. The first detection camera is used to collect the top view of the rubber product to be tested, the second detection camera is used to collect the bottom view of the rubber product to be tested, and the third detection camera is used to collect the side view of the rubber product to be tested.

4. A rubber product defect detection device according to claim 1, characterized in that: The belt feeder includes a lifting belt and a feeding belt, one end of the lifting belt is connected to the bottom of the loading tray, the other end of the lifting belt is connected to one end of the feeding belt, and the other end of the feeding belt is connected to the edge of the glass tray, the lifting belt is used to transport the rubber product to be tested from the loading tray to the feeding belt, and the feeding belt is used to transport the rubber product to be tested from the lifting belt to the glass tray.

5. The rubber product defect detection device according to claim 2, characterized in that: The air blowing port includes a first air blowing port, a second air blowing port and a third air blowing port, the sorting box includes a qualified product box, a waste box and a to-be-processed box, the qualified product box is located below the first air blowing port, the waste box is located below the second air blowing port, and the to-be-processed box is located below the third air blowing port, and the first air blowing port, the second air blowing port and the third air blowing port are used to blow the rubber product to be tested into the qualified product box, the waste box or the to-be-processed box according to the defect detection result.

6. A defect detection method of a rubber product defect detection device, which is used to be performed by the rubber product defect detection device according to any one of claims 1 to 5, characterized in that: The following steps are involved: Place the rubber product to be tested on the loading tray; The rubber product to be tested is conveyed from the loading tray to the glass tray by a belt feeder; The rubber product to be tested is passed through multiple groups of detection cameras in sequence through the glass plate; Collect three views of the rubber product to be tested by using multiple groups of the detection cameras, wherein the three views include a top view, a bottom view, and a side view; Defect detection is performed on the three views using a defect detection model pre-trained in the processor to obtain a defect detection result.

7. The defect detection method of a rubber product defect detection device according to claim 6, characterized in that: The defect detection method further comprises the step of pre-training the defect detection model, wherein the pre-training of the defect detection model specifically comprises: Acquire a training sample set, wherein the training sample set includes a plurality of top view samples, bottom view samples, side view samples and corresponding detection labels of the rubber product to be tested; Determine defect judgment rules, perform multi-stage training on a pre-constructed convolutional neural network according to the training sample set, the detection labels and the defect judgment rules, and obtain the defect detection model.

8. The defect detection method of a rubber product defect detection device according to claim 7, characterized in that: The determining of the defect judgment rule, performing multi-stage training on a pre-built convolutional neural network according to the training sample set, the detection label and the defect judgment rule to obtain the defect detection model, specifically includes: Determining a defect determination rule, and then inputting the defect determination rule into the convolutional neural network; Inputting the top view sample and the side view sample into the convolutional neural network for first-stage training, and then randomly selecting the top view sample and the side view sample to test the convolutional neural network after the first-stage training to obtain a first defect detection result; The defect determination rule is updated according to the first defect detection result until the first defect detection result reaches a preset first test condition and the top view sample and the side view sample randomly input reach a preset first quantity threshold; Inputting the top view sample and the side view sample at the critical point into the convolutional neural network for second-stage training to obtain a second defect detection result; The defect determination rule is updated according to the second defect detection result and the detection label until the second defect detection result reaches a preset second test condition and the input top view sample and the side view sample at the critical point reach a preset second quantity threshold; Determine the lettering features of the rubber product to be tested, import the lettering features into a preset shielding list, and then input the bottom view sample and the shielding list into the convolutional neural network for third-stage training to obtain a third defect detection result; The defect judgment rule is updated according to the third defect detection result and the detection label until the third defect detection result reaches a preset third test condition and the input bottom view sample reaches a preset third quantity threshold, and the training is stopped to obtain the trained defect detection model.

9. The defect detection method of a rubber product defect detection device according to claim 6, characterized in that: The defect detection is performed on the three views by using a defect detection model pre-trained in the processor to obtain a defect detection result, specifically including: Inputting the top view, the bottom view and the side view into the defect detection model, and then dividing the top view, the bottom view and the side view into regions to obtain a plurality of regions to be tested; Performing defect detection on each of the test areas to obtain defect detection results; Wherein, the defect detection includes at least one of glue deficiency detection, bubble detection, flower chipping detection, crack detection, inclusion detection, deformation detection or size detection.

10. The defect detection method of a rubber product defect detection device according to claim 6, characterized in that: The defect detection method further includes the step of blowing the rubber product to be tested into a corresponding sorting box through a plurality of air ports according to the defect detection result, the air ports include a first air port, a second air port and a third air port, the sorting box includes a qualified product box, a waste product box and a to-be-processed box, and blowing the rubber product to be tested into a corresponding sorting box through a plurality of air ports according to the defect detection result specifically includes: When the defect detection result shows that the rubber product to be tested is a qualified product, the rubber product to be tested is blown into the qualified product box through the first blowing port; When the defect detection result shows that the rubber product to be tested is waste, the rubber product to be tested is blown into the waste box through the second blowing port; When the defect detection result shows that the rubber product to be tested is a product to be processed, the rubber product to be tested is blown into the processing box through the third blowing port.

Images