CN112557399A - Point inspection method and device for smoke machine equipment quality detection system - Google Patents
Point inspection method and device for smoke machine equipment quality detection system Download PDFInfo
- Publication number
- CN112557399A CN112557399A CN202011369942.9A CN202011369942A CN112557399A CN 112557399 A CN112557399 A CN 112557399A CN 202011369942 A CN202011369942 A CN 202011369942A CN 112557399 A CN112557399 A CN 112557399A
- Authority
- CN
- China
- Prior art keywords
- cigarette packet
- image
- point inspection
- image template
- packet image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
- G01N21/956—Inspecting patterns on the surface of objects
- G01N21/95607—Inspecting patterns on the surface of objects using a comparative method
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/75—Determining position or orientation of objects or cameras using feature-based methods involving models
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
- G01N21/956—Inspecting patterns on the surface of objects
- G01N21/95607—Inspecting patterns on the surface of objects using a comparative method
- G01N2021/95615—Inspecting patterns on the surface of objects using a comparative method with stored comparision signal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a point inspection method and a related device for a quality detection system of cigarette making machine equipment, and particularly belongs to the field of tobacco wrapping equipment. The point inspection method comprises a configuration stage and an operation stage, wherein: in a configuration stage, configuring a point inspection panel, an image identification module and a tobacco bale image template library for the quality detection system; training the image recognition module by using a cigarette packet image template of a cigarette packet image template library so that the image recognition module marks a typical defect cigarette packet image template in the cigarette packet image template; in the operation stage, when analog detection, automatic point inspection and/or system validation are required, the typical defect cigarette packet image template is extracted through the point inspection panel to replace the cigarette packet image of the imaging detector and is sent to the quality detection system. The point inspection device comprises a data server, a communication interface and a point inspection panel. The technical scheme of the invention realizes the simulation detection, the automatic point detection and the system validation of quality detection systems such as a small packet imaging detector of a tobacco wrapping machine and the like.
Description
Technical Field
The invention belongs to the field of tobacco wrapping equipment, and particularly relates to a point inspection method and a related device for a quality inspection system of cigarette making machine equipment.
Background
In the wrapping process of tobacco production, the appearance of the finished packet produced by a packaging machine needs to be checked. At present, cigarette machine equipment quality detection systems adopting an imaging detection technology are adopted for appearance detection of finished product packets of packaging machines such as ZB45 and the like, CCD cameras are used as imaging detectors, defects such as leakage of cigarette packet paperboard, loss of inner lining paper, turning, creasing, skewing, opening, fouling and the like are detected by photographing, and online real-time detection and automatic elimination are realized. The detection principle is that a camera is used for shooting, the standard images of the qualified small packets are stored, and a standard model is established; in the production process of the ZB45 packaging machine, a camera is triggered through a synchronous sensor to shoot in real time, real-time images of each packet of cigarettes are shot, the images of each packet of cigarettes are analyzed and compared with stored standard images, the detected values are within set tolerance, the packet of cigarettes is qualified in appearance, otherwise, the packet of cigarettes is judged to be unqualified, and the unqualified cigarettes are marked and shifted by an image processing industrial personal computer and are automatically removed at a removing opening.
In order to ensure the appearance quality of cigarette products and the effectiveness of the functions of an imaging detection system, in the prior art, a point inspection method for performing point inspection verification of an imaging detector is adopted: the method comprises the steps of performing point inspection test on an imaging detector during handover of each shift, manually simulating a defective cigarette packet, putting the defective cigarette packet into a production line, destructively performing defective cigarette packet test, identifying the defective cigarette packet if the defective cigarette packet passes through an imaging detection system, successfully eliminating the defective cigarette packet, and judging that the detection function of the current imaging detector system is intact and effective. The point inspection method can effectively judge the effectiveness of the detection system and plays an important role in ensuring the product quality, but has a plurality of defects: on one hand, unqualified cigarette packets need to be manufactured manually every shift change, which is tedious and difficult to ensure that unqualified cigarette packet templates are uniform; on the other hand, due to the fact that the defects to be tested are various, various unqualified cigarette packet templates need to be manufactured every time, the occupied time is long, and waste is serious; during simulation, the machine needs to be stopped and operated at a low speed, a plurality of unqualified cigarette packet templates are manually placed into a production channel in a time-sharing manner one by one, and if all unqualified cigarette packet templates are detected once, all machines are verified, so that the consumption is high, and the production efficiency is seriously influenced; in another aspect, the newly introduced production line equipment such as the high-speed wrapping machine has high tobacco bale conveying speed on the production line and is difficult to manually put in.
Disclosure of Invention
The invention aims to provide a point inspection method of a cigarette machine equipment quality detection system, which is used for the simulation detection, the automatic point inspection and the system validation of a packet imaging detector of a tobacco wrapping machine.
The invention provides a point inspection method of a cigarette machine equipment quality detection system, which comprises a configuration stage and an operation stage, wherein:
in a configuration stage, configuring a point inspection panel, an image identification module and a tobacco bale image template library for the quality detection system; training the image recognition module by using a cigarette packet image template of a cigarette packet image template library so that the image recognition module marks a typical defect cigarette packet image template in the cigarette packet image template; the cigarette packet image template comprises a qualified cigarette packet image template and a defective cigarette packet image template; the typical defects include missing, turned up, crumples, skewing, openings, and/or blemishes.
In the operation stage, when analog detection, automatic point inspection and/or system validation are required, the typical defect cigarette packet image template is extracted through the point inspection panel to replace the cigarette packet image of the imaging detector and is sent to the quality detection system.
One improvement of the point inspection method is that: comprises an updating stage; and in the updating stage, the image recognition module is retrained so that the image recognition module can re-mark the typical defective cigarette packet image template.
One improvement of the point inspection method is that: and during the normal operation of the quality detection system, acquiring a defective cigarette packet image, and if the defective cigarette packet image is classified by the image identification module and the defect characteristic index is atypical, storing the defective cigarette packet image as a defective cigarette packet image template in the cigarette packet image template library.
One improvement of the point inspection method is that: the image recognition module comprises a first image neural network unit;
in the configuration stage, defects are marked for defective cigarette packet image templates in the cigarette packet image template library, all cigarette packet image templates are divided into a training set and a verification set, and the first image neural network unit is trained, so that for a given defective cigarette packet image after training is finished, the first image neural network unit outputs the confidence coefficient of the defective cigarette packet image on each defect; the defect cigarette packet image template with the highest confidence level on a defect in the defect cigarette packet image templates of cigarette packets with the same specification is marked as a typical defect cigarette packet image template of the cigarette packet with the specification under the defect;
in the operation stage, when the first image neural network unit reads that each defect of a defective cigarette packet image provided by the quality detection system is atypical, one or more defects are distributed to the first image neural network unit and then the first image neural network unit is used as a new defective cigarette packet image template to be placed into the defective cigarette packet image template set.
One improvement of the point inspection method is that: when simulation detection is needed, a typical defect tobacco bale image template under a typical defect is extracted through the point detection panel to replace a tobacco bale image and is sent to the quality detection system, so that the elimination function of the typical defect of the quality detection system is detected.
One improvement of the point inspection method is that: when automatic point inspection is needed, typical defective cigarette packet image templates of cigarette packets with the same specification are extracted through the point inspection panel and inserted into the cigarette packet image stream to be sent to the quality detection system, so that the elimination function of the defective cigarette packets by the quality detection system during continuous production is detected.
One improvement of the point inspection method is that: when system verification is needed, the typical defect cigarette packet image template is extracted through the point inspection panel and is sent to the quality inspection system without operating other equipment on a production line.
An improvement of the above-mentioned spot inspection method is that the automatic spot inspection process comprises the following steps:
step 1, triggering light sources to turn on lights according to stations in sequence and obtaining a current cigarette packet image;
step 2, calling a typical defect cigarette packet image template of the cigarette packets with the same specification to replace the current cigarette packet image and sending the image to the quality detection system;
step 3, the quality detection system judges whether the defect exists, if not, the current production line is judged to be unqualified, and if so, the step 4 is executed;
step 4, triggering a rejection sensor, judging that the current production line is unqualified if the rejection sensor cannot be successfully triggered, and executing step 5 if the rejection sensor can be triggered;
and 5, finishing the automatic point inspection and judging that the current production line is qualified.
And 6, outputting an automatic point inspection report of the current production line.
The invention provides a full-automatic simulation point inspection method, which aims at the existing small bag appearance imaging detection device and develops a set of processing program. On the basis of an imaging detection device, a simulation point inspection interface is created, unqualified cigarette packet templates are collected in advance and stored in a system in advance to serve as templates of unqualified cigarette packets; and then, under the conditions that the packaging machine does not stop running and does not slow down, unqualified cigarette packet templates can be taken through the simulation point inspection interface at any time, and the functions of simulation detection, automatic point inspection and system verification are realized.
The invention adopts a standard modeling principle, and realizes the functions of simulation detection, automatic point inspection and system validation by establishing two image templates, namely a qualified cigarette packet image template and a defective cigarette packet image template and adopting two different template data analysis and comparison.
The problem to be solved by the first aspect of the invention is that when the system is just put into use, the initial modeling of the qualified cigarette packet image template and the defective cigarette packet image template is not problematic, and the point inspection can be normally implemented, but as the time for putting into use of the image detection system is originally longer, the LED light source gradually attenuates, the actual detection values of the characters, colors, patterns and positions of the photographed cigarette packet image become lower and lower, the actual detection values are lower than the initially set control threshold value, but the cigarette packet is normal at the moment, is a qualified cigarette packet, and can be subjected to false inspection or non-inspection.
The problem to be solved by the second aspect of the invention is that the defects of the cigarette packet, such as paperboard leakage, inner lining paper loss, turning, crumpling, skewing, opening, fouling and the like, are various, the patterns are not fixed or are completely the same, the manual simulation of the defective cigarette packet is limited to the defects which often occur and are representative, and the purpose of simulation point inspection cannot be tested and is difficult to realize for accidental defects or defects which are extremely difficult to occur.
The invention adopts a neural network algorithm technology, continuously identifies qualified cigarette packet images and defective cigarette packet images by calling an application interface of the existing image neural network unit and HALCON image processing software and applying a multi-feature identification technology, establishes and simulates a neural network for analyzing and learning human brain, and then classifies the images to realize AI artificial intelligent control of simulation point inspection.
In some embodiments, the learning training is performed for two image templates of the system, one for a qualified cigarette packet image template and one for a defective cigarette packet image template.
Wherein, an aim at of qualified tobacco bale image template is as follows: the method comprises the steps of taking a pre-collected qualified cigarette packet template as a reference, continuously photographing through a camera in the production process, greatly identifying four main characteristics of characters, colors, patterns and positions of the cigarette packet, simulating human brain through a neural network algorithm to analyze and learn, continuously training and correcting the parameters of the characters, the colors, the patterns and the positions, and further adjusting a monitoring threshold value in real time. Through a large amount of learning and correction of the method, the qualified cigarette packet image template is real-time, the problem that the LED light source in the first aspect gradually attenuates can be effectively solved, and the automation of the updating of the qualified cigarette packet image template is realized.
One purpose of the defective cigarette packet image template is as follows: for the defects of the cigarette packet, such as the leakage of paperboard, the loss of lining paper, the turning, the crease, the skew, the opening, the fouling and the like, the initial manual template can not be used too much, and the continuous study in the production is also needed. The invention classifies the characteristics of the defects of the cigarette packet such as cardboard leakage, inner lining paper loss, reverse wrapping, cockling, skewing, opening, fouling and the like in the process, the quality defect characteristics of redundant manual templates can appear in the production process, and the system can store the newly appeared quality defect characteristics into the characteristic classes, so that the data of each quality defect characteristic class is continuously improved and increased through a large number of characteristic identification technologies of a neural network algorithm, the defect characteristics are more and more improved along with the longer and longer time of the system in use, the second difficulty can be effectively solved, and the intellectualization of the defective cigarette packet image templates is realized.
In some embodiments, the method of the present invention is implemented using a dedicated device, and the device operation process include:
establishing a simulation point inspection interface on a programmable human-computer interface as a point inspection panel, and configuring an image identification module and a cigarette packet image template library which are in butt joint with software on a defect identification computer system of a quality inspection system: template picture windows, template parameters and point inspection buttons of various unqualified cigarette packets are edited in advance on the point inspection panel.
And initializing and inputting a point inspection template for a cigarette packet image template library of the defect identification computer system, wherein the template of the qualified cigarette packet is manufactured by photographing and sampling all unqualified templates through a camera of the detection device. The template for making defective cigarette packets can be manually and uniformly used for making templates for unqualified cigarette packets at one time, wherein the templates comprise deletion, reverse wrapping, creasing, skewing, opening, fouling and the like.
In the operation stage, the system continuously finishes the updating and accumulation of the templates in the tobacco bale image template library through a large amount of data in the production process.
When an inspector needs the small-package imaging detection device to perform simulation detection, the template of an unqualified cigarette package is only required to be taken from a simulation point detection interface, a point detection button is pressed, simulation detection comprises the steps of missing, turning, creasing, skewing, opening, fouling and the like of the unqualified cigarette package, and whether the unqualified cigarette package can be detected or not is verified and is automatically removed at a removal opening. Through the automatic point inspection function of the simulation point inspection interface, the point inspection can be carried out on the small bag imaging detection device at any time or regularly, inspection personnel do not need to manufacture unqualified cigarette packet templates and manually add cigarettes during the automatic point inspection process, the normal operation of the packaging machine is not stopped and does not slow down, the point inspection of the detection function of the small bag imaging detection device is directly completed, the efficiency is improved, and the point inspection of the packaging machine at each speed is realized.
When the detection device runs for a long time, technicians need to check the system of the detection device to detect whether the detection device works normally, unqualified cigarette packet templates do not need to be manufactured and cigarettes are added manually, only any unqualified cigarette packet template needs to be called on a simulation point detection interface, a point detection button is pressed, if the device can perform normal detection and automatic elimination, the detection parameters and functions of the system are indicated to be normal, and the detection device is very convenient and fast.
According to the technical scheme of the method, a special point inspection device is provided for the quality detection system of the existing cigarette making machine equipment in one embodiment, and the point inspection device comprises a data server, a communication interface and a virtual or real point inspection panel which are independent or share the same platform with the quality detection system; wherein the content of the first and second substances,
the data server is provided with a tobacco bale image template library; the tobacco bale image template library stores qualified tobacco bale image templates and defective tobacco bale image templates; the defective cigarette packet image template comprises a typical defective cigarette packet image template; the typical defective cigarette packet image template is dynamically marked in the defective cigarette packet image template by an image identification module;
the communication interface is in communication connection with an image input interface of the quality detection system;
the point inspection panel comprises a point inspection button; and after the point inspection button of the point inspection panel is triggered, extracting a typical defect tobacco bale image template from the data server, and inputting the typical defect tobacco bale image template into an image input interface of the quality inspection system through the communication interface.
It can be seen that the above-mentioned apparatus does not necessarily include an image recognition module, and is only used for storing the marking result of the image recognition module, and in other embodiments, the image recognition module is integrated into the spot inspection apparatus. The point inspection device is integrated or not integrated with the image identification module and is in communication connection with the data server so as to read the cigarette packet image templates of the cigarette packet image template library and write mark attributes into the cigarette packet image templates of the cigarette packet image template library; the marking property includes a typical defect.
According to the above description of the technical solution of the present invention, it can be easily understood that the technical effects of the present invention include:
1. the effectiveness is that the point inspection effect of the point inspection method of the invention achieves the effect of artificial simulation test; even make up for the difference of detection effect caused by the difference of manual point detection methods;
2. the method has the advantages that the method is simple to operate and easy to realize, and the analog point inspection machine is not convenient to set particularly for a high-speed wrapping machine;
3. the economy is different from the waste caused by a large amount of damage simulation tests in the prior art, and the point inspection method does not need to damage the cigarette packet and consume 0;
4. the safety is high, the possibility that defective products enter finished products by mistake in manual simulation point inspection is high, and the point inspection method completely avoids the risk.
Drawings
FIG. 1 is a system block diagram of a system for quality detection of cigarette making machine equipment in an embodiment of the present invention;
FIG. 2 is a flowchart of an example process of a spot check method according to an embodiment of the present invention.
Detailed Description
Firstly, it should be noted that, in the invention, the cigarette making machine equipment is a cigarette wrapping machine special for tobacco, the finished product is a small cigarette packet, the quality detection system of the cigarette making machine equipment is used for carrying out visual analysis on the small cigarette packet, if a problem occurs, the unqualified small cigarette packet is at least removed from the production line, or the problem occurs in the cigarette wrapping machine special for tobacco, the production line problem should be timely processed, and whether the machine is stopped or not is considered. The existing checking method of the cigarette machine equipment quality detection system aims to detect whether the quality detection system has functional problems or not and prevent the quality detection system from missing and mistakenly rejecting functions due to the equipment problems. However, in the era of high-speed cigarette packaging, the problem of matching of unexpected cigarette making machine equipment with the point inspection process of the production line occurs. For example, for the shift spot inspection of the high-speed production line, the shift spot inspection needs to be performed synchronously, on one hand, if each equipment performs spot inspection for 5 minutes each time, each responsible person is responsible for 6 equipment, and each person takes 30 minutes each time; each simulation test destroys 5 boxes of cigarettes and 20 devices, every two shifts are carried out, every day is 200 boxes of cigarettes, sometimes, every shift needs two times, the number is doubled, and the time cost of shifting is enlarged by a new production mode; at the same time. On the other hand, in the production process, the types of the defects are various and some defects are unforeseen, the starting and stopping states of production equipment are different every shift, the manual simulation of the defective cigarette packet is limited to the defects which often occur and are representative, the accidental defects or the defects which are extremely difficult to occur cannot be tested, the purpose of point inspection is difficult to achieve, and the quality hidden danger exists.
The following describes the pointing scheme and pointing device of the present invention with reference to the embodiments and the drawings. The embodiment is a point inspection method of a cigarette machine equipment quality detection system based on the technical improvement of the existing equipment.
The existing quality inspection system mainly comprises an imaging detector, a defect identification computer system and a window setting panel. The imaging detector comprises an image acquisition unit, an illumination unit and a controller, and is used for acquiring one or more cigarette packet images on a production line according to a received image acquisition setting signal; the defect identification computer system receives the cigarette packet image of the imaging detector and compares the cigarette packet image with a qualified image template of the computer system, if a visual difference larger than a detection threshold value exists, a defect prompt signal is output to the production line, and a rejection sensor of the production line is triggered to execute rejection action; the window setting panel is used for setting parameters of a monitoring window of the imaging detector through the defect identification computer system so as to adjust the acquisition requirement of the imaging detector at any time according to factors such as specifications, equipment conditions, production standards and the like and the detection requirement of the defect identification computer system.
Referring to fig. 1, the embodiment adds a separate program module to the existing defect recognition computer system, which includes: a user interface module, which is matched with a display to realize the point inspection panel of the invention; the database module is used for providing reading and writing service of the cigarette packet image template library; and the image recognition module is used for reading the cigarette packet image templates in the cigarette packet image template library and marking the cigarette packet image templates, and is provided with an independent storage space so as to store training data of the image recognition neural network.
It is easy to understand that in other embodiments, the point inspection device may be an independent device including a data server, a communication interface, and a point inspection panel, wherein the data server is used to establish a tobacco bale image template library so as to store tobacco bale image templates including qualified tobacco bale image templates and defective tobacco bale image templates and provide read-write services for the tobacco bale image templates, and the data server may be either embedded in the field or shared in the cloud; the communication interface is in communication connection with an image input interface of the quality detection system, and is integrated in the defect identification computer system, and the communication interface is in communication with an image input process or thread of the quality detection system by using internal sockets, APIs and other virtual interfaces so as to exchange data; the point inspection panel realizes that a user interface is displayed on the touch display by simulating a point inspection interface through the user interface module, and the user interface comprises a point inspection button; the image identification module is in communication connection with the data server, and is triggered to read the cigarette packet image templates in the cigarette packet image template library through an external request, and typical defect mark attributes are written into the cigarette packet image templates in the cigarette packet image template library.
In the embodiment, in a configuration stage, an unqualified cigarette packet template is collected firstly and stored in a cigarette packet image template library in advance to serve as a defective cigarette packet image template, an original qualified cigarette packet template is used or a qualified cigarette packet image is collected again to serve as a qualified cigarette packet image template and stored in a cigarette packet image template library, the attributes of the cigarette packet image template in the cigarette packet image template library at least comprise specifications, quality marks, collection parameters, defect characteristics, defect characteristic original confidence coefficients and defect characteristic mark confidence coefficients, and then an image recognition module is called to assign values to the defect characteristic confidence coefficients of all defect characteristics of the cigarette packet image template; when the cigarette packet template is used in the operation stage, the typical defect cigarette packet template can be called through a simulation point inspection interface in a user interface of a point inspection panel at any time under the conditions that the packaging machine does not stop and does not reduce the speed during operation, and a point inspection button is triggered; after the point inspection button is triggered, a user interface module of a point inspection panel extracts a typical defect tobacco bale image template from the data server and inputs the typical defect tobacco bale image template into an image input interface of the quality inspection system through the communication interface; so as to realize the functions of simulation detection, automatic point inspection and system verification of production line equipment such as a wrapping machine, an imaging detector, a defect identification computer system and the like in production.
The spot inspection method of the embodiment further comprises an updating stage; and in the updating stage, the image recognition module is retrained so that the image recognition module can re-mark the typical defective cigarette packet image template, and the typical defective cigarette packet image template is dynamically marked in the defective cigarette packet image template by the image recognition module. In order to realize dynamic updating, a defective cigarette packet image is acquired during the normal operation of a quality detection system, if the defective cigarette packet image is classified by an image identification module, a defective characteristic index is atypical, if the confidence coefficient of any defective characteristic of one acquired defective cigarette packet image is less than 0.95, the defective cigarette packet image template is stored in a cigarette packet image template library through a user interface of a point inspection panel, and during storage, a defective characteristic attribute and the original confidence coefficient of each defective characteristic are configured. In this embodiment, the image recognition module is pre-manufactured with a default image neural network unit, and evaluates the reliability of the difference part between the tobacco bale image and the standard tobacco bale image on each defect feature by using the defect feature as a classification. It is easy to understand that a large number of cigarette packet image templates with qualified or defective quality are stored in the cigarette packet image template and are manually marked with the original confidence degrees of the cigarette packet image templates on each defect feature, if the defect does not exist, the defect is 0, and the defect exists, the defect is 1. In an updating stage, the cigarette packet image template library of the supplementary sample is utilized to retrain the image neural network in the image recognition module so as to update the number of output nodes of the image neural network, the pooling layer connection structure or relevant parameters.
It can be seen that the image recognition of the embodiment uses a simple supervised learning over-scale to mark the cigarette packet image template in advance, so as to separate the training set and the verification set and iterate in the supervised learning. In other embodiments, more complex semi-supervised or unsupervised learning and corresponding image neural networks may be employed, including resist generation networks, generating virtual qualified cigarette packet image templates and defective cigarette packet image templates.
In some other embodiments, the image recognition module includes a first image neural network unit, and the first image neural network unit is trained by using a detection threshold value when a qualified cigarette packet image template is collected in a qualified cigarette packet image template set as a classification label, so that after training is completed, for a given qualified cigarette packet image, the first image neural network unit outputs the detection threshold value of the qualified cigarette packet image, and stores a new qualified cigarette packet image generated in a countermeasure generation set as the qualified cigarette packet image template under the new detection threshold value to the cigarette packet image template for the quality detection system or the image recognition module to call.
In some other embodiments, the image recognition module comprises a second image neural network unit, the defect status problem relates to window selection, the illumination attenuation problem relates to picture updating so as to extract the difference of the qualified cigarette packet, so that the image recognition module extracts the cigarette packet quality defect feature and corrects the qualified cigarette packet image template set and the defective product image template set. The second image neural network unit adopts an unsupervised learning neural network structure to automatically classify partial tobacco bale image templates of the tobacco bale image template set, wherein the classification is as follows, (missing 0.75, reverse wrapping 0.99, creasing 0.46, skewing 0.05, opening 0.92 and fouling 0.85). With the accumulation of the defective cigarette packet image templates in the cigarette packet image template library in the using process, the parameters of the image neural network unit of the image recognition module are updated through iterative training, the confidence intervals on the characteristics of each defect become narrower, the judgment results of the shapes (missing 0.01, anti-wrap 0.99, wrinkle 0.01, skew 0.01, opening 0.99 and fouling 0.99) are easy to mark, and therefore when the defects are called and the anti-wrap, opening or fouling is detected at a simulation point, the defective cigarette packet image templates are used as typical defective cigarette packet image templates of each defect.
Referring to fig. 2, it is a variant that includes a configuration phase and an operational phase, in which:
in a configuration stage, configuring a point inspection panel, an image identification module and a tobacco bale image template library for the quality detection system; training the image recognition module by using a cigarette packet image template of a cigarette packet image template library so that the image recognition module marks a typical defect cigarette packet image template in the cigarette packet image template; the cigarette packet image template comprises a qualified cigarette packet image template and a defective cigarette packet image template; the typical defects include missing, turned up, crumples, skewing, openings, and/or blemishes. In a configuration stage, marking defects for defective cigarette packet image templates in the cigarette packet image template library, dividing all cigarette packet image templates into a training set and a verification set, and training the first image neural network unit so that the first image neural network unit outputs the confidence coefficients of the defective cigarette packet images on the defects for the given defective cigarette packet images after the training is finished; the defective cigarette packet image template with the highest confidence level on a defect in the defective cigarette packet image templates of the cigarette packets with the same specification is marked as a typical defective cigarette packet image template of the cigarette packet with the specification under the defect.
In the operation stage, when analog detection, automatic point inspection and/or system validation are required, the typical defect cigarette packet image template is extracted through the point inspection panel to replace the cigarette packet image of the imaging detector and is sent to the quality detection system. In the operation stage, when the first image neural network unit reads that each defect of a defective cigarette packet image provided by the quality detection system is atypical, one or more defects are distributed to the defective cigarette packet image provided by the quality detection system and then are used as a new defective cigarette packet image template to be placed into the defective cigarette packet image template set.
When system verification is needed, the typical defect tobacco bale image template is extracted through the point inspection panel and is sent to the quality detection system, so that the defect identification system is judged to work normally. The method comprises two modes, wherein one mode is that other equipment on a production line is not operated, and a qualified cigarette packet image template and a defective cigarette packet image template with the same specification are selected to be randomly mixed so as to judge whether a current defect identification computer system works normally; and the other method is to merge production line test operation, and randomly insert a defective cigarette packet image template into a normal cigarette packet image stream so as to judge whether the current defect identification computer system works normally.
When simulation detection is needed, a typical defect tobacco bale image template under a typical defect is extracted through the point detection panel to replace a tobacco bale image and is sent to the quality detection system, so that the elimination function of the typical defect of the quality detection system is detected. Specifically, a typical defective cigarette packet image template applicable to the current process needs to be selected to replace an artificial cigarette packet according to the process description.
When automatic point inspection is needed, the typical defective cigarette packet image templates of the automatic point inspection in the current production are specified through the point inspection panel, the typical defective cigarette packet image templates of the cigarette packets with the same specification are extracted through the point inspection panel and inserted into the cigarette packet image stream to be sent to the quality inspection system, and therefore the elimination function of the defective cigarette packets by the quality inspection system during continuous production is detected conveniently. In other embodiments, when automatic point inspection is required, template images designated in the defect product image template set are extracted through a point inspection panel and inserted into an image acquisition stream of the quality detection system imaging detector for transmission;
specifically, referring to fig. 2, the automatic spot inspection process includes the following steps:
step 1, triggering light sources to turn on lights according to stations in sequence and obtaining a current cigarette packet image;
step 2, calling a typical defect cigarette packet image template of the cigarette packets with the same specification to replace the current cigarette packet image and sending the image to the quality detection system;
step 3, the quality detection system judges whether the defect exists, if not, the current production line is judged to be unqualified, and if so, the step 4 is executed;
step 4, triggering a rejection sensor, judging that the current production line is unqualified if the rejection sensor cannot be successfully triggered, and executing step 5 if the rejection sensor can be triggered;
and 5, finishing the automatic point inspection and judging that the current production line is qualified.
And 6, outputting an automatic point inspection report of the current production line.
In this embodiment, if the difference is not consistent, the influence of other equipment factors, such as the breakthrough influence of the LED random aging on the production line judgment threshold in the slow accumulation, is eliminated first. Taking the light source of the imaging detector as an example, adjusting the light source to detect again, if the abnormality is still judged, considering that the current production line can not carry out production, stopping the production line, and if the abnormality is recovered, recovering the light source and continuing normal automatic point inspection; if the images are not consistent, the current cigarette packet image is prompted to be abnormal, and the production line problem is continuously checked manually.
It will be appreciated that the same or different typical defective cigarette packet image templates, with or without the use of qualified cigarette packet image templates, may be used in performing the simulation test, the automatic point inspection and the system validation.
It is easy to understand that this embodiment is a set of processing program developed for the existing imaging detection device for the appearance of a small bag, and provides a fully automatic simulation point inspection method. In the embodiment, the standard modeling principle is adopted, two image models are pre-established, one is a qualified cigarette packet image model, the other is a defective cigarette packet image model, and the functions of simulation detection, automatic point inspection and system validation are realized by adopting two different model data analysis and comparison.
The software processing program adopts a neural network algorithm technology, the neural network algorithm has the functions of parallel computing, self-organizing and self-learning characteristics and global approximation, the software processing process of the invention mainly uses the self-organizing neural network to complete image characteristic extraction and modeling, and uses the fuzzy neural network to complete image correction and reconstruction. The program software adopts the programming of a high-level language VB.Net of a computer, and programming functions of Boolean conditions, attributes and events, delegation and event management, condition compilation and the like of the VB.Net are applied to complete the neural network algorithm.
Claims (10)
1. A point inspection method of a cigarette machine equipment quality detection system comprises a configuration stage and an operation stage, wherein:
in a configuration stage, configuring a point inspection panel, an image identification module and a tobacco bale image template library for the quality detection system; training the image recognition module by using a cigarette packet image template of a cigarette packet image template library so that the image recognition module marks a typical defect cigarette packet image template in the cigarette packet image template; the cigarette packet image template comprises a qualified cigarette packet image template and a defective cigarette packet image template; the typical defects include missing, turned, wrinkled, skewed, open, and/or stained;
in the operation stage, when analog detection, automatic point inspection and/or system validation are required, the typical defect cigarette packet image template is extracted through the point inspection panel to replace the cigarette packet image of the imaging detector and is sent to the quality detection system.
2. The spot inspection method according to claim 1, characterized in that: comprises an updating stage; and in the updating stage, the image recognition module is retrained so that the image recognition module can re-mark the typical defective cigarette packet image template.
3. The spot inspection method according to claim 2, wherein: and during the normal operation of the quality detection system, acquiring a defective cigarette packet image, and if the defective cigarette packet image is classified by the image identification module and the defect characteristic index is atypical, storing the defective cigarette packet image as a defective cigarette packet image template in the cigarette packet image template library.
4. The spot inspection method according to claim 3, wherein: the image recognition module comprises a first image neural network unit;
in the configuration stage, defects are marked for defective cigarette packet image templates in the cigarette packet image template library, all cigarette packet image templates are divided into a training set and a verification set, and the first image neural network unit is trained, so that for a given defective cigarette packet image after training is finished, the first image neural network unit outputs the confidence coefficient of the defective cigarette packet image on each defect; the defect cigarette packet image template with the highest confidence level on a defect in the defect cigarette packet image templates of cigarette packets with the same specification is marked as a typical defect cigarette packet image template of the cigarette packet with the specification under the defect;
in the operation stage, when the first image neural network unit reads that each defect of a defective cigarette packet image provided by the quality detection system is atypical, one or more defects are distributed to the first image neural network unit and then the first image neural network unit is used as a new defective cigarette packet image template to be placed into the defective cigarette packet image template set.
5. The spot inspection method according to claim 1, characterized in that: when simulation detection is needed, a typical defect tobacco bale image template under a typical defect is extracted through the point detection panel to replace a tobacco bale image and is sent to the quality detection system, so that the elimination function of the typical defect of the quality detection system is detected.
6. The spot inspection method according to claim 1, characterized in that: when automatic point inspection is needed, typical defective cigarette packet image templates of cigarette packets with the same specification are extracted through the point inspection panel and inserted into the cigarette packet image stream to be sent to the quality detection system, so that the elimination function of the defective cigarette packets by the quality detection system during continuous production is detected.
7. The spot inspection method according to claim 1, characterized in that: when system verification is needed, the typical defect cigarette packet image template is extracted through the point inspection panel and is sent to the quality inspection system without operating other equipment on a production line.
8. The point inspection method according to claim 6, wherein the automatic point inspection process comprises the steps of:
step 1, triggering light sources to turn on lights according to stations in sequence and obtaining a current cigarette packet image;
step 2, calling a typical defect cigarette packet image template of the cigarette packets with the same specification to replace the current cigarette packet image and sending the image to the quality detection system;
step 3, the quality detection system judges whether the defect exists, if not, the current production line is judged to be unqualified, and if so, the step 4 is executed;
step 4, triggering a rejection sensor, judging that the current production line is unqualified if the rejection sensor cannot be successfully triggered, and executing step 5 if the rejection sensor can be triggered;
and 5, finishing the automatic point inspection and judging that the current production line is qualified.
And 6, outputting an automatic point inspection report of the current production line.
9. A cigarette machine equipment quality detection system point inspection device comprises a data server, a communication interface and a point inspection panel;
the data server is provided with a tobacco bale image template library; the tobacco bale image template library stores qualified tobacco bale image templates and defective tobacco bale image templates; the defective cigarette packet image template comprises a typical defective cigarette packet image template; the typical defective cigarette packet image template is dynamically marked in the defective cigarette packet image template by an image identification module;
the communication interface is in communication connection with an image input interface of the quality detection system;
the point inspection panel comprises a point inspection button; and after the point inspection button of the point inspection panel is triggered, extracting a typical defect tobacco bale image template from the data server, and inputting the typical defect tobacco bale image template into an image input interface of the quality inspection system through the communication interface.
10. The cigarette making machine equipment quality detection system spot check device according to claim 9, comprising an image recognition module; the image identification module is in communication connection with the data server so as to read the cigarette packet image templates of the cigarette packet image template library and write mark attributes into the cigarette packet image templates of the cigarette packet image template library; the marking property includes a typical defect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011369942.9A CN112557399B (en) | 2020-11-30 | 2020-11-30 | Point inspection method and device for smoke machine equipment quality detection system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011369942.9A CN112557399B (en) | 2020-11-30 | 2020-11-30 | Point inspection method and device for smoke machine equipment quality detection system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112557399A true CN112557399A (en) | 2021-03-26 |
| CN112557399B CN112557399B (en) | 2022-11-22 |
Family
ID=75045251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011369942.9A Active CN112557399B (en) | 2020-11-30 | 2020-11-30 | Point inspection method and device for smoke machine equipment quality detection system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112557399B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113942714A (en) * | 2021-11-10 | 2022-01-18 | 云南烟草机械有限责任公司 | Cigarette packet dismantling and inspecting device capable of performing appearance inspection on cigarettes and dismantling and inspecting method thereof |
| CN117890379A (en) * | 2024-02-22 | 2024-04-16 | 宁德时代新能源科技股份有限公司 | Tab ultrasonic welding detection system and spot inspection method |
Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110157226A1 (en) * | 2009-12-29 | 2011-06-30 | Ptucha Raymond W | Display system for personalized consumer goods |
| JP2012194059A (en) * | 2011-03-16 | 2012-10-11 | Tokyo Electron Ltd | Image creation method, substrate inspection method, recording medium with program for executing image creation method or substrate inspection method recorded thereon, and substrate inspection device |
| CN103512888A (en) * | 2013-06-05 | 2014-01-15 | 北京化工大学 | Cigarette packet seal defect detecting system based on image identification technology |
| CN108107054A (en) * | 2017-12-08 | 2018-06-01 | 云南昆船设计研究院 | A kind of online cigarette defective vision detecting system and method |
| US20180342050A1 (en) * | 2016-04-28 | 2018-11-29 | Yougetitback Limited | System and method for detection of mobile device fault conditions |
| CN109241983A (en) * | 2018-09-13 | 2019-01-18 | 电子科技大学 | A kind of cigarette image-recognizing method of image procossing in conjunction with neural network |
| CN109345513A (en) * | 2018-09-13 | 2019-02-15 | 红云红河烟草(集团)有限责任公司 | A kind of cigarette package defect inspection method with cigarette packet Attitude Calculation |
| CN109507209A (en) * | 2019-01-22 | 2019-03-22 | 中科院金华信息技术有限公司 | A kind of film printing defect detecting system and method |
| CN109911280A (en) * | 2019-01-18 | 2019-06-21 | 红云红河烟草(集团)有限责任公司 | A kind of detection method and system of cigarette packet open defect |
| US20190303717A1 (en) * | 2018-03-28 | 2019-10-03 | Kla-Tencor Corporation | Training a neural network for defect detection in low resolution images |
| CN110378618A (en) * | 2019-07-26 | 2019-10-25 | 杭州安脉盛智能技术有限公司 | Quality evaluating method and system based on online pipe tobacco surface defects detection |
| KR102043664B1 (en) * | 2019-04-29 | 2019-12-05 | 주식회사 에이치비테크놀러지 | Auto Repair System and Method for Detecting and Repairing Defects in a Substrate |
| WO2019233341A1 (en) * | 2018-06-08 | 2019-12-12 | Oppo广东移动通信有限公司 | Image processing method and apparatus, computer readable storage medium, and computer device |
| US20200005449A1 (en) * | 2018-06-29 | 2020-01-02 | Utechzone Co., Ltd. | Defect inspection system and method using artificil intelligence |
| CN110907472A (en) * | 2019-11-15 | 2020-03-24 | 河南中烟工业有限责任公司 | Image recognition system for online cigarette product packaging |
| CN111096473A (en) * | 2019-12-23 | 2020-05-05 | 中国电子科技集团公司第四十一研究所 | Cigarette filter tip defect online detection device and method of cigarette making and plug assembling machine |
| CN111122597A (en) * | 2019-12-12 | 2020-05-08 | 河南中烟工业有限责任公司 | Visual detection system and detection method for defects of cigarette packets of packaging machine |
| CN111721947A (en) * | 2019-03-21 | 2020-09-29 | 红塔烟草(集团)有限责任公司 | Automatic analysis and optimization system for cigarette packet image detection |
| US20200349875A1 (en) * | 2018-07-02 | 2020-11-05 | Beijing Baidu Netcom Science Technology Co., Ltd. | Display screen quality detection method, apparatus, electronic device and storage medium |
-
2020
- 2020-11-30 CN CN202011369942.9A patent/CN112557399B/en active Active
Patent Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110157226A1 (en) * | 2009-12-29 | 2011-06-30 | Ptucha Raymond W | Display system for personalized consumer goods |
| JP2012194059A (en) * | 2011-03-16 | 2012-10-11 | Tokyo Electron Ltd | Image creation method, substrate inspection method, recording medium with program for executing image creation method or substrate inspection method recorded thereon, and substrate inspection device |
| CN103512888A (en) * | 2013-06-05 | 2014-01-15 | 北京化工大学 | Cigarette packet seal defect detecting system based on image identification technology |
| US20180342050A1 (en) * | 2016-04-28 | 2018-11-29 | Yougetitback Limited | System and method for detection of mobile device fault conditions |
| CN108107054A (en) * | 2017-12-08 | 2018-06-01 | 云南昆船设计研究院 | A kind of online cigarette defective vision detecting system and method |
| US20190303717A1 (en) * | 2018-03-28 | 2019-10-03 | Kla-Tencor Corporation | Training a neural network for defect detection in low resolution images |
| WO2019233341A1 (en) * | 2018-06-08 | 2019-12-12 | Oppo广东移动通信有限公司 | Image processing method and apparatus, computer readable storage medium, and computer device |
| US20200005449A1 (en) * | 2018-06-29 | 2020-01-02 | Utechzone Co., Ltd. | Defect inspection system and method using artificil intelligence |
| US20200349875A1 (en) * | 2018-07-02 | 2020-11-05 | Beijing Baidu Netcom Science Technology Co., Ltd. | Display screen quality detection method, apparatus, electronic device and storage medium |
| CN109345513A (en) * | 2018-09-13 | 2019-02-15 | 红云红河烟草(集团)有限责任公司 | A kind of cigarette package defect inspection method with cigarette packet Attitude Calculation |
| CN109241983A (en) * | 2018-09-13 | 2019-01-18 | 电子科技大学 | A kind of cigarette image-recognizing method of image procossing in conjunction with neural network |
| CN109911280A (en) * | 2019-01-18 | 2019-06-21 | 红云红河烟草(集团)有限责任公司 | A kind of detection method and system of cigarette packet open defect |
| CN109507209A (en) * | 2019-01-22 | 2019-03-22 | 中科院金华信息技术有限公司 | A kind of film printing defect detecting system and method |
| CN111721947A (en) * | 2019-03-21 | 2020-09-29 | 红塔烟草(集团)有限责任公司 | Automatic analysis and optimization system for cigarette packet image detection |
| KR102043664B1 (en) * | 2019-04-29 | 2019-12-05 | 주식회사 에이치비테크놀러지 | Auto Repair System and Method for Detecting and Repairing Defects in a Substrate |
| CN110378618A (en) * | 2019-07-26 | 2019-10-25 | 杭州安脉盛智能技术有限公司 | Quality evaluating method and system based on online pipe tobacco surface defects detection |
| CN110907472A (en) * | 2019-11-15 | 2020-03-24 | 河南中烟工业有限责任公司 | Image recognition system for online cigarette product packaging |
| CN111122597A (en) * | 2019-12-12 | 2020-05-08 | 河南中烟工业有限责任公司 | Visual detection system and detection method for defects of cigarette packets of packaging machine |
| CN111096473A (en) * | 2019-12-23 | 2020-05-05 | 中国电子科技集团公司第四十一研究所 | Cigarette filter tip defect online detection device and method of cigarette making and plug assembling machine |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113942714A (en) * | 2021-11-10 | 2022-01-18 | 云南烟草机械有限责任公司 | Cigarette packet dismantling and inspecting device capable of performing appearance inspection on cigarettes and dismantling and inspecting method thereof |
| CN117890379A (en) * | 2024-02-22 | 2024-04-16 | 宁德时代新能源科技股份有限公司 | Tab ultrasonic welding detection system and spot inspection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112557399B (en) | 2022-11-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112557399B (en) | Point inspection method and device for smoke machine equipment quality detection system | |
| CN105955889B (en) | A kind of graphical interfaces automated testing method | |
| CN110274908A (en) | Flaw detection apparatus, defect detecting method and computer readable recording medium | |
| CN110648490B (en) | Multi-factor flame identification method suitable for embedded platform | |
| AU2008337296B2 (en) | Assisting failure mode and effects analysis of a system comprising a plurality of components | |
| CN108809948A (en) | A kind of abnormal network connecting detection method based on deep learning | |
| CN110378618A (en) | Quality evaluating method and system based on online pipe tobacco surface defects detection | |
| CN109918296A (en) | Automatic software test method and device | |
| KR102174424B1 (en) | Method for Inspecting Component basesd Server and system and apparatus therefor | |
| CN115128033A (en) | Tobacco leaf detection method, device and system and storage medium | |
| CN109685756A (en) | Image feature automatic identifier, system and method | |
| US20220284699A1 (en) | System and method of object detection using ai deep learning models | |
| KR20230023263A (en) | Deep learning-based sewerage defect detection method and apparatus | |
| DE102020215930A1 (en) | VISUAL ANALYSIS PLATFORM FOR UPDATING OBJECT DETECTION MODELS IN AUTONOMOUS DRIVING APPLICATIONS | |
| CN106934425A (en) | A kind of industrial products counting method based on deep learning | |
| CN206286224U (en) | A kind of smoke box recycles mistake proofing error correction device | |
| WO2021170518A1 (en) | Automatic spreader for fabrics comprising an identification system for defects and relative method | |
| CN107729635A (en) | Semiconductor chip qualification inspection method and device | |
| CN113192068A (en) | AI visual inspection system for printed wiring board | |
| CN111707677A (en) | PCB appearance detection method and device | |
| CN110252686A (en) | The determination method and device of remote controler quality | |
| US11797646B2 (en) | Method, electronic device, and computer program product for standardizing image annotation | |
| CN111366582A (en) | Automatic judgment method based on automatic optical detection | |
| JP4155497B2 (en) | Defect classification method, program, and defect classification apparatus | |
| CN117935174B (en) | Intelligent management system and method for vacuum bag film production line |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |