CN116071298A - Tool loss identification method and device based on target detection model - Google Patents

Tool loss identification method and device based on target detection model Download PDF

Info

Publication number
CN116071298A
CN116071298A CN202211521759.5A CN202211521759A CN116071298A CN 116071298 A CN116071298 A CN 116071298A CN 202211521759 A CN202211521759 A CN 202211521759A CN 116071298 A CN116071298 A CN 116071298A
Authority
CN
China
Prior art keywords
cutter
tool
image data
data
groups
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.)
Pending
Application number
CN202211521759.5A
Other languages
Chinese (zh)
Inventor
杨丹丹
李冰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Suzhou Nuowei Industrial Technology Service Co ltd
Original Assignee
Suzhou Nuowei Industrial Technology Service Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Suzhou Nuowei Industrial Technology Service Co ltd filed Critical Suzhou Nuowei Industrial Technology Service Co ltd
Priority to CN202211521759.5A priority Critical patent/CN116071298A/en
Publication of CN116071298A publication Critical patent/CN116071298A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/761—Proximity, similarity or dissimilarity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection
    • 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)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a tool loss identification method and device based on a target detection model. The method comprises the following steps: acquiring image data of the pre-classified cutter; extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter; respectively comparing the similarity between the plurality of sets of characteristic data of the cutter and the plurality of sets of characteristic data of the preset cutter to generate a plurality of similarity values of the plurality of sets of characteristic data of the cutter; and determining the loss level of the cutter according to a plurality of similarity values of a plurality of groups of characteristic data of the cutter. By the tool loss identification method based on the target detection model, tool loss identification detection with lower cost is realized.

Description

Tool loss identification method and device based on target detection model
Technical Field
The invention relates to the field of tool detection, in particular to a tool loss identification method and device based on a target detection model.
Background
The current manufacturing industry is indispensable to use and recycle the cutting tools, so far, various tool cabinets and various borrowing and recycling processes appearing on the market also show that the factory is always searching for a more advanced, more convenient and intelligent management method for recycling and recycling the tools, and the management of the residual service life of the tools is an aspect of cost reduction of each factory, and more of the factories are currently used for manually identifying and analyzing the residual service life of old tools and coping values and the like.
In implementing the prior art process, the inventors found that:
when the cutter is returned after being used, manual damage assessment is mostly adopted to determine the subsequent availability of the cutter. The manual damage determination has insufficient experience of workers, and whether the recovered cutter can be polished or not can not be completely and correctly identified, so that the recovered cutter can be directly determined as waste, and a brand new cutter is required to be reused, thereby increasing the cost of the cutter.
Therefore, it is desirable to provide a method for identifying and detecting tool loss at a low cost to solve the problem of cost increase caused by false tool detection in the prior art.
Disclosure of Invention
The invention mainly solves the technical problem of cost rise caused by false detection of a cutter in the prior art.
In order to solve the technical problems, the invention adopts a technical scheme that:
the utility model provides a cutter loss identification method based on a target detection model, which comprises the following steps:
acquiring image data of the pre-classified cutter;
extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter;
respectively comparing the similarity between the plurality of sets of characteristic data of the cutter and the plurality of sets of characteristic data of the preset cutter to generate a plurality of similarity values of the plurality of sets of characteristic data of the cutter;
and determining the loss level of the cutter according to a plurality of similarity values of a plurality of groups of characteristic data of the cutter.
In a preferred embodiment of the present invention, the acquiring image data of the pre-classified tool specifically includes:
acquiring an initial image of the cutter to generate cutter initial image data;
pre-classifying the cutter initial image data by adopting a cutter classification model to generate pre-classified cutter image data;
at least bit image data and milling cutter image data are prestored in the cutter classification model;
the pre-classifying the cutter initial image data by adopting a cutter classification model specifically comprises the following steps:
comparing the cutter initial image data with a plurality of cutter image data stored in advance to generate comparison data values of a plurality of cutters;
and determining the image data of the pre-classified cutter according to the comparison data value.
In a preferred embodiment of the present invention, the sets of characteristic data of the tool include:
profile group features of the tool;
coating set characteristics of the tool;
cutting group characteristics of the cutter;
the profile group features of the cutter at least comprise shape features of the cutter and profile features of the cutter.
In a preferred embodiment of the invention, sets of characteristic data of the tool in the image data of the tool are extracted at least by means of a convolutional neural network.
In a preferred embodiment of the present invention, determining the wear level of the tool according to the similarity values of the several sets of features of the tool specifically includes:
respectively carrying out standardization processing on the similarity values of a plurality of groups of characteristics of the cutter to generate the similarity values of a plurality of groups of characteristics of the cutter after the standardization processing;
respectively carrying out weight assignment on a plurality of groups of characteristics of the cutter;
and determining the loss level of the cutter according to the similarity value between the weight assignment of the plurality of groups of characteristics of the cutter and the plurality of groups of characteristics of the cutter after the standardized processing.
In a preferred embodiment of the present invention, the normalizing process is performed on the similarity values of the plurality of sets of features of the tool, and the generating the normalized similarity values of the plurality of sets of features of the tool specifically includes:
and respectively carrying out normalization processing on the similarity values of the plurality of groups of features to generate the similarity values of the plurality of groups of features after normalization processing.
In a preferred embodiment of the present invention, after determining the wear level of the tool according to the similarity values of the sets of features of the tool, the method further comprises:
and (3) carrying out one of maintenance treatment, scrapping treatment and no treatment on the cutter according to the loss grade of the cutter.
The application child provides a cutter loss recognition device based on target detection model, includes:
the acquisition module is used for acquiring image data of the pre-classified cutter;
the extraction module is used for extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter;
the comparison module is used for respectively comparing the similarity between the plurality of sets of characteristic data of the cutter and the plurality of sets of characteristic data of the preset cutter to generate a plurality of similarity values of the plurality of sets of characteristic data of the cutter;
and the determining module is used for determining the loss level of the cutter according to a plurality of similarity values of a plurality of groups of characteristic data of the cutter.
The beneficial effects of the invention are as follows: through the tool loss identification method and device based on the target detection model, the problem that the tool is possibly misjudged as waste can be reduced, the rising of the tool cost is reduced, and meanwhile, the time cost of manual identification is greatly reduced. Meanwhile, the method can accurately improve the time of an auditing link and the complexity in the process, the service life of the cutter can be prolonged by reasonably polishing the cutter, and the improvement of time cost, labor cost and workpiece recycling cost of enterprises is facilitated.
Drawings
FIG. 1 is a schematic flow diagram of a method for identifying tool loss based on a target detection model;
FIG. 2 is a block diagram of a tool loss recognition device based on a target detection model;
the components in the drawings are marked as follows:
a tool loss recognition device-100 based on a target detection model; obtaining a module-10; extracting module-20; a comparison module-30; a determination module-40.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making clear and defining the scope of the present invention.
Referring to fig. 1, a method for identifying tool loss based on a target detection model provided in the present application specifically includes:
s110: and acquiring image data of the tool after the pre-classification.
Specifically, the image data of the tool after the pre-classification may be understood as being obtained after the image data of the tool is initially classified. In a preferred embodiment provided in the present application, the acquiring image data of the pre-classified tool specifically includes: acquiring an initial image of the cutter to generate cutter initial image data; pre-classifying the cutter initial image data by adopting a cutter classification model to generate pre-classified cutter image data; at least drill bit image data and milling cutter image data are prestored in the cutter classification model. The pre-classifying the cutter initial image data by adopting a cutter classification model specifically comprises the following steps: comparing the cutter initial image data with a plurality of cutter image data stored in advance to generate comparison data values of a plurality of cutters; and determining the image data of the pre-classified cutter according to the comparison data value.
It should be noted that, in addition to the method of pre-classifying the image data of the tool described above, the corresponding classification of the tool may be performed manually according to the type of the tool. For example, the returned tools are manually placed on corresponding tool cabinet return tracks for acquisition of image data of the tools.
It will be appreciated that the pre-classification of the tools may be either by classifying the initial images of the tools using a tool classification model or by manually placing the tools in the corresponding tool chest return trajectory. It will also be appreciated that the classification of the initial image of the tool by using the tool classification model may be a normal use case approach, whereas the pre-classification by manually placing the tool in the corresponding tool chest return track is used for manual processing that the approach cannot do. It can be understood that by setting the method of pre-classifying multiple tools, the image data of the tools after the pre-classifying in the step S110 can have multiple processing methods, so that the suspension of the loss identification of the tools caused by one of the methods when the other method cannot run is avoided.
It should be noted that, the cutter classification model is adopted to perform pre-classification processing on the cutter initial image data, and data of a plurality of cutter type images are preset in the cutter classification model in the image data of the cutter after pre-classification. For example, the tool classification model is provided with drill bit image data, milling cutter image data and tap image data, and each image data comprises image data of the tool in various views. That is, regardless of the state of the acquired initial image data of the tool, there is data of a state image corresponding thereto. For example, the drill bit image data stores therein front view data, left view data, right view image data, and front view data, left view data, right view image data of the drill bit in the transverse direction, and when the photographed tool initial image data is right view image data of the transverse direction, it can be classified as right view image data of the drill bit in the transverse direction.
It should be further noted that, in the pre-classifying the tool-initial-image data by using the tool classification model, the comparison data value may be understood as whether the tool-initial-image data is highest in the left-view comparison data value in the standing and standing states of the tool in the tool-initial-image data and the pre-stored tool-image data. Of course, the image data of the tool after the pre-classification can also be determined according to the pre-contrast data values of the pre-stored contrast data values of the tool in other states of the tool in the plurality of tool image data.
It should be noted that the acquisition of the image data of the pre-classified tool may be performed by R-CNN in the target detection algorithm or by YOLO. The generic name of R-CNN is Region-CNN, which is an algorithm that applies deep learning to target detection. The R-CNN is based on Convolutional Neural Network (CNN), linear regression, support Vector Machine (SVM) and other algorithms, and the target detection technology is realized. Compared with the R-CNN algorithm, the YOLO algorithm has relatively small calculated amount while realizing the functions, and can realize real-time detection of the target.
S120: extracting a plurality of sets of characteristic data of the cutter in the image data of the cutter.
Specifically, the plurality of sets of characteristic data of the cutter specifically include: profile group features of the tool; coating set characteristics of the tool; cutting group characteristics of the cutter; the profile group features of the cutter at least comprise shape features of the cutter and profile features of the cutter.
It should be noted that the profile group features of the tool include, in particular, the shape features of the tool, and the profile features of the tool. It will be appreciated that the extracted shape features of the tool may be similar to or significantly missing from the shape features of the entirely new tool. Likewise, the coating of the tool in the coating set characteristic of the tool may be similar to the shape characteristic of the completely new tool or may be a characteristic of coating discoloration. The cutting group characteristics of the cutter can be similar to the shape characteristics of the brand new cutter, and can also be the deformation and missing characteristics of main parts such as teeth, blades and the like.
In another embodiment, sets of feature data of the tool in the image data of the tool are extracted at least by a convolutional neural network.
Specifically, extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter is realized through a convolutional neural network. May be based on convolutional neural networks in the YOLO algorithm. The concrete steps are as follows: the method comprises the steps of extracting profile group characteristics of a cutter, extracting coating group characteristics of the cutter and extracting cutting group characteristics of the cutter. The extraction of the profile group features of the tool may be expressed specifically as: dividing image data of the cutter into SxS grids; predicting position information and confidence information of each grid; setting a threshold according to the position information and the confidence information of each grid, setting the threshold, filtering out the boxes with low scores, and performing NMS processing on the reserved boxes to obtain a final detection result. Correspondingly, the network model extracts the characteristics of the cutter profile group from the image data of the cutter through convolution of a plurality of convolution layers, and predicts output probability and coordinate values through the full connection layer. It will be appreciated that the extraction of profile set features of the tool, i.e. the identification of profile features of the tool. And extracting the coating group characteristics of the cutter, namely extracting the coating color of the cutter. And extracting the characteristics of the cutting groups of the cutters, and extracting the characteristics of the teeth and the blades of the cutters by a machine.
It should also be noted that in the extraction of several sets of feature data of the tool in the image data of the tool, the YOLO algorithm has been preset with a variety of profile features of the tool, coating color features of the tool, features of the teeth, edges of the tool. The characteristics of the teeth and the blades of the cutter can be the characteristics of the intact teeth and the blades, or the characteristics of the teeth and the blades with gaps.
S130: and respectively comparing the similarity between the plurality of sets of characteristic data of the cutter and the plurality of sets of characteristic data of the preset cutter to generate a plurality of similarity values of the plurality of sets of characteristic data of the cutter.
Specifically, similarity comparison is performed, namely, comparing a plurality of sets of acquired characteristic data of the cutter with a plurality of sets of characteristic data of a preset cutter. Presetting profile group characteristic data, tool coating data and cutting group characteristic data of a tool, wherein the profile group characteristic data, the tool coating data and the cutting group characteristic data are of the complete tool, in a plurality of groups of characteristic data of the preset tool. And obtaining a plurality of similarity values of the profile group characteristic data of the cutter, the cutter coating data and the cutting group characteristic data of the cutter through similarity comparison.
S140: and determining the loss level of the cutter according to a plurality of similarity values of a plurality of groups of characteristic data of the cutter.
In another embodiment, determining the wear level of the tool according to the similarity values of the sets of features of the tool specifically includes: respectively carrying out standardization processing on the similarity values of a plurality of groups of characteristics of the cutter to generate the similarity values of a plurality of groups of characteristics of the cutter after the standardization processing; respectively carrying out weight assignment on a plurality of groups of characteristics of the cutter; and determining the loss level of the cutter according to the similarity value between the weight assignment of the plurality of groups of characteristics of the cutter and the plurality of groups of characteristics of the cutter after the standardized processing.
Specifically, the similarity values of the plurality of groups of features of the tool are respectively subjected to standardization processing, and the similarity values of the plurality of groups of features of the tool after the standardization processing are generated may be obtained by respectively carrying out the standardization processing on the similarity values of the plurality of groups of features, so as to generate the similarity values of the plurality of groups of features after the standardization processing.
The data standardization is to convert the original data according to a certain proportion by a certain mathematical transformation mode, so that the original data falls into a small specific interval, for example, an interval of 0-1 or-1, the difference of characteristic attributes such as properties, dimensions, orders of magnitude and the like among different variables is eliminated, the original data is converted into a dimensionless relative value, namely, a standardized value, and the values of all indexes are in the same number level, thereby being convenient for the indexes of different units or orders of magnitude to be comprehensively analyzed and compared. The normalization process is a way of simplifying the calculation, namely, an expression with dimension is transformed into a non-dimension expression to become a scalar.
The weight refers to the importance of a factor or index relative to a thing, which is different from the general proportion, and is represented by not only the percentage of the factor or index, but also the relative importance of the factor or index, and tends to contribute to the degree or importance.
Specifically, each of the profile set characteristics of the tool, the coating set characteristics of the tool, and the cutting set characteristics of the tool has an important effect on whether the tool can be directly recycled for use. The sizes of the influences on whether the cutter can be recycled are different. For example, a cutter is a scrapped cutter in which the profile set features of the cutter are the deformed profile or the cutting teeth or cutting edges of the cutting set features of the cutter are completely ground flat. In a preferred embodiment provided in the application, firstly, normalization processing is performed on a plurality of similarity values of profile group characteristics of a cutter, coating group characteristics of the cutter and cutting group characteristics of the cutter, which may be that 3 characteristics occupy 1/3, and then weight assignment is performed on subsequent use influences of the cutter according to the profile group characteristics of the cutter, the coating group characteristics of the cutter and the cutting group characteristics of the cutter, so as to determine a loss level of the cutter. For example, the weights of the profile group feature of the tool, the coating group feature of the tool and the cutting group feature of the tool are respectively 4, 2.5 and 3.5, then the normalized value is multiplied by the weight assignment, finally the evaluation value of the tool is obtained, and the loss level of the tool is determined according to the evaluation value.
In another embodiment, after determining the wear level of the tool according to the similarity values of the sets of features of the tool, the method further comprises: and (3) carrying out one of maintenance treatment, scrapping treatment and no treatment on the cutter according to the loss grade of the cutter.
Specifically, according to the evaluation values of the cutters, which are obtained by the similarity values of the plurality of groups of characteristics of the cutters, the loss level of the cutters is determined. Specifically, the method can be expressed as follows: when the evaluation value of the cutter is more than 2, processing is not needed; when the evaluation value of the cutter is 1-2, performing maintenance treatment; and when the evaluation value of the cutter is lower than 1, scrapping.
It should also be noted that, depending on the wear level of the tool, the tool may be subjected to one of maintenance, scrapping, and no-need processing, or may be subjected to one of maintenance, scrapping, and no-need processing manually through the evaluation value of the tool. When the cutter is scrapped, the cutter slides into the corresponding scrapped storage box according to the track arrangement, and recovery checking is completed. When the cutter is subjected to maintenance treatment, the cutter slides into the corresponding repairable storage box according to the track arrangement, and recovery and verification are completed.
Referring to fig. 2, a tool loss recognition device based on a target detection model provided in the present application includes:
an acquisition module 10 for acquiring image data of the pre-classified tool.
Specifically, the image data of the tool after the pre-classification may be understood as being obtained after the image data of the tool is initially classified. In a preferred embodiment provided in the present application, the acquiring image data of the pre-classified tool specifically includes: acquiring an initial image of the cutter to generate cutter initial image data; pre-classifying the cutter initial image data by adopting a cutter classification model to generate pre-classified cutter image data; at least drill bit image data and milling cutter image data are prestored in the cutter classification model. The pre-classifying the cutter initial image data by adopting a cutter classification model specifically comprises the following steps: comparing the cutter initial image data with a plurality of cutter image data stored in advance to generate comparison data values of a plurality of cutters; and determining the image data of the pre-classified cutter according to the comparison data value.
It should be noted that, in addition to the method of pre-classifying the image data of the tool described above, the corresponding classification of the tool may be performed manually according to the type of the tool. For example, the returned tools are manually placed on corresponding tool cabinet return tracks for acquisition of image data of the tools.
It will be appreciated that the pre-classification of the tools may be either by classifying the initial images of the tools using a tool classification model or by manually placing the tools in the corresponding tool chest return trajectory. It will also be appreciated that the classification of the initial image of the tool by using the tool classification model may be a normal use case approach, whereas the pre-classification by manually placing the tool in the corresponding tool chest return track is used for manual processing that the approach cannot do. It can be understood that by setting the method of pre-classifying multiple tools, the image data of the tools after the pre-classifying in the step S110 can have multiple processing methods, so that the suspension of the loss identification of the tools caused by one of the methods when the other method cannot run is avoided.
It should be noted that, the cutter classification model is adopted to perform pre-classification processing on the cutter initial image data, and data of a plurality of cutter type images are preset in the cutter classification model in the image data of the cutter after pre-classification. For example, the tool classification model is provided with drill bit image data, milling cutter image data and tap image data, and each image data comprises image data of the tool in various views. That is, regardless of the state of the acquired initial image data of the tool, there is data of a state image corresponding thereto. For example, the drill bit image data stores therein front view data, left view data, right view image data, and front view data, left view data, right view image data of the drill bit in the transverse direction, and when the photographed tool initial image data is right view image data of the transverse direction, it can be classified as right view image data of the drill bit in the transverse direction.
It should be further noted that, in the pre-classifying the tool-initial-image data by using the tool classification model, the comparison data value may be understood as whether the left view of the tool-initial-image data and the pre-stored several-tool-image data in the standing state has the highest comparison data value in the comparison of the tool-initial-image data and the pre-stored several-tool-image data. Of course, the image data of the tool after the pre-classification can also be determined according to the pre-contrast data values of the pre-stored contrast data values of the tool in other states of the tool in the plurality of tool image data.
It should be noted that the acquisition of the image data of the pre-classified tool may be performed by R-CNN in the target detection algorithm or by YOLO. The generic name of R-CNN is Region-CNN, which is the first algorithm to successfully apply deep learning to target detection. The R-CNN is based on Convolutional Neural Network (CNN), linear regression, support Vector Machine (SVM) and other algorithms, and the target detection technology is realized. Compared with the R-CNN algorithm, the YOLO algorithm has relatively small calculated amount while realizing the functions, and can realize real-time detection of the target.
An extraction module 20 for extracting sets of feature data of the tool in the image data of the tool.
Specifically, the plurality of sets of characteristic data of the cutter specifically include: profile group features of the tool; coating set characteristics of the tool; cutting group characteristics of the cutter; the profile group features of the cutter at least comprise shape features of the cutter and profile features of the cutter.
It should be noted that the profile group features of the tool include, in particular, the shape features of the tool, and the profile features of the tool. It will be appreciated that the extracted shape features of the tool may be similar to or significantly missing from the shape features of the entirely new tool. Likewise, the coating of the tool in the coating set characteristic of the tool may be similar to the shape characteristic of the completely new tool or may be a characteristic of coating discoloration. The cutting group characteristics of the cutter can be similar to the shape characteristics of the brand new cutter, and can also be the deformation and missing characteristics of main parts such as teeth, blades and the like.
In another embodiment, sets of feature data of the tool in the image data of the tool are extracted at least by a convolutional neural network.
Specifically, extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter is realized through a convolutional neural network. May be based on convolutional neural networks in the YOLO algorithm. The concrete steps are as follows: the method comprises the steps of extracting profile group characteristics of a cutter, extracting coating group characteristics of the cutter and extracting cutting group characteristics of the cutter. The extraction of the profile group features of the tool may be expressed specifically as: dividing image data of the cutter into SxS grids; predicting position information and confidence information of each grid; setting a threshold according to the position information and the confidence information of each grid, setting the threshold, filtering out the boxes with low scores, and performing NMS processing on the reserved boxes to obtain a final detection result. Correspondingly, the network model extracts the characteristics of the cutter profile group from the image data of the cutter through convolution of a plurality of convolution layers, and predicts output probability and coordinate values through the full connection layer. It will be appreciated that the extraction of profile set features of the tool, i.e. the identification of profile features of the tool. And extracting the coating group characteristics of the cutter, namely extracting the coating color of the cutter. And extracting the characteristics of the cutting groups of the cutters, and extracting the characteristics of the teeth and the blades of the cutters by a machine.
It should also be noted that in the extraction of several sets of feature data of the tool in the image data of the tool, the YOLO algorithm has been preset with a variety of profile features of the tool, coating color features of the tool, features of the teeth, edges of the tool. The characteristics of the teeth and the blades of the cutter can be the characteristics of the intact teeth and the blades, or the characteristics of the teeth and the blades with gaps.
The comparison module 30 is configured to perform similarity comparison on the plurality of sets of feature data of the tool and a plurality of sets of feature data of a preset tool, respectively, to generate a plurality of similarity values of the plurality of sets of feature data of the tool.
Specifically, similarity comparison is performed, namely, comparing a plurality of sets of acquired characteristic data of the cutter with a plurality of sets of characteristic data of a preset cutter. Presetting profile group characteristic data, tool coating data and cutting group characteristic data of a tool, wherein the profile group characteristic data, the tool coating data and the cutting group characteristic data are of the complete tool, in a plurality of groups of characteristic data of the preset tool. And obtaining a plurality of similarity values of the profile group characteristic data of the cutter, the cutter coating data and the cutting group characteristic data of the cutter through similarity comparison.
A determining module 40, configured to determine a wear level of the tool according to a plurality of similarity values of a plurality of sets of characteristic data of the tool.
In another embodiment, determining the wear level of the tool according to the similarity values of the sets of features of the tool specifically includes: respectively carrying out standardization processing on the similarity values of a plurality of groups of characteristics of the cutter to generate the similarity values of a plurality of groups of characteristics of the cutter after the standardization processing; respectively carrying out weight assignment on a plurality of groups of characteristics of the cutter; and determining the loss level of the cutter according to the similarity value between the weight assignment of the plurality of groups of characteristics of the cutter and the plurality of groups of characteristics of the cutter after the standardized processing.
Specifically, the similarity values of the plurality of groups of features of the tool are respectively subjected to standardization processing, and the similarity values of the plurality of groups of features of the tool after the standardization processing are generated may be obtained by respectively carrying out the standardization processing on the similarity values of the plurality of groups of features, so as to generate the similarity values of the plurality of groups of features after the standardization processing.
The data standardization is to convert the original data according to a certain proportion by a certain mathematical transformation mode, so that the original data falls into a small specific interval, for example, an interval of 0-1 or-1, the difference of characteristic attributes such as properties, dimensions, orders of magnitude and the like among different variables is eliminated, the original data is converted into a dimensionless relative value, namely, a standardized value, and the values of all indexes are in the same number level, thereby being convenient for the indexes of different units or orders of magnitude to be comprehensively analyzed and compared. The normalization process is a way of simplifying the calculation, namely, an expression with dimension is transformed into a non-dimension expression to become a scalar.
The weight refers to the importance of a factor or index relative to a thing, which is different from the general proportion, and is represented by not only the percentage of the factor or index, but also the relative importance of the factor or index, and tends to contribute to the degree or importance.
Specifically, each of the profile set characteristics of the tool, the coating set characteristics of the tool, and the cutting set characteristics of the tool has an important effect on whether the tool can be directly recycled for use. The sizes of the influences on whether the cutter can be recycled are different. For example, a cutter is a scrapped cutter in which the profile set features of the cutter are the deformed profile or the cutting teeth or cutting edges of the cutting set features of the cutter are completely ground flat. In a preferred embodiment provided in the application, firstly, normalization processing is performed on a plurality of similarity values of profile group characteristics of a cutter, coating group characteristics of the cutter and cutting group characteristics of the cutter, which may be that 3 characteristics occupy 1/3, and then weight assignment is performed on subsequent use influences of the cutter according to the profile group characteristics of the cutter, the coating group characteristics of the cutter and the cutting group characteristics of the cutter, so as to determine a loss level of the cutter. For example, the weights of the profile group feature of the tool, the coating group feature of the tool and the cutting group feature of the tool are respectively 4, 2.5 and 3.5, then the normalized value is multiplied by the weight assignment, finally the evaluation value of the tool is obtained, and the loss level of the tool is determined according to the evaluation value.
In another embodiment, after determining the wear level of the tool according to the similarity values of the sets of features of the tool, the method further comprises: and (3) carrying out one of maintenance treatment, scrapping treatment and no treatment on the cutter according to the loss grade of the cutter.
Specifically, according to the evaluation values of the cutters, which are obtained by the similarity values of the plurality of groups of characteristics of the cutters, the loss level of the cutters is determined. Specifically, the method can be expressed as follows: when the evaluation value of the cutter is more than 2, processing is not needed; when the evaluation value of the cutter is 1-2, performing maintenance treatment; and when the evaluation value of the cutter is lower than 1, scrapping.
It should also be noted that, depending on the wear level of the tool, the tool may be subjected to one of maintenance, scrapping, and no-need processing, or may be subjected to one of maintenance, scrapping, and no-need processing manually through the evaluation value of the tool. When the cutter is scrapped, the cutter slides into the corresponding scrapped storage box according to the track arrangement, and recovery checking is completed. When the cutter is subjected to maintenance treatment, the cutter slides into the corresponding repairable storage box according to the track arrangement, and recovery and verification are completed.
The foregoing description is only illustrative of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or equivalent processes or direct or indirect application in other related technical fields are included in the scope of the present invention.

Claims (8)

1. The tool loss identification method based on the target detection model is characterized by comprising the following steps of:
acquiring image data of the pre-classified cutter;
extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter;
respectively comparing the similarity between the plurality of sets of characteristic data of the cutter and the plurality of sets of characteristic data of the preset cutter to generate a plurality of similarity values of the plurality of sets of characteristic data of the cutter;
and determining the loss level of the cutter according to a plurality of similarity values of a plurality of groups of characteristic data of the cutter.
2. The tool loss recognition method according to claim 1, wherein the acquiring of the image data of the pre-classified tool specifically includes:
acquiring an initial image of the cutter to generate cutter initial image data;
pre-classifying the cutter initial image data by adopting a cutter classification model to generate pre-classified cutter image data;
at least bit image data and milling cutter image data are prestored in the cutter classification model;
the pre-classifying the cutter initial image data by adopting a cutter classification model specifically comprises the following steps:
comparing the cutter initial image data with a plurality of cutter image data stored in advance to generate comparison data values of a plurality of cutters;
and determining the image data of the pre-classified cutter according to the comparison data value.
3. The tool loss recognition method according to claim 1, wherein the sets of characteristic data of the tool specifically include:
profile group features of the tool;
coating set characteristics of the tool;
cutting group characteristics of the cutter;
the profile group features of the cutter at least comprise shape features of the cutter and profile features of the cutter.
4. The tool loss recognition method according to claim 1, wherein sets of feature data of a tool in the image data of the tool are extracted at least by a convolutional neural network.
5. The tool loss identification method according to claim 1, wherein determining the loss level of the tool based on the similarity values of the sets of features of the tool comprises:
respectively carrying out standardization processing on the similarity values of a plurality of groups of characteristics of the cutter to generate the similarity values of a plurality of groups of characteristics of the cutter after the standardization processing;
respectively carrying out weight assignment on a plurality of groups of characteristics of the cutter;
and determining the loss level of the cutter according to the similarity value between the weight assignment of the plurality of groups of characteristics of the cutter and the plurality of groups of characteristics of the cutter after the standardized processing.
6. The method for identifying the loss of the cutter according to claim 3, wherein the method for identifying the loss of the cutter according to claim 3 is characterized in that the method for identifying the loss of the cutter according to claim 3 comprises the following steps:
and respectively carrying out normalization processing on the similarity values of the plurality of groups of features to generate the similarity values of the plurality of groups of features after normalization processing.
7. The tool loss identification method according to claim 1, further comprising, after determining a loss level of the tool from similarity values of several sets of features of the tool:
and (3) carrying out one of maintenance treatment, scrapping treatment and no treatment on the cutter according to the loss grade of the cutter.
8. A tool loss recognition device based on a target detection model, comprising:
the acquisition module is used for acquiring image data of the pre-classified cutter;
the extraction module is used for extracting a plurality of groups of characteristic data of the cutter in the image data of the cutter;
the comparison module is used for respectively comparing the similarity between the plurality of sets of characteristic data of the cutter and the plurality of sets of characteristic data of the preset cutter to generate a plurality of similarity values of the plurality of sets of characteristic data of the cutter;
and the determining module is used for determining the loss level of the cutter according to a plurality of similarity values of a plurality of groups of characteristic data of the cutter.
CN202211521759.5A 2022-11-30 2022-11-30 Tool loss identification method and device based on target detection model Pending CN116071298A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211521759.5A CN116071298A (en) 2022-11-30 2022-11-30 Tool loss identification method and device based on target detection model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211521759.5A CN116071298A (en) 2022-11-30 2022-11-30 Tool loss identification method and device based on target detection model

Publications (1)

Publication Number Publication Date
CN116071298A true CN116071298A (en) 2023-05-05

Family

ID=86179376

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211521759.5A Pending CN116071298A (en) 2022-11-30 2022-11-30 Tool loss identification method and device based on target detection model

Country Status (1)

Country Link
CN (1) CN116071298A (en)

Similar Documents

Publication Publication Date Title
DE102014214448B4 (en) Device and method for recognizing a traffic sign
Yoo et al. Development of a crack recognition algorithm from non-routed pavement images using artificial neural network and binary logistic regression
CN109614488B (en) Text classification and image recognition-based distribution network live working condition judgment method
CN110555819A (en) Equipment monitoring method, device and equipment based on infrared and visible light image fusion
CN103106265A (en) Method and system of classifying similar images
CN106950945A (en) A kind of fault detection method based on dimension changeable type independent component analysis model
CN105023027A (en) Sole trace pattern image retrieval method based on multi-feedback mechanism
US9189841B2 (en) Method of checking the appearance of the surface of a tyre
CN111325410A (en) General fault early warning system based on sample distribution and early warning method thereof
CN110794360A (en) Method and system for predicting fault of intelligent electric energy meter based on machine learning
CN114863464A (en) Second-order identification method for PID drawing picture information
CN108375729A (en) Degenerate state appraisal procedure is clustered based on the motor SOM that Fisher differentiates
CN117010666B (en) Intelligent management method and system for production workstation of automobile repair tool
CN115984158A (en) Defect analysis method and device, electronic equipment and computer readable storage medium
CN116071298A (en) Tool loss identification method and device based on target detection model
CN113179389A (en) System and method for identifying crane jib of power transmission line dangerous vehicle
CN109635008B (en) Equipment fault detection method based on machine learning
CN116503809A (en) Post-processing method for intelligent factory behavior wearing false alarm filtering
CN103793913A (en) Spectral clustering image segmenting method combined with mean shift
CN116665272A (en) Airport scene face recognition fusion decision method and device, electronic equipment and medium
CN114596439A (en) Image-based intelligent damage assessment and check method and device for car insurance and electronic equipment
CN113705229B (en) High-risk enterprise illegal production identification method based on electric power big data
Mia et al. An efficient image segmentation method based on linear discriminant analysis and K-means algorithm with automatically splitting and merging clusters
CN113592043B (en) Carton classification method
CN114155522A (en) Point cloud data quality inspection repairing method and system

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