WO2019203924A1 - Automatisation de classements de parties de machine visuelle - Google Patents

Automatisation de classements de parties de machine visuelle Download PDF

Info

Publication number
WO2019203924A1
WO2019203924A1 PCT/US2019/018867 US2019018867W WO2019203924A1 WO 2019203924 A1 WO2019203924 A1 WO 2019203924A1 US 2019018867 W US2019018867 W US 2019018867W WO 2019203924 A1 WO2019203924 A1 WO 2019203924A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
predictive model
machine part
rating
images
Prior art date
Application number
PCT/US2019/018867
Other languages
English (en)
Inventor
George A. KHOURY
Benjamin C. GAMOKE
Bashkar PRABHAKAR
Original Assignee
Exxonmobil Research And Engineering Company
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 Exxonmobil Research And Engineering Company filed Critical Exxonmobil Research And Engineering Company
Publication of WO2019203924A1 publication Critical patent/WO2019203924A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/28—Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • 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
    • G06T7/001—Industrial image inspection using an image reference approach
    • 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
    • G06V2201/06—Recognition of objects for industrial automation

Definitions

  • This invention relates to systems and methods for automatically rating machine parts based upon predictive models built utilizing images and prior ratings of the machine parts shown in the images.
  • Ratings based upon images of objects are used in many tests for determining the quality, effectiveness, and/or expected performance of the objects shown in the images, as well as any associated operating media (e.g ., hydraulic fluid, lubricant, or oil) or operating conditions used during operation of the object.
  • image-based rating of machine parts can be used to determine quality and/or expected performance of a fluid or conditions in which the machine part(s) that are the subject of various images are operated.
  • an oil filter z.e., machine part
  • image-based ratings of machine parts can include, without limitation, rating piping in a manufacturing process using images of normal and defective piping, rating normalcy or defects in a compressor or a pump based on an expert operator or technician’s trained eye, and rating a flare as having black smoke, incipient or white smoke automatically rather than literally observing the flare or watching it through a non-recording monitor.
  • methods for building automatic predictive models for machine part rating.
  • the methods can include receiving a plurality of images. Each image can show one of a plurality of machine parts. Each machine part can have a particular model, type, variation, etc. Images that show machine parts having the same particular model, type, variation, etc. can be homogenized or transformed such that differences among images of similar machine parts can be minimized. At least one prior rating (e.g ., a prior human-attributed rating) for the particular machine part shown in each homogenized image can be received.
  • One or more features can be extracted from each homogenized image. Utilizing the features extracted from each homogenized image and the prior rating(s) for the particular machine part shown in each homogenized image the automatic predictive model can be built.
  • the features extracted from each homogenized image and the prior rating(s) for the particular machine part shown in each homogenized image can be utilized to build a linear predictive model.
  • the linear predictive model can be generated by partial least squares regression (PLSR), multiple regressions, and/or linear models containing cross-terms, among others.
  • the features extracted from each homogenized image and the prior rating(s) for the particular machine part shown in each homogenized image can be utilized to build a multilinear predictive model.
  • the multilinear predictive model can include nonlinear components, such as cross-terms, ratios, and/or polynomial forms.
  • the features extracted from each homogenized image and the prior rating(s) for the particular machine part shown in each homogenized image can be utilized to build a nonlinear predictive model.
  • nonlinear predictive models can be directly built which include cross-terms, ratios, and/or polynomial forms.
  • the nonlinear predictive model can be a predictive model defined by support vector regression, decision tree, elastic network, neural network, deep neural network, convolutional neural network, and/or blended models consisting of one or more prior-stated models, among others.
  • methods are provided for automatically rating machine parts based upon machine-part images.
  • the method can include receiving an image showing a particular machine part.
  • the particular machine part can have a particular model, type, variation, or the like.
  • the image can be automatically compared to a plurality of homogenized images of machine parts having the particular model, type, variation or the like.
  • homogenized images can have one or more prior ratings (e.g., human-attributed rating) associated therewith. Each prior rating can be for the machine part shown in the associated image. A rating can be automatically assigned to the particular machine part based upon the automatic comparison.
  • prior ratings e.g., human-attributed rating
  • a rating can be automatically assigned to the particular machine part based upon the automatic comparison.
  • systems are provided for building an automatic predictive model for machine part rating.
  • the system can include a processor and one or more computer storage media.
  • the one or more computer storage media can store computer-useable instructions that, when used by the processor, cause the processor to perform certain functions.
  • the functions performed can include receiving a plurality of images. Each image can show one of a plurality of machine parts. Each machine part can have a particular model, type, variation, or the like.
  • the functions performed further can include homogenizing the images that show machine parts having the same particular model, type, variation, or the like.
  • the functions performed further can include receiving at least one prior rating (e.g ., human-attributed rating) for the particular machine part shown in each homogenized image.
  • the functions performed further can include extracting one or more features from each homogenized image.
  • the functions performed further can include building the automatic predictive model (for instance, a linear model (e.g., a partial least squares regression model) or a nonlinear model (e.g., a support vector regression model) utilizing the one or more features extracted from each homogenized image and the at least one prior rating for the particular machine part shown in each homogenized image.
  • a linear model e.g., a partial least squares regression model
  • a nonlinear model e.g., a support vector regression model
  • FIG. 1 shows an exemplary system for building an automatic predictive model for machine part rating.
  • FIG. 2 shows a flowchart illustrating an exemplary method for building an automatic predictive model for machine part rating.
  • FIG. 3 shows a flowchart illustrating an exemplary method for automatically rating machine parts based on machine part images.
  • FIG. 4 is a block diagram of an exemplary computing environment suitable for use in accordance with some aspects of the present invention.
  • systems and methods are provided for automatically rating machine parts to aid in determining the quality, effectiveness, and/or expected performance of the objects shown in the images, as well as the associated operating media or operating conditions used during operation of the object.
  • “automatically” refers to the rating of machine parts utilizing one or more reference data stores and one or more machine learning algorithms. In this way, machine parts can be rated without direct human intervention, saving spot checking, supervising, and the like.
  • a data store having a plurality of stored images of various machine parts can be compiled. Each stored image can show a machine part having a particular model, type, variation, etc. Images showing machine parts having the same particular model, type, variation, etc.
  • image capture devices utilized to capture images of machine parts can be transformed or homogenized such that differences among images of similar machine parts due to, for instance, image capture devices utilized to capture images of machine parts, settings of image capture devices utilized to capture images of machine parts, angles and/or zoom levels at which images of machine parts are captured, distances between image capture devices and the captured machine parts during image capture, and the like, can be minimized and the quantity of different images of machine parts having the same particular model, type, variation, etc. that can be utilized to compile the data store can be maximized.
  • At least one feature can be extracted from each image.
  • Features can be characteristics of an image that may be indicative of the quality and/or expected performance of the machine part that is the subject of the image.
  • Features can be characteristics of an image that, in prior image-based ratings systems, a human rating technician might have evaluated in rating the quality and/or expected performance of the machine part shown. While the number of features extracted from each image can be as low as one, the more features that are extracted from an image, the more accurate and robust the data store can be, resulting in more accurate and reliable automated ratings. Desired features for extraction may be based on a recommended standard set of features, a customized set of features, or any combination thereof.
  • exemplary features that can be extracted from images of, e.g., oil filters can include: the number of connected components in the image (that is, regions of the image that have similar color content and have a clear and distinguishable boundary), the average area of the connected components in the image, the standard deviation of the connected components in the image, the mean centroid location of the connected components in the image, the standard deviation of the centroid location of the connected components in the image, the percentage of the image designated as a“deposits” (also referred to as“external coverings,” sludge,” surface deposits,” and/or“solids”), the mean pixel intensity of the region of the image identified as“deposits,” the standard deviation pixel intensity of the region of the image identified as“deposits,” the mean grayscale pixel intensity of the region of the image identified as“deposits,” the standard deviation grayscale pixel intensity of the region of the image identified as“deposits,” the red pixel intensity of the region of the image identified as“de
  • the features extracted from each image, along with prior human-assigned ratings of the particular machine part that is the subject of each image, can be utilized to build one or more predictive models for machine part rating.
  • Predictive machine-part-rating models can be machine learned models that are capable of predicting a quality, effectiveness, and/or expected performance rating of a particular machine part that is shown in an input image, as well as any associated operating media or operating conditions used during operation of the object. Such predictive rating can be based upon comparison to a plurality of prior images showing machine parts having the same model, type, variation, etc. of the machine part shown in the input image and the quality and/or performance rating(s) associated with each prior image.
  • Predictive linear models for machine part rating can be built, for instance, using partial least squares regression.
  • Predictive nonlinear models can be built, for instance, using support vector regression.
  • human-assigned ratings can be supplied to the systems and methods described herein such that the predictive models can be updated in a fashion that improves model performance and/or accuracy as the model gains additional data over time.
  • FIG. 1 a block diagram is provided illustrating an exemplary computing system 100 in which aspects of the present invention may be employed.
  • the computing system 100 illustrates an environment in which linear, multilinear, and/or nonlinear predictive models for machine part rating may be built and implemented, in accordance with the methods, for instance, illustrated in FIGS. 2 and 3 (more fully described below).
  • the computing system 100 generally includes a computing device 110, a model building and implementing engine 112, and a data store 114, all in communication with one another via a network 116.
  • the network 116 may include, without limitation, one or more local area networks (LANs) and/or wide area networks (WANs). Such networking environments are commonplace in offices, enterprise-wide computer networks, intranets and the Internet. Accordingly, the network 116 is not further described herein.
  • LANs local area networks
  • WANs wide area networks
  • computing devices 110 and/or model building and implementing engines 112 may be employed in the computing system 100 within the scope of aspects of the present invention. Each may comprise a single device/interface or multiple devices/interfaces cooperating in a distributed environment.
  • the model building and implementing engine 112 may comprise multiple devices and/or modules arranged in a distributed environment that collectively provide the functionality of the model building and implementing engine 112 described herein. Additionally, other components or modules not shown also may be included within the computing system 100.
  • one or more of the illustrated components/modules may be implemented as stand-alone applications. In other embodiments, one or more of the illustrated components/modules may be implemented via the model building and implementing engine 112 or as an Internet-based service. It will be understood by those having ordinary skill in the art that the components/modules illustrated in FIG. 1 are exemplary in nature and in number and should not be construed as limiting. Any number of components/modules may be employed to achieve the desired functionality within the scope of aspects hereof. Further, components/modules may be located on any number of model building and implementing engines 112. By way of example only, the model building and implementing engine 112 might be provided as a single computing device, a cluster of computing devices, or a computing device remote from one or more of the remaining components.
  • the computing device 110 may include any type of computing device, such as the computing device 400 described with reference to FIG. 4, for example.
  • computing device 400 described with reference to FIG. 4, for example.
  • the computing device 110 is a mobile computing device (e.g ., a mobile telephone, a tablet computer, a laptop PC, a smart band or watch, or the like) that a user feasibly routinely may have in proximity to his or her presence.
  • the computing device 110 is associated with a microphone, a speaker and one or more I/O components, such as a stylus or keypad, for permitting alpha-numeric and/or textual input into a designated region (e.g., text box).
  • a designated region e.g., text box
  • the functionality described herein as being performed by the computing device 110 may be performed by any other application, application software, user interface, or the like capable of, at least in part, transmitting images of machine parts and associated human-assigned ratings.
  • aspects of the present invention are equally applicable to devices accepting gesture, touch and/or voice input. Any and all such variations, and any combination thereof, are contemplated to be within the scope of aspects of the present invention.
  • FIG. 1 is configured to, among other things, receive machine part images, build predictive models, and apply predictive models to rate machine parts based upon images.
  • the model building and implementing engine 112 has access to the data store 114.
  • the data store 114 is configured to be searchable for one or more of the items stored in association therewith.
  • the information stored in association with the data store 114 may be configurable and may include any information relevant to, by way of example only, machine part images, human- assigned quality, effectiveness and/or expected performance ratings, predictive model quality and/or expected performance ratings, machine part models, and the like. The content and volume of such information are not intended to limit the scope of aspects of the present invention in any way.
  • the data store 114 may be a single, independent component (as shown) or a plurality of storage devices, for instance a database cluster, portions of which may reside in association with the model building and implementing engine 112, another external computing device (not shown), and/or any combination thereof. Additionally, the data store 114 may include a plurality of unrelated data stores within the scope of aspects of the present invention.
  • the model building and implementing engine 112 of FIG. 1 includes an image receiving component 118, an image homogenizing component 120, a ratings receiving component 122, a feature extracting component 124, an automatic predictive model building component 126, and an automatic predictive model applying component 128.
  • the image receiving component 118 is configured to receive a plurality of images. Images can be received, for instance from one or more data stores (e.g ., data store 114), one or more computing devices (e.g ., computing device 110), one or more image capture devices (not shown), or any combination of images.
  • data stores e.g ., data store 114
  • computing devices e.g ., computing device 110
  • image capture devices not shown
  • Each image can show one of a plurality of machine parts.
  • Each machine part can have a particular model, type, variation, etc.
  • model is utilized herein to refer to machine parts having a particular model, type, variation, etc. That is, the term“model” is utilized herein to refer to machine parts that are similar enough to one another that comparison among images and ratings of such parts is useful for purposes of comparison. For instance, an oil filter having a first manufacturer and model number can be different enough from an oil filter having a second manufacturer and/or model number such that comparisons among images thereof would prove of little use or value in rating quality and/or expected performance.
  • Machine parts having the same model can be machine parts that are similar enough that comparisons between them prove useful, whether human or machine-assisted ratings are being assigned.
  • images received by the image receiving component 118 of the model building and implementing engine 112 can be captured from relatively consistent angles (for instance, angles differing by less than 20°).
  • the image homogenizing component 120 is configured to homogenize or transform received images such that differences among images of similar machine parts due to, e.g., image capture devices utilized to capture images of machine parts, settings of image capture devices utilized to capture images of machine parts, angles and/or zoom levels at which images of machine parts are captured, distances between image capture devices and the captured machine parts during image capture, and the like, can be minimized and the quantity of different images of machine parts having the same model that can be utilized to compile the data store can be maximized.
  • Homogenization can include, by way of example only, zooming in on a central region of a machine part so that different angles, image capture devices, and distance between the image capture devices and the machine part can be utilized.
  • the ratings receiving component 122 is configured to receive at least one prior rating for the particular machine part shown in each homogenized image. In aspects, for instance, upon first building of an automatic predictive model for machine part rating and/or upon updating an automatic predictive model to improve accuracy and the like, received ratings can be prior human-assigned ratings for the particular machine part shown in each homogenized image.
  • the feature extracting component 124 is configured to extract one or more features from each homogenized image. Features can be characteristics of an image that may be indicative of the quality, effectiveness, and/or expected performance of the machine part that is shown in the image, as well as any associated operating media or operating conditions used during operation of the machine part.
  • Features can be characteristics of an image that, in prior image- based ratings systems, a human rating technician might have evaluated in rating the quality, effectiveness, and/or expected performance of the machine part shown, or any associated operating media or operating conditions used during machine part operation. While the number of features extracted from each image can be as low as one, the more features that are extracted from an image, the more accurate and robust the data store being built can be, resulting in more accurate and reliable automated ratings when the model is utilized to automatically predict machine part ratings. Desired features for extraction may be based on a recommended standard set of features, a customized set of features, or any combination thereof.
  • exemplary features that can be extracted from images of, e.g., oil filters can include: the number of connected components in the image, the average area of the connected components in the image, the standard deviation of the connected components in the image, the mean centroid location of the connected components in the image, the standard deviation of the centroid location of the connected components in the image, the percentage of the image designated as“deposits” (e.g.,“oil deposits”), the mean pixel intensity of the region of the image identified as“deposits,” the standard deviation pixel intensity of the region of the image identified as“deposits,” the mean grayscale pixel intensity of the region of the image identified as“deposits,” the standard deviation grayscale pixel intensity of the region of the image identified as“deposits,” the red pixel intensity of the region of the image identified as “deposits,” the green pixel intensity of the region on the image identified as“deposits,” the blue pixel intensity of the region of the image identified as“deposits
  • the region of an image defined as“deposits” can be defined as pixels that are darker than the other pixels in a given image.
  • L*a*b decomposition is a method of converting the representation of each pixel in a color image (i.e., rgb) to lightness (L), red- green (a), and blue-yellow (b). Such representation enables an easier analysis of color and darkness in a given image.
  • L*a*b is known to those having ordinary skill in the relevant art and, accordingly, is not further described herein.
  • the automatic predictive model building component 126 is configured to build automatic predictive models utilizing the features extracted from each homogenized image and the prior rating(s) for the particular machine part shown in each homogenized image. In aspects, individual features or ratings that are outside of three standard deviations of the mean for machine parts having the same model are disregarded as outliers.
  • Linear, multilinear, and/or nonlinear automatic predictive models may be built in accordance with aspects of the present invention.
  • partial least squares regression PLSR
  • PLSR is a predictive model building technique known to those having ordinary skill in the relevant art and, accordingly, is not further described herein.
  • support vector regression SVR
  • SVR is a predictive model building technique known to those having ordinary skill in the relevant art and, accordingly, is not further described herein.
  • the predictive model applying component 128 is configured to apply a predictive model to automatically rate a machine part based upon a machine part image.
  • the predictive model applying component 128 includes a comparing component 130 and a ratings calculating component 132.
  • the comparing component 130 is configured to automatically compare an input image, the image showing a particular machine part having a particular model, to a plurality of homogenized images of machine parts having the particular mode.
  • Each of the plurality of homogenized images can have one or more prior ratings associated therewith. At least one of the one or more prior ratings can be a human-assigned rating.
  • the ratings calculating component 132 can be configured to calculate a rating to the particular machine part shown in the input image based upon the automatic comparison.
  • FIG. 2 a flow diagram is shown illustrating an exemplary method 200 for building an automatic predictive model for machine part rating.
  • a plurality of images is received. Each image can show one of a plurality of machine parts. Each machine part can have a particular model.
  • the received images are homogenized such that differences among images of similar machine parts due to, e.g., image capture devices utilized to capture images of machine parts, settings of image capture devices utilized to capture images of machine parts, angles and/or zoom levels at which images of machine parts are captured, distances between image capture devices and the captured machine parts during image capture, and the like, can be minimized and the quantity of different images of machine parts having the same model that can be utilized to compile the data store can be maximized.
  • Homogenization can include, by way of example only, zooming in on a central region of a machine part so that different angles, image capture devices, and distance between the image capture devices and the machine part can be utilized.
  • At least one prior rating for the particular machine part shown in each homogenized image is received.
  • received ratings can be prior human-assigned ratings for the particular machine part shown in each homogenized image.
  • features can be characteristics of an image that may be indicative of the quality, effectiveness, and/or expected performance of the machine part that is shown in the image, as well as the associated operating media or operating conditions used during operation of the object.
  • features can be characteristics of an image that, in prior image- based ratings systems, a human rating technician might have evaluated in rating the quality, effectiveness, and/or expected performance of the machine part shown, and/or the associated operating media or operating conditions used during operation of the machine part.
  • Desired features for extraction may be based on a recommended standard set of features, a customized set of features, or any combination thereof.
  • an automatic predictive model is built utilizing the one or more features extracted from each homogenized image and the at least one prior rating for the particular machine part shown in each homogenized image.
  • Linear, multilinear, and/or nonlinear automatic predictive models may be built in accordance with aspects of the present invention.
  • partial least squares regression can be used, among other methods.
  • support vector regression can be used, among other methods.
  • FIG. 3 a flowchart illustrating an exemplary method 300 is shown for automatically rating machine parts based on machine part images.
  • an image showing a particular machine part is received.
  • the image can have a particular model.
  • the image is automatically compared to a plurality of homogenized images of machine parts having the particular model.
  • Each of the plurality of homogenized images can have one or more prior ratings associated therewith.
  • Each of the prior ratings can be for the machine part shown in the associated image.
  • the prior rating can be prior human-attributed ratings for the particular machine part shown in each homogenized image.
  • a rating is automatically calculated for the particular machine part based upon the automatic comparison.
  • FIG. 4 An exemplary operating environment in which aspects of the present invention may be implemented is described below in order to provide a general context for various aspects of the present invention. Referring to FIG. 4, an exemplary operating environment for
  • computing device 400 implementing aspects of the present invention is shown and designated generally as computing device 400.
  • the computing device 400 is but one example of a suitable computing environment and is not intended to suggest any limitation as to the scope of use or functionality of aspects of the invention. Neither should the computing device 400 be interpreted as having any dependency or requirement relating to any one or combination of components illustrated.
  • aspects of the invention may be described in the general context of computer code or machine-useable instructions, including computer-executable instructions such as program modules, being executed by a computer or other machine, such as a personal data assistant or other handheld device.
  • program modules including routines, programs, objects, components, data structures, etc., refer to code that perform particular tasks or implement particular abstract data types.
  • aspects of the invention may be practiced in a variety of system configurations, including hand-held devices, consumer electronics, general-purpose computers, more specialty computing devices, etc.
  • aspects of the invention also may be practiced in distributed computing environments where tasks are performed by remote-processing devices that are linked through a communications network.
  • the computing device 400 includes a bus 410 that directly or indirectly couples the following devices: a memory 412, one or more processors 414, one or more presentation components 416, one or more input/output (EO) ports 418, one or more input/output components 420, and an illustrative power supply 422.
  • the bus 410 represents what may be one or more busses (such as an address bus, data bus, or combination thereof).
  • busses such as an address bus, data bus, or combination thereof.
  • FIG. 4 is merely illustrative of an exemplary computing device that can be used in connection with one or more aspects of the present invention. Distinction is not made between such categories as“workstation,”“server,”“laptop,” “hand-held device,” etc., as all are contemplated within the scope of FIG. 4 and reference to “computing device.”
  • the computing device 400 typically includes a variety of computer-readable media.
  • Computer-readable media can be any available media that can be accessed by the computing device 400 and includes both volatile and nonvolatile media, removable and non-removable media.
  • Computer-readable media may comprise computer storage media and communication media.
  • Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data.
  • Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by the computing device 400.
  • Computer storage media does not comprise signals per se.
  • Communication media typically embodies computer-readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media.
  • modulated data signal means a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal.
  • communication media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared and other wireless media.
  • the memory 412 includes computer storage media in the form of volatile and/or nonvolatile memory.
  • the memory may be removable, non-removable, or a combination thereof.
  • Exemplary hardware devices include solid-state memory, hard drives, optical-disc drives, etc.
  • the computing device 400 includes one or more processors that read data from various entities such as the memory 412 or the I/O component(s) 420.
  • the presentation component(s) 416 present data indications to a user or other device. Exemplary presentation components include, without limitation, a display device, speaker, printing component, vibrating component, etc.
  • the I/O ports 418 allow the computing device 400 to be logically coupled to other devices including the EO component s) 420, some of which may be built in.
  • Illustrative components include, without limitation, a microphone, joystick, game pad, satellite dish, scanner, printer, wireless device, etc.
  • the I/O components 420 may provide a natural user interface (NUI) that processes air gestures, voice, or other physiological inputs generated by a user. In some instance, inputs may be transmitted to an appropriate network element for further processing.
  • NUI may implement any combination of speech recognition, touch and stylus recognition, facial recognition, biometric recognition, gesture recognition both on screen and adjacent to the screen, air gestures, head and eye-tracking, and touch recognition associated with displays on the computing device 400.
  • the computing device 400 may be equipped with depth cameras, such as, stereoscopic camera systems, infrared camera systems, RGB camera systems, and combinations of these for gesture detection and recognition. Additionally, the computing device 400 may be equipped with accelerometers or gyroscopes that enable detection of motion.
  • depth cameras such as, stereoscopic camera systems, infrared camera systems, RGB camera systems, and combinations of these for gesture detection and recognition.
  • the computing device 400 may be equipped with accelerometers or gyroscopes that enable detection of motion.
  • the computing device 400 additionally may include a radio 424.
  • the radio 424 transmits and receives radio communications.
  • the computing device 400 may be a wireless terminal adapted to receive communications and media over various wireless networks.
  • the computing device 400 may communicate via wireless protocols, such as code division multiple access (“CDMA”), global system for mobiles (“GSM”), or time division multiple access
  • CDMA code division multiple access
  • GSM global system for mobiles
  • the radio communications may be a short-range connection, a long-range connection, or a combination of both a short-range and a long-range wireless telecommunications connection.
  • a short-range connection may include a Wi-Fi ® connection to a device (e.g ., a mobile hotspot) that provides access to a wireless communications network, such as a WLAN connection using the 802.11 protocol.
  • a Bluetooth ® connection to another computing device is a second example of a short- range connection.
  • a long-range connection may include a connection using one or more of CDMA, GPRS, GSM, TDMA, and 802.16 protocols.
  • Embodiment 1 A method for building an automatic predictive model for machine part rating, comprising: receiving a plurality of images, each image showing one of a plurality of machine parts, each machine part of the plurality of machine parts having a particular model; homogenizing the images of the plurality of images that show machine parts having the same particular model; receiving at least one prior rating for the particular machine part shown in each homogenized image; extracting one or more features from each homogenized image; and building the automatic predictive model utilizing the one or more features extracted from each homogenized image and the at least one prior rating for the particular machine part shown in each homogenized image.
  • Embodiment 2 The method of embodiment 1, wherein extracting the one or more features from each homogenized image comprises extracting at least one of the one or more features utilizing L*a*b decomposition.
  • Embodiment 3 The method of any of the above embodiments, wherein building the automatic predictive model comprises building a linear automatic predictive model.
  • Embodiment 4 The method of embodiment 3, wherein building the linear automatic predictive model comprises building the linear automatic predictive model utilizing partial least squares regression.
  • Embodiment 5 The method of any of embodiments 1 through 3, wherein building the automatic predictive model comprises building a nonlinear automatic predictive model.
  • Embodiment 6 The method embodiment 5, wherein building the nonlinear automatic predictive model comprises building the nonlinear automatic predictive model utilizing support vector regression.
  • Embodiment 7 A computerized system for building an automatic predictive model for machine part rating, comprising: a processor; and one or more computer storage media storing computer-useable instructions that, when used by the processor, cause the processor to: receive a plurality of images, each image showing one of a plurality of machine parts, each machine part of the plurality of machine parts having a particular model; homogenize the images of the plurality of images that show machine parts having the same particular model; receive at least one prior rating for the particular machine part shown in each homogenized image; extract one or more features from each homogenized image; and build the automatic predictive model utilizing the one or more features extracted from each homogenized image and the at least one prior rating for the particular machine part shown in each homogenized image.
  • Embodiment 8 The system of embodiment 7, wherein causing the processor to extract one or more features from each homogenized image comprises causing the processor to extract at least one of the one or more features utilizing L*a*b decomposition.
  • Embodiment 9. The system of any of embodiments 7 and 8, wherein causing the processor to build the automatic predictive model comprises causing the processor to build a linear automatic predictive model.
  • Embodiment 10 The system of embodiment 9, wherein causing the processor to build the linear automatic predictive model comprises causing the processor to build the linear automatic predictive model utilizing partial least squares regression.
  • Embodiment 11 The system of any of embodiments 7 and 8, wherein causing the processor to build the automatic predictive model comprises causing the processor to build a nonlinear automatic predictive model.
  • Embodiment 12 The system of embodiment 11, wherein causing the processor to build the nonlinear automatic predictive model comprises causing the processor to build the nonlinear automatic predictive model utilizing support vector regression.
  • Embodiment 13 A method for automatically rating machine parts based on machine- part images, comprising: receiving an image showing a particular machine part, the particular machine part having a particular model; automatically comparing the image to a plurality of homogenized images of machine parts having the particular model, each of the plurality of homogenized images having one or more prior human-assigned ratings associated therewith, each of the one or more prior human-assigned ratings being for the machine part shown in the associated image; and automatically calculating a rating of the particular machine part based upon the automatic comparison.
  • Embodiment 14 The method of embodiment 13, wherein automatically calculating the rating to the particular machine part based upon the automatic comparison comprises automatically calculating the rating of the particular machine part utilizing a linear automatic predictive model.
  • Embodiment 15 The method of embodiment 13, wherein automatically calculating the rating to the particular machine part based upon the automatic comparison comprises automatically calculating the rating of the particular machine part utilizing a nonlinear automatic predictive model.
  • Embodiment 16 The method of embodiments 1 to 15, wherein the parts being rated are being rated for level of deposits.
  • Embodiment 17 The method of embodiments 1 to 15, wherein the parts being rated are being rated for level of oil deposits.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

L'invention concerne des procédés et des systèmes de construction de modèles prédictifs automatique pour un classement de partie de machine. Les procédés peuvent comprendre la réception d'une pluralité d'images. Chaque image peut présenter l'une d'une pluralité de parties de machine. Chaque partie de machine peut avoir un modèle particulier. Des images qui montrent des parties de machine ayant le même modèle particulier peuvent être homogénéisées de telle sorte que des différences entre des images de parties de machine similaires peuvent être réduites au minimum. Au moins une évaluation préalable pour la partie de machine particulière représentée dans chaque image homogénéisée peut être reçue. Une ou plusieurs caractéristiques peuvent être extraites de chaque image homogénéisée. En utilisant les caractéristiques extraites de chaque image homogénéisée et de l'évaluation préalable pour la partie de machine particulière représentée dans chaque image homogénéisée, un modèle prédictif automatique linéaire ou non linéaire peut être construit.
PCT/US2019/018867 2018-04-16 2019-02-21 Automatisation de classements de parties de machine visuelle WO2019203924A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US201862658055P 2018-04-16 2018-04-16
US62/658,055 2018-04-16

Publications (1)

Publication Number Publication Date
WO2019203924A1 true WO2019203924A1 (fr) 2019-10-24

Family

ID=65767297

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2019/018867 WO2019203924A1 (fr) 2018-04-16 2019-02-21 Automatisation de classements de parties de machine visuelle

Country Status (1)

Country Link
WO (1) WO2019203924A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1394727A1 (fr) * 2002-08-30 2004-03-03 MVTec Software GmbH Reconnaissance d'objets à base de segmentation hierarchique
US20050259868A1 (en) * 2004-05-19 2005-11-24 Applied Vision Company, Llc Vision system and method for process monitoring
US20160026900A1 (en) * 2013-04-26 2016-01-28 Olympus Corporation Image processing device, information storage device, and image processing method
WO2017055878A1 (fr) * 2015-10-02 2017-04-06 Tractable Ltd. Étiquetage semi-automatique d'ensembles de données

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1394727A1 (fr) * 2002-08-30 2004-03-03 MVTec Software GmbH Reconnaissance d'objets à base de segmentation hierarchique
US20050259868A1 (en) * 2004-05-19 2005-11-24 Applied Vision Company, Llc Vision system and method for process monitoring
US20160026900A1 (en) * 2013-04-26 2016-01-28 Olympus Corporation Image processing device, information storage device, and image processing method
WO2017055878A1 (fr) * 2015-10-02 2017-04-06 Tractable Ltd. Étiquetage semi-automatique d'ensembles de données

Similar Documents

Publication Publication Date Title
KR102229594B1 (ko) 디스플레이 스크린 품질 검출 방법, 장치, 전자기기 및 저장매체
CN111310841B (zh) 医学图像分类方法、装置、设备、计算机设备和存储介质
US11631230B2 (en) Method, device, system and computer-program product for setting lighting condition and storage medium
AU2017204494B2 (en) Diagnosis assisting device, image processing method in diagnosis assisting device, and non-transitory storage medium having stored therein program
CN109084955A (zh) 显示屏质量检测方法、装置、电子设备及存储介质
KR102308437B1 (ko) 대상체의 외부의 검사를 최적화하기 위한 장치 및 그 방법
US20200202508A1 (en) Image-based maintenance prediction and detection of operating errors
JP6765791B2 (ja) パターンマッチング用の基準画像セットの作成方法、パターンマッチング用の基準画像セットの作成装置、ワークの認識方法、プログラム、及び記録媒体
CN111758117B (zh) 检查系统、识别系统以及学习数据生成装置
CN112767366A (zh) 基于深度学习的图像识别方法、装置、设备及存储介质
CN110378227B (zh) 修正样本标注数据的方法、装置、设备及存储介质
US20220164952A1 (en) Capture and Storage of Magnified Images
CN113837079A (zh) 显微镜的自动对焦方法、装置、计算机设备和存储介质
CN109087281B (zh) 显示屏外围电路检测方法、装置、电子设备及存储介质
US8139862B2 (en) Character extracting apparatus, method, and program
US20200082297A1 (en) Inspection apparatus and machine learning method
CN112926483A (zh) 标准机柜状态指示灯识别监测方法、装置和系统
CN118193595A (zh) 工业物联网的数据采集分析方法、装置、设备及存储介质
KR102436786B1 (ko) 인공지능 기반 외관 표면 품질 검사 장치 및 방법
WO2019203924A1 (fr) Automatisation de classements de parties de machine visuelle
CN116596928B (zh) 基于图像特征的花生油杂质快速检测方法
CN114359764A (zh) 一种基于图像数据的违章建筑识别方法、系统及相关设备
CN109829389B (zh) 机器移位判断方法、装置以及计算机存储介质
JP2018005505A (ja) 画像認識パラメータ設定装置、画像認識パラメータ設定方法、および画像認識パラメータ設定プログラム
CN110705627A (zh) 目标检测方法、目标检测系统、装置及可读存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19711193

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19711193

Country of ref document: EP

Kind code of ref document: A1