Image-based analysis of HOSE assemblies

Abstract

Analysis of hose assemblies using images. An application for a computing device can be configured to generate information about hoses and end fittings of hose assemblies, using different machine learning models to analyze images of the hose assembly component. A selection of a component type or type of information sought causes the image data to be fed corresponding one of the machine learning models. The machine learning models are configured to, for example, identify equivalent hose assembly parts and check acceptability of hose routings based on the image data provided.

Description

IMAGE-BASED ANALYSIS OF HOSE ASSEMBLIESCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is being filed on January 21, 2026, as a PCT International application and claims the benefit of and priority to U.S. Provisional Application Nos.63,887,615 filed on September 25, 2025; 63 / 748,129, filed January 22, 2025; the disclosures of which are hereby incorporated by reference in their entirety.TECHNICAL FIELD

[0002] This disclosure relates to image analysis of hoses and other components of hose assemblies for purposes of component replacement and / or routing improvement.BACKGROUND

[0003] Hose assemblies are critical, physical components in a variety of applications, such as hydraulics systems, heating and cooling systems, automotive, aircraft, and other vehicular systems, fuel systems, and the like. Typically, the hose assembly must satisfy several operational specifications to be suitable for a particular installation in a particular application. Such operational specifications vary from installation to installation and from application to application and can include, for example, hose length, hose outer diameter, hose inner diameter, material composition, strength, bend resistance, interior pressure rating, exterior pressure rating, porosity, type of fluid to be carried by the hose, operational temperature range, hose end configuration, end fitting configuration, end fitting type, end fitting material, end fitting coupling mechanism, and the like.

[0004] When a hose assembly or portion of a hose assembly fails or otherwise needs replacement, the hose assembly or portion needs to be replaced with an operationally suitable replacement.SUMMARY

[0005] Aspects of the present disclosure relate to generating insights about a hose assembly and / or components that can be used to make hose assemblies or portions of hose assemblies based on different images of the components. In some examples, different machine learning models are used to generate insights corresponding to the different components.

[0006] According to certain specific aspects, the present disclosure relates to a method of mapping a hose assembly, including: using a server to: generate, on a displaydevice of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of a layline of the hose assembly; and process a digital image transmitted from the electronic device and received by the server in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the layline, wherein when the server determines that the digital image includes at least the minimum set of data of the layline, the server is configured to map a hose of the hose assembly to a different hose of equivalent mechanical specifications to the hose; and wherein when the server determines that the digital image does not include at least the minimum set of data of the layline, the server is configured to reject the digital image.

[0007] According to further specific aspects, the present disclosure relates to a method of mapping a hose assembly, including: using a server to: generate, on a display device of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a first prompt to transmit a first image of a first portion of the hose assembly and a second prompt to transmit a second image of a second portion of the hose assembly, the second portion being different from the first portion; and determine whether a digital image transmitted from the electronic device and received by the server includes an image of the first portion of the hose assembly or of the second portion of the hose assembly, wherein when the digital image is received by the server in response to the first prompt and the server determines that the digital image does not include the first image of the first portion, the server is configured to reject the digital image; wherein when the digital image is received by the server in response to the first prompt and the server determines that the digital image does include the first image of the first portion, the server is configured to accept the digital image; wherein when the digital image is received by the server in response to the second prompt and the server determines that the digital image does not include the second image of the second portion, the server is configured to reject the digital image; and wherein when the digital image is received by the server in response to the second prompt and the server determines that the digital image does include the second image of the second portion, the server is configured to accept the digital image.

[0008] According to further specific aspects, the present disclosure relates to a method of mapping a hose assembly, including: using a server to: generate, on a displaydevice of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of an end fitting of the hose assembly; and process a digital image transmitted from the electronic device and received by the server in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the end fitting, wherein when the server determines that the digital image includes at least the minimum set of data of the end fitting, the server is configured to map the end fitting of the hose assembly to a different end fitting of equivalent mechanical specifications to the end fitting; wherein when the server determines that the digital image does not include at least the minimum set of data of the end fitting, the server is configured to reject the digital image; and wherein the minimum set of data includes image data corresponding to a thread of the end fitting and image data corresponding to a flare of the end fitting.

[0009] According to further specific aspects, the present disclosure relates to system for mapping hose assemblies, including: one or more processors; and non-transitory computer readable storage storing instructions which, when executed by the one or more processors, cause the system to: generate, on a display device of an electronic device, a graphical interface, the graphical interface including a first prompt to transmit a first image of a first component of a hose assembly and a second prompt to transmit a second image of a second component of the hose assembly, the first component being different from the second component; in response to receipt of a selection of the first prompt and receipt of the first image: feed the first image to a first machine learning model and not to a second machine learning model; and determine, using the first machine learning model, if the first image includes a first minimum set of data corresponding to the first component; and in response to receipt of a selection of the second prompt and receipt of the second image: feed the second image to the second machine learning model and not to the first machine learning model; and determine, using the second machine learning model, if the second image includes a second minimum set of data corresponding to the second component.

[0010] According to further aspects, the present disclosure relates to a method of mapping a hose assembly or a hose assembly component, including: using a server to: generate, on a display device of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of a layline of the hose assembly or the hose assemblycomponent; and process a digital image transmitted from the electronic device and received by the server in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the layline, wherein when the server determines that the digital image includes at least the minimum set of data of the layline, the server is configured to map the hose assembly component or a hose of the hose assembly to a different hose of equivalent mechanical specifications to the hose assembly component or the hose; and wherein when the server determines that the digital image does not include at least the minimum set of data of the layline, the server is configured to reject the digital image.

[0011] According to further aspects, the present disclosure relates to a method of mapping a hose assembly or a hose assembly component, including: using a server to: generate, on a display device of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of the hose assembly component or an end fitting of the hose assembly; and process a digital image transmitted from the electronic device and received by the server in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the hose assembly component or the end fitting, wherein when the server determines that the digital image includes at least the minimum set of data of the hose assembly component or the end fitting, the server is configured to map the hose assembly component or the end fitting to a different end fitting of equivalent mechanical specifications to the hose assembly component or the end fitting; wherein when the server determines that the digital image does not include at least the minimum set of data of the hose assembly component or the end fitting, the server is configured to reject the digital image; and wherein the minimum set of data includes image data corresponding to a thread of the hose assembly component or the end fitting and image data corresponding to a flare of the hose assembly component or the end fitting.

[0012] According to further aspects, the present disclosure relates to a method of classifying an end fitting, including: extracting frames from a video; providing the frames to a machine learning model; calculating, by the machine learning model and for each of the frames, an attribute score; and classifying an attribute of the fitting based on each attribute score.

[0013] According to further aspects, the present disclosure relates to method of mapping hose assembly component, including: generating, on a display device of anelectronic device, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of the hose assembly component or an end fitting of the hose assembly; and processing a digital image captured by a camera of the electronic device in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the hose assembly component or the end fitting, wherein when the electronic device determines that the digital image includes at least the minimum set of data of the hose assembly component or the end fitting, the electronic device is configured to map the hose assembly component or the end fitting to a different end fitting of equivalent mechanical specifications to the hose assembly component or the end fitting; wherein when the electronic device determines that the digital image does not include at least the minimum set of data of the hose assembly component or the end fitting, the electronic device is configured to reject the digital image; and wherein the minimum set of data includes image data corresponding to a thread of the hose assembly component or the end fitting and image data corresponding to a flare of the hose assembly component or the end fitting.

[0014] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. Additional aspects, features, and / or advantages of examples will be set forth in part in the description which follows and, in part, will be apparent from the description, or may be learned by practice of the disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Non-limiting and non-exhaustive examples are described with reference to the following figures, wherein like numbers correspond to like parts, components, or features.

[0016] FIG. 1 schematically depicts an example system for mapping hose assemblies in accordance with the present disclosure.

[0017] FIG. 2 depicts details of a portion of the system of FIG. 1.

[0018] FIG. 3 depicts additional components of the system of FIG. 1.

[0019] FIG. 4 depicts an example graphical interface generated by the system of FIG. 1.

[0020] FIG. 5 depicts another example graphical interface generated by the system of FIG 1.

[0021] FIG. 6 depicts another example graphical interface generated by the system of FIG 1.

[0022] FIG. 7 depicts another example graphical interface generated by the system of FIG 1.

[0023] FIG. 8 depicts another example graphical interface generated by the system of FIG 1.

[0024] FIG. 9 depicts an image of a portion of an example hose assembly.

[0025] FIG. 10 depicts an image of another portion of another example hose assembly.

[0026] FIG. 11 depicts example methods that can be performed using the system of FIG. 1.

[0027] FIG. 12 depicts example components of the end fitting mapping model of FIG. 2.

[0028] FIG. 13 is an image of an end fitting overlayed with aspects detected by the system of FIG. 1.

[0029] FIG. 14 is a two-dimensional representation of the aspects detected in FIG.13.

[0030] FIG. 15 depicts an example method that can be performed using, at least in part, the end fitting mapping model of FIG. 2.

[0031] FIG. 16 depicts a further example embodiment of the client device of the system of FIG. 1.DETAILED DESCRIPTION

[0032] The present disclosure relates to image-based analysis of hose assemblies and components capable of making up hose assemblies or portions of hose assemblies. Such components can include, for example, a hose and an end fitting. One non-limiting example of an end fitting is a fitting that can be attached to an end of a hose to enable the hose to be sealingly connected to another piece of equipment. Another non-limiting example of an end fitting is an adapter that allows two other end fittings to sealingly connect to each other. As used herein, images include photographs and videos. For example, an image can be a photograph, or a frame of a video.

[0033] Atypical hose assembly includes a hose with an end fitting on either end of the hose. The physical attributes needed for the hose and end fittings depend on howthe hose assembly is being used. For example, hose assemblies can be used to carry different types of fluid under different conditions (pressure, humidity, temperature), with other different stresses exerted on the hose assembly (e.g., tight bend radii in tight, compact spaces), and so forth.

[0034] When a portion of a hose assembly fails, it is often necessary to replace it, and to do so as quickly as possible to minimize system downtime and avoid problems (e.g., leaks, machine failure, explosions, fire) that could arise from an old or failing hose assembly. In many cases, the exact make and model of a hose assembly or component of a hose assembly may no longer be sold, may be on back order, and the like. Quickly identifying operationally equivalent hose assemblies and components therefore becomes critical.

[0035] A given component of a hose assembly can have many different attributes. For instance, for a hose, such attributes can include the hose inner diameter, outer diameter, maximum and minimum operating temperatures, type of application (e.g., hydraulics, cooling), material of the hose, pressure and environmental ratings for the hose, and the configuration of the end fittings required to connect the hose to a given system.

[0036] In many situations, the values of all of these different attributes may not be available. For example, a portion of the hose’s layline may be rubbed out or obscured by other equipment, or the part number of an end fitting may no longer be visible. In these situations, it can be impossible to locate a replacement hose assembly or component that has operational equivalence.

[0037] The present disclosure alleviates one or more of these challenges at least by providing a system capable of mapping what would otherwise be incomplete attribute information regarding a hose assembly or component of a hose assembly to an operationally equivalent assembly or component. Advantageously, the system can perform equivalence mapping based on image data only, without requiring a user to write down or type values of specific hose assembly attributes.

[0038] Further technological advantages are borne out by the present disclosure. For example, the system of the present disclosure is configured to feed different images of different components of hose assemblies to different machine learning models that are separately and individually trained on processing image data (e.g., pixel data) relating to different, specific physical features of hose assemblies. Such physical features can include, for example, a layline or portion of a layline that appears between pixelatedboundary lines of a hose. Such physical features can also include, for example, an end fitting including pixelated representations of flanges and threads of the end fitting. Such physical features can also include, for example, routing paths of hoses from which routing path characteristics such as bend radius and torsion are derived from pixelated boundary lines of a hose.

[0039] FIG. 1 schematically depicts an example system 100 for mapping hose assemblies in accordance with the present disclosure.

[0040] The system 100 includes various computing devices that are operatively linked to one another. In some examples, the computing devices are operatively linked via a network 112.

[0041] The network 112 includes network connections. Each network connection can be established by one or more of a local-area network (LAN), a wide-area network (WAN), point-to-point, Bluetooth, radio frequency (RF), and the like.

[0042] The computing devices of the example system 100 include a server 110, a client device 114, 814 and one or more databases 116, all operatively connected via the network 112.

[0043] Typically, the client device 114 is physically remote from the server 110. In some examples, the database(s) 116 is / are physically remote from the server 110, whereas in other examples the database(s) 116 can be locally supported by the server 110. In some examples, the data of the database(s) 116 and / or the software of the server 110 can be supported by a Cloud computing system with access to such components being provided via the network 112 to the user’s local system components.

[0044] In some examples, the server 110 and / or functionality of the server 110 described herein can be hosted privately, e.g., by a proprietor of hose assembly components. In some examples, data stored on the database 116 and used by the server 110 to perform the functionality described herein can be proprietary data, e.g., data privately collected, curated, cleaned, and / or flattened by a proprietor of hose assembly components. In some examples, the functionality of the server 110 described herein is distributed across multiple computing devices networked together. In some examples, the functionality of the server 110 is hosted entirely or partially locally on the client device 114, 814.

[0045] In some examples, the client device 114, 814 is operated by a user, such as a customer of a hose assembly proprietor. For example, the client device 114, 814 can be associated with a hose assembly distributor, with a manufacturer or operator ofmachinery that uses hose assemblies, with a datacenter that uses hose assemblies, e.g., for thermal cooling, and the like.

[0046] FIG. 3 depicts additional components of the system 100 of FIG. 1. In particular, FIG. 3 depicts example computing components of each of the server 110 and the client device 114, 814.

[0047] Referring to FIG. 3, each device 110, 114, 814 can include computing components 152. The computing components 152 include at least one processor 154 and memory 150.

[0048] The memory 150 can include a non-transitory computer readable medium. Depending on the exact configuration, the memory 150 (storing, among other things, the machine learning models and other software modules described herein) can be volatile (such as RAM), non-volatile (such as ROM, flash memory, etc.), or some combination of the two.

[0049] The server 110 can include one or more graphics processing units (GPUs) configured to expedite model training and / or model predictions.

[0050] Each device 110, 114, 814 can also include storage devices (removable storage 156, and / or non-removable storage 158) including, but not limited to, solid-state devices, magnetic or optical disks, or tape. Each device 110, 114, 814 can also include input device(s) 116 such as touch screens, keyboard, mouse, pen, voice input, etc., and / or output device(s) 160 such as a display devices, speakers, printers, etc. Each device 110, 114, 814 can also include one or more communication connections 164 for communication via the network 112, such as local -area network (LAN) connections, wide-area network (WAN) connections, point-to-point connections, Bluetooth connections, RF connections, and the like.

[0051] FIG. 2 depicts further details of a portion of the system 100 of FIG. 1.

[0052] Referring to FIG. 2, the client device 114 includes an input / output (I / O) device 128, which can correspond to the input devices 162 and output devices 160, respectively, of FIG. 3. In some examples, the I / O device 128 includes a display screen configured to display the graphical interfaces of, e.g., FIGS. 4-8. In some examples, the I / O device 128 can include a touch screen for making selections and entering other data and commands in response to prompts displayed via the display device.

[0053] The client device 114 includes a camera 126. In some examples, the camera 126 is linked to a dedicated camera software application also stored locally on the client device 114. The camera 126 can include a lens with a shutter that is controllablevia the I / O device 128 to capture digital images and videos. In some examples, the digital images and videos captured with the camera 126 are stored locally, or remotely, in a digital image and video album. Digital copies of the stored digital images and videos can be accessed and selected from the album and fed to the client application 124 to initiate various types of backend functionality at the server 110, as described herein. For example, the camera 126 can be used to capture digital images and videos of a laylines, hose assembly routings, and end fittings, which can then be fed, via the client application 124, to the server 110 via the network 112.

[0054] As mentioned, the client device 114 includes the client application 124. The client application 124 can be a dedicated software application stored locally on the client device 114 and configured to, e.g., generate the graphical interfaces of FIGS. 4-8 and to provide two-way communication between the I / O device 128 and the server 110 in connection with the various functionalities of the system 100 as described herein. That is, the client application 124 can be configured to process input data packets transmitted from the server 110 to generate components of the graphical interfaces of FIGS. 4-8 encoded in those input data packets. In addition, the client application 124 can be configured to process selections and other data inputs received via the I / O device 128 (e.g., touch inputs to the graphical interfaces of FIGS. 4-8) and generate outgoing data packets encoding those selections and other data inputs that are transmitted from the client device 114 to the server device 110.

[0055] The server 110 includes an application programming interface (API) 130. The API 130 defines computational protocols that enable the client application 124 to communicate with the back-end functionality of the server 110 that will be described below. When a user opens the client application 124, the client application generates an API call that is transmitted to the API 130. The API 130 then establishes protocols for a communication session between the server 110 and the client application 124 to enable the functionality of the system 100 as performed by the operational unit of the client device 114 and the server 110.

[0056] The server stores and runs modules that include a STAMPED module 132, a part number module 134, and an eCommerce module 136.

[0057] STAMPED is an industry acronym that stands for physical and operational attributes of a hose or hose assembly. The attributes can include Size, Temperature, Application, Materials / Media, Pressure, Ends, and Delivery. The size can include the axial length of the hose (where the axis is defined parallel to the primary direction offluid flow), as well as an inner diameter range and outer diameter range. The temperature can define the range of temperatures for one or both of the external environment and the material conveyed internally by the hose that the hose can tolerate. The application can define one or more of configuration, routing, orientation, anticipated movements of the hose, as well as other factors, such as whether the hose is used for intermittent or continuous service, the magnitude of mechanical loads it is subjected to, the magnitude of abrasion it is subjected to, electrical conductively requirements, equipment type, and other external conditions (e.g., the presence of corrosive materials such as acid, ozone, salt water, and the like). The material / media can define the type (e.g., chemical composition) of material that flows through the hose, its viscosity, the velocity or flow rate of the material, and the like. The pressure can define the maximum and minimum system operating pressures of which the hose is a part. The ends define attributes of the end fittings of a hose, such as the style, type, orientation, and attachment method of an end fitting, conductivity requirements of an end fitting, and the like. The delivery includes additional customer-specific requirements, such as testing requirements, certification requirements, packaging and shipping requirements and the like.

[0058] The STAMPED module 132 is configured to process STAMPED data provided via the client application 124 on the client device 114 and map the processed STAMPED data to an equivalent hose assembly or hose assembly component. That is, the STAMPED module 132 is configured to identify a component, e.g., a hose, that meets the STAMPED criteria entered via the client application 124. The STAMPED module 132 is configured to cause the server 110 to transmit data about the identified hose assembly or component to the client application 124 such that the client application 124 can display the data via the I / O device 128. In some examples, the data can include STAMPED data, an image, and / or other specifications of the identified hose assembly component.

[0059] The database 116 can include STAMPED data for many different hoses and hose assemblies. The STAMPED module 132 can access this STAMPED data to perform the mapping operations and identify equivalently STAMPED hoses and hose assemblies for given input STAMPED data. For example, STAMPED data for various hoses and hose assemblies can be stored in one or more data tables that are stored on the database 116. The STAMPED module 132 performs a look-up function where itcompares and atempts to match input STAMPED data with the various rows of STAMPED data in the one or more data tables, to thereby identify equivalents.

[0060] The part number module 134 is configured to process part number data provided via the client application 124 on the client device 114 and map the processed part number data to an equivalent hose assembly or hose assembly component. That is, the part number module 134 is configured to identify a component, e.g., a hose, that has equivalent operational criteria (e.g., equivalent STAMPED characteristics) as the component corresponding to the received part number. The part number module 134 is configured to cause the server 110 to transmit data about the identified hose assembly or component to the client application 124 such that the client application 124 can display the data via the I / O device 128. In some examples, the data can include STAMPED data, an image, a video and / or other specifications of the identified hose assembly component.

[0061] The database 116 can include part number data for many different hoses and hose assemblies. The part number module 134 can access this part number data to perform the mapping operations and identify operationally equivalent part numbers. For example, part numbers, together with various operational data about those parts, can be stored in one or more data tables that are stored on the database 116. The part number module 134 performs a look-up function where it compares and atempts to match the operational data corresponding to an input part number with the various rows of operational data for other hose assemblies and components in the one or more data tables, to thereby identify equivalents.

[0062] The eCommerce module 136 is configured to enable purchasing or ordering of a mapped hose assembly or component identified by the module 132, the module 136, or one of the machine learning models 146. For example, the eCommerce module 136 can include a web crawler configured to search for and identify websites hosted on one or more other servers that sell the mapped equivalent hose assemblies and components. The eCommerce module 136 can also be configured to generate links to those websites. The eCommerce module 136 can be configured to transmit the generated links to the client application 124, which displays the selectable links via the I / O device 128. In other examples, the eCommerce module 136 provides a dedicated eCommerce platform locally hosted by the server 110 and linked to a physical inventory of hose assemblies and components. Such a platform can be accessed via theclient application 124 on the I / O device 128 to purchase the identified, mapped, equivalent hose assembly or component.

[0063] The server 110 also includes multiple machine learning models 146. In this example, the machine learning models include a layline mapping model 140, an end fitting mapping model 142, and a hose routing model 144.

[0064] The machine learning models 146 can be configured to process pixel data of digital images and / or videos to derive insights from which the machine learning models 146 can map, by predictive analysis using one more machine learning algorithms, the input image and / or video to an equivalent hose assembly or hose assembly component. In some examples, the images and / or videos can include images and / or videos showing text. The machine learning models 146 can be configured to perform natural language processing on the images to extract the text from the images and input the text into one or more algorithms to map a hose assembly or component corresponding to the text to an equivalent hose assembly or component. For example, one or more of the machine learning models 146 can be configured to perform name entity recognition (NER) on digital images that are fed to the machine learning models 146.

[0065] The machine learning models 146 can employ different algorithms one from another. The machine learning models 146 can be configured to employ one or more types of artificial intelligence. The machine learning models 146 can include one or more pixel clustering algorithms configured to process pixel data (e.g., color data and brightness data of individual pixels and / or predefined shapes of clusters of pixels) in images and / or videos to identify certain physical features in clusters of pixels by comparing the pixel data of those pixel clusters to pixel data of other pixel clusters, such as other pixel clusters that are adjacent to the clusters that define the feature. Such physical features that one or more of the machine learning models 146 can be trained to recognize in digital images can include, for example, an edge of a hose, a textual or non-textual logo of a layline, an end fitting, a flare of an end fitting, a color or material (e.g., texture) of a hose or end fitting, a thread of an end fitting, a bend radius of a hose, a magnitude of torsion of a hose, and the like.

[0066] The machine learning models 146 are trained on training data. Each of the models 146 is trained on its own, model-specific training data. In addition, each of the models 146 is configured to learn and improve its algorithm(s) via a feedback loop based on data outputs of sessions with the client application 124.

[0067] The layline mapping model 140 is trained to map an image of a layline to an operationally equivalent hose or hose assembly. The layline mapping model 140 is initially trained using layline training data 122 stored on the database 116. The layline training data 122 can include images of laylines as well as data identifying, for each layline image, zero, one or more operationally equivalent hoses. The layline training data 122 can be updated via a feedback loop based on data outputs of sessions with the client application 124, providing further real time training of the layline mapping model 140. For example, images submitted via the client application 124 that are routed via the API 130 to the layline mapping model 140 can then be transmitted to the database 116 and stored in the hose routing training data 118, together with the mapped equivalent(s) of the imaged layline as identified by the layline mapping model 140. The end layline mapping model 142 can employ one or more algorithms that weight parameters (e.g., parameters corresponding to features) identified in the pixels of the received images to output a predicted operational equivalent or equivalents.

[0068] The end fitting mapping model 142 is trained to map an image of an end fitting to an operationally equivalent end fitting. The end fitting mapping model 142 is initially trained using end fitting training data 120 stored on the database 116. The end fitting training data 120 can include images and videos of end fittings and hose assemblies including end fittings as well as data identifying, for each such image, zero, one or more operationally equivalent end fittings. The end fitting training data 120 can be updated via a feedback loop based on data outputs of sessions with the client application 124, providing further real time training of the end fitting mapping model 142. For example, images and / or videos submitted via the client application 124 that are routed via the API 130 to the end fitting mapping model 142 can then be transmitted to the database 116 and stored in the end fitting training data 120, together with the mapped equivalent(s) of the imaged end fittings as identified by the end fitting mapping model 142. The end fitting mapping model 142 can employ one or more algorithms that weight parameters (e.g., parameters corresponding to features, such as end fitting flares, end fitting material, end fitting threads, end fitting thread angle, end fitting thread depth, physical features of a hose connected to the end fitting in the digital image) identified in the pixels of the received images and / or videos to output a predicted operational equivalent or equivalents.

[0069] The hose routing model 144 is trained to identify (via one or more predictive algorithms) routing issues in images of hose assembly routings and to suggest (via oneor more other predictive algorithms) routing modifications to improve performance of those assemblies. The hose routing model 144 is initially trained using hose routing training data 118 stored on the database 116. The hose routing training data 118 can include images and / or videos of routings of hose assemblies in situ as well as data identifying, for each such image, the type and other characteristics of the hose, and any issues with the imaged routing that could impact performance according to some predefined metric, such as time before part failure, probability of leakage, or maximum sustainable flow rate or pressure of the medium within the hose. The hose routing training data 118 can be updated via a feedback loop based on data outputs of sessions with the client application 124, providing further real time training of the hose routing model 144. For example, images submitted via the client application 124 that are routed via the API 130 to the hose routing model 144 can then be transmitted to the database 116 and stored in the hose routing training data 118, together with geometries and / or images of modified routings with the same or operationally equivalent hose assembly components that would improve or optimize the performance. The hose routing model 144 can employ one or more algorithms that weight parameters (e.g., parameters corresponding to features, such as hose type, hose length, hose bend radius in the routing, hose torsion in the routing in the digital image) identified in the pixels of the received images and / or videos to output an assessment of the quality of the routing and, in some examples, images or data suggesting ways to modify the routing for improved performance.

[0070] The API 130 is configured to route images submitted via the I / O device 128 to the appropriate one of the models 146 based on the functionality of the system 100 being selected at the I / O device 128. For example, based on the button selected via the I / O device 128, a subsequently uploaded image or video will be routed to the corresponding one of the layline mapping model 140, the end fitting mapping model 142, or the hose routing model 144. Each such selected button encodes a data packet header that the API 130 interprets to determine which of the machine learning models 146 to submit the subsequently uploaded image as model input.

[0071] FIG. 4 depicts an example graphical interface 170 generated by the system 100 of FIG. 1.

[0072] The interface 170 is a dashboard that includes a variety of selectable buttons for performing differential functions of the system 100. Data entered through the I / O device 128 following selection of the STAMPED button 172 is routed via the API 130to the STAMPED module 132. Data entered through the I / O device 128 following selection of the Scan Layline button 174 is routed via the API 130 to the layline mapping model 140. Data entered through the I / O device 128 following selection of the Hose Part Number button 176 is routed via the API 130 to the part number module 136. Data entered through the I / O device 128 following selection of the Scan End Fitting button 178 is routed via the API 130 to the end fitting mapping model 142. Data entered through the I / O device 128 following selection of the Scan Routing button 180 is routed via the API 130 to the hose routing model 144.

[0073] In some examples, selection of each button generates a module or model API call specific to that button and that facilitates the routing of the request to the appropriate module or model stored on the server 110. Thus, the interface 170 can provide technological advantages in its ability to route different types of requests relating to hose assembly analysis from the same graphical interface to an appropriate machine learning model trained specifically to perform image data analysis on digital images corresponding to the requested task.

[0074] Following selection of the STAMPED button 172, the client application 124 prompts entry of STAMPED information for a desired hose or hose assembly. The data entered in response to the STAMPED prompts is routed to the STAMPED module 132 for further processing as described herein and identification, if available, of one or more equivalent hoses or hose assemblies.

[0075] Following selection of the Hose Part Number button 176, the client application 124 prompts entry of a hose part number for a desired hose or hose assembly. The data entered in response to the prompt is routed to the part number module 134 for further processing as described herein and identification, if available, of one or more equivalent hoses or hose assemblies.

[0076] Following selection of the Scan Layline button 174, the client application 124 can, in some examples, generate the graphical interface 182 of FIG. 5.

[0077] Referring to FIG. 5, the graphical interface 182 includes a prompt button 184. Selection of the prompt button 184 can, e.g., cause another application to activate the camera 126 on the client device 114, 814 or pull up an album of stored digital images. A digital image of a layline can be taken with the camera or pulled from the album and uploaded to the server 110 for processing by the layline mapping model 140.

[0078] The layline mapping model 140 then causes the data field 186 to populate with the part number extracted or otherwise determined by the layline mapping model140 from analyzing the digital image of the layline. If the layline mapping model identifies one or more equivalent hoses or hose assemblies, the layline mapping model 140 also causes the data field 188 to populate with the part number of at least one operationally equivalent hose or hose assembly, e.g., a hose that has STAMPED data equivalent to the hose that has been scanned.

[0079] Once the data fields 186 and 188 are populated, selection of the confirm button 190 generates the graphical interface 200 of FIG. 7, which is described below.

[0080] Referring to FIG. 9, a representation of a digital image 214 of a layline of a hose and that can be submitted to the server 110 via the client application 124 is shown. In this example, the layline includes a brand name and a brand logo, a part number, hose size data, sub-brand data, certifications data, pressure rating data, and temperature rating data.

[0081] By performing pixel cluster analysis on the digital image 214, the layline mapping model 140 can recognize physical features of the hose, such as radial edges of the hose, edges of the printed strip that includes the layline, the location and orientation of the layline, the logo of the layline, the part number of the layline, and the like. Based on recognition of one of more of these physical features, the layline mapping model can then perform its mapping function to determine if there are one or more operationally equivalent hoses.

[0082] Following selection of the Scan End Fitting button 178 (FIG. 4), the client application 124 can, in some examples, generate the graphical interface 192 of FIG. 6. Referring to FIG. 6, the graphical interface 192 includes a prompt button 194. Selection of the prompt button 194 can, e.g., cause another application to activate the camera 126 on the client device 114, 814 or pull up an album of stored digital images. A digital image or video of an end fitting can be taken with the camera or pulled from the album and uploaded to the server 110 for processing by the end fitting mapping model 142.

[0083] The end fitting mapping model 142 then causes the data field 196 to populate with the part number extracted or otherwise determined by the end fitting mapping model 142 from analyzing the digital image or video of the end fitting. If the end fitting mapping model identifies one or more equivalent end fittings, the end fitting mapping model 142 also causes the data field 198 to populate with the part number of at least one operationally equivalent end fitting, e.g., an end fitting that has a particular flare configuration, thread configuration, material, and operational compatibility with a given hose.

[0084] Once the data fields 186 and 188 are populated, selection of the confirm button 199 generates the graphical interface 200 of FIG. 7, which is described below.

[0085] Referring to FIG. 10, a representation of a digital image 250 of an end fitting of a hose and that can be submitted to the server 110 via the client application 124 is shown.

[0086] By performing pixel cluster analysis on the digital image 250, the end fitting mapping model 142 can recognize physical features of the end fitting, such as flare type and diameter, material, coating, thread depth, thread density, thread pitch, whether the thread is external or internal to the fitting, crimp type, type of hose to which the end fitting is crimped, and the like. Based on recognition of one of more of these physical features, the end fitting mapping model 142 can then perform its mapping function to determine if there are one or more operationally equivalent end fittings.

[0087] Additional features and functions of the end fitting mapping model 142, and interface features relating to the scan end fitting button 178 will now be described.

[0088] Generally, an end fitting of a hose assembly in, e.g., an industrial, electrical, aerospace, or hydraulics application, is a connector used to connect a hose to another piece of equipment, such as a machine, a valve, another hose, and the like. Typically, the end fittings of hoses and other hydraulic fittings must be identified by a user or manufacturer as a part of the standard hose assembly using the STAMPED approach described above. Each end fitting belongs to a particular design type and may vary largely in size of diameter as well as flare angle.

[0089] Automating detection of end fitting attributes sufficient to identify the fitting for purposes of, e.g., replacement, is technically challenging. This is due, in part, to end fittings having few features to uniquely identify, e.g., different sizes, genders, flare angles and classes. Manually, e.g., to perform the STAMPED approach for end fittings, measurements must be made with a vernier caliper and specialized angle tools to properly identify the part. Practically, this is a time consuming and cumbersome process prone to human error.

[0090] The system 100 is configured to solve one or more of these technical problems. In particular, the system 100 is configured to automatically identify an end fitting for a hose from one or mor images (e.g., video frames) such that a suitable replacement end fitting (and hose) can be ordered. The end fitting mapping model 142 is configured to extract different predefined attributes from an image and / or video of anexisting connector (such as the image 250 of FIG. 10) to identify the type of end fitting with sufficient attribute particularity that a suitable replacement part can be identified.

[0091] In some examples, one or more of the machine learning models 146 are configured to parse individual frames from a video that is fed to it. The machine learning model(s) can determine information about the hose assembly component imaged in the video, such as the component’s front, back side, top, bottom, interior exterior, and size, based on the sequence of frames.

[0092] Referring to FIG. 12, the end fitting mapping model 142 can include a depth estimator 400, a contour detector 402, an edge detector 404, and a ridge detector 406.

[0093] The depth estimator is configured to estimate the spatial distance of the object (e.g., the end fitting) from the camera 126. The system then processes each frame of the captured video to determine the type of fitting by detecting one or more contours with the contour detector 402, one or more edges with the edge detector 404, and one or more ridges with the ridge detector 406.

[0094] The end fitting mapping model 142 processes detections of one or more contours of the end fitting, one or more edges of the end fitting, and / or one or more ridges of the end fitting to identify the different regions and features of a standard end fitting such as a hexagonal mounting, a circular opening, thread location (interior or exterior), thread dimension, and ratio of a size of a cavity of the end fitting to a size of the overall body of the end fitting itself.

[0095] For example, the contour detector 402 detects a curved surface, the ridge detector 406 detects a number of interior threads and / or exterior threads, and the edge detector 404 detects an inner diameter and outer diameter of a frusto-conical flare with an interior opening.

[0096] Using one or more of a detected contour, a detected edge and / or a detected ridge, the end fitting mapping model 142 determines whether the end fitting is male or female.

[0097] The end fitting mapping model 142 can also include an intensity analyzer 408. Output from the intensity analyzer 408 can be used by one or more of the depth estimator 400, the contour detector 402, the edge detector 404, and / or the ridge detector 406. The intensity analyzer 408 is configured to measure pixel intensity (e.g., brightness) of the pixels of an image (e.g., a frame of a video). For example, the intensity analyzer can generate a pixel intensity map of the image that can be fed to,and processed by, one or more of the depth estimator 400, the contour detector 402, the edge detector 404, and / or the ridge detector 406.

[0098] Based on differentiation of pixel intensity across an image as measured by the intensity analyzer 142, different aspects of the end fitting can be identified, such as depths, relatively dimensions, contours, edges, and ridges, with the end fitting mapping model 142 being trained to recognize patterns in pixel intensity that correspond to each of these physical aspects. From these aspect identifications, which are based on outputs from the depth estimator 400, the contour detector 402, the edge detector 404, and / or the ridge detector 406, the end fitting mapping model 140 can identify the fitting opening and the rest of the fitting body, e.g., by referencing a look-up table that provides a list of available end fittings and their corresponding end fitting connection requirements. The end fitting connection requirements of the correct classification correspond to the identified aspects. The identified aspects can include, for example, the size of the fitting (small radius, medium radius, large radius), whether the fitting is male or female, the angle of an exterior surface of the fitting such as a flare, the location (interior or exterior), number, and diameter of screw threads, and the like.

[0099] By comparing intensity of the fitting opening with respect to the rest of the fitting body, the end fitting mapping model 142 can determine the two highest peaks of intensity, which can help identity the size of the end fitting and, thereby, the type of end fitting, which allows the end fitting mapping model 142 to classify the imaged end fitting.

[0100] According to another process of classifying the end fitting, the end fitting mapping model 142 uses outputs of one or more of the depth estimator 400, the contour detector 402, the edge detector 404, and / or the ridge detector 406 across multiple images (e.g., multiple frames of a video) to detect different shapes (e.g., polygons) that make up the front, side and isometric views of the end fitting. For example, the end fitting mapping model 142 can determine that a particular frame of a video is front view image of the end fitting and contains a circular inner diameter and circular outer diameter, while another frame of the video is an isometric view image of the end fitting that shows a hexagonal mounting portion immediately behind the circular openings. Detection of these shapes and their relative positions and depths relative to one another on the images can allow the end fitting mapping model 142 to classify the end fitting based on a databased of data (e.g., end fitting part catalogs) that the end fitting mapping model 142 has trained on.

[0101] According to another process of classifying the end fitting, the end fitting mapping model 142, using one or more of the depth estimator 400, the contour detector 402, the edge detector 404, and the intensity analyzer 408, is configured to detect two-dimensional polygons or other shapes in different video frames of the end fitting. The detected shapes are assigned confidence scores. The shapes with the highest confidence scores are selected. The selected shapes map to measurements of diameters of hex lengths and side lengths of the end fitting.

[0102] Referring to FIGS. 13-14 exterior surface angles (e.g., angle of a flare) are determined by the end fitting mapping model 142 by calculating a ratio of the inner and outer diameters from one or more frames of the video or other images of the end fitting. The image 500 (which can be a frame of a video) can be captured by the camera 126. For example, the edge detector 404, using an intensity map of the image 500 generated by the intensity analyzer 408, detects an edge 506. The contour detector 402, using the intensity map, detects a contour 505. The edge detector 404, using the intensity map, detects the edge 504. In addition, the depth estimator 400 determines that that the edge 504 is deeper into the image 500 than the edge 506.

[0103] Referring to Fig. 14, and using the foregoing data, the end fitting mapping model 142 calculates a ratio of diameter of the shape 602 corresponding to the edge 506 and the shape 600 corresponding to the edge 504. This ratio, together with the depth data, enable the end fitting mapping model 142 to calculate the angle of the contoured flare 505.

[0104] Armed with the angle of the contoured flare 505, and other aspects of the end fitting 502, such as the hex length and side length as described above, the end fitting mapping model can map the end fitting 502 to a particular classification and, from there, to a suitable replacement part.

[0105] In some examples, where traditional imaging of hose assemblies is impaired due to dirt or dust obscuring the hose assemblies, super resolution imaging techniques can be employed to capture the image data needed for the machine learning models 146 to perform their image analyses, aspect detections, and part classifications and mappings.

[0106] FIG. 15 depicts an example method 700 that can be performed using, at least in part, the end fitting mapping model 142 of FIG. 2.

[0107] At a step 702 of the method 700, a video of an end fitting taken by the camera 126 is uploaded to the server 110.

[0108] At a step 704 of the method 700, frames are extracted, e.g., by parsing them, from the uploaded video. The extraction can be performed by the server 110.

[0109] At a step 706 of the method 700, each frame is analyzed to detect features of the imaged end fitting, such as cavity shapes, intensity patterns, edges, contours, ridges, and the like.

[0110] Based on the data generated at the step 706, at a step 708 of the method 700 the end fitting mapping model 142 calculates a gender score for each frame.

[0111] At a step 710 of the method 700, for each frame, if the gender score is greater than zero, then the end fitting mapping model 142 classifies the imaged end fitting for that frame as female. If the gender score is less than zero, then the end fitting mapping model 142 classifies the imaged end fitting for the frame as male.

[0112] At steps 714 and 716 of the method 700, the end fitting mapping model 142 determines whether the majority of frames have been classified as male or female. If the majority of frames have been classified as female, the end fitting mapping model classifies the imaged end fitting as female. If the majority of frames have been classified as male, the end fitting mapping model classifies the imaged end fitting as male.

[0113] The end fitting mapping model 142 uses the gender classification output from the method 700, together with other detected and identified attributes of the end fitting determined from the frames of the video as described herein, to map the imaged end fitting to a functionally equivalent replacement part.

[0114] As noted above, in some examples, the functionality of the server 110 is hosted, entirely or partially, locally on the client device. For example, instead of the client device 114, the client device can correspond to the client device 814 (FIG. 16) and can include one or more of the machine learning models 146, such as one or more of the layline mapping model 140, the end fitting mapping model 142, and the hose routing model 144.

[0115] The client device 814 can be any suitable electronic device capable of both capturing the photo or video (e.g., with the camera 126) storing one or more software applications (e.g., the application 124) and the underlying one or more machine learning models 146 that the application 124 relies upon to process the photo or video. The client device 814 can be a smartphone, a tablet device, and the like.

[0116] By hosting the functionality of one or more of the machine learning models 146 locally on the client device 814, local and offline processing of photos and / orvideos can be performed quickly and in real-time to classify a hose assembly or hose assembly component (e.g., an end fitting) and / or find a suitable replacement component. Local, real-time processing can be particularly advantageous in certain use cases where network connectivity is poor, e.g., due to the remote location of the component (e.g., out at sea), the component being positioned deep underground, or positioned within an electromagnetic shield, such as a faraday cage. For instance, a suitable replacement part may be available on hand, but without the classification replacement identification functionalities of the machine learning models 146, the technician would be unable to determine which replacement part is needed.

[0117] In situations where it is needed to perform processing remote from the client device 814, the client device 814 is capable of network interactions with the server 110, just like the client device 114, allowing the server 110 to execute one or more functions described herein.

[0118] In some examples, other components of the server device 110 can also be stored on the client device 814, such as one or more of the STAMPED module 132 and the part number module 134.

[0119] Referring to FIG. 7, the graphical interface 200 can be generated following selection of each of the confirm button 190 and the confirm button 199. The graphical interface 200 includes information about the one or more equivalent hose assemblies or hose assembly components that have been identified by the model 140 or the model 142. In some examples, the graphical interface 200 can be generated to display information about equivalent components identified by the module 132 and / or the module 134.

[0120] In the example graphical interface 200, various information about an identified equivalent component is displayed. The information can include an image 202 of the mapped equivalent component, and other specific operational and mechanical information 204 of the mapped equivalent component. The information 204 can include a link to a specification sheet for the mapped equivalent component. In some examples, the ecommerce module 136 can generate a link 206 on the graphical interface 200, selection of which pulls up and displays a website on the I / O device 128 from which the mapped equivalent component can be purchased.

[0121] Following selection of the Scan Routing button 180 (FIG. 4), the client application 124 can, in some examples, generate the graphical interface 208 of FIG. 8.

[0122] Referring to FIG. 8, the graphical interface 208 includes a prompt button 210. Selection of the prompt button 210 can, e.g., cause another application to activate the camera 126 on the client device 114 or pull up an album of stored digital images and / or videos. A digital image or video of a proposed or existing hose assembly routing (e.g., in situ) can be taken with the camera or pulled from the album and uploaded to the server 110 for processing by the hose routing model 144.

[0123] The hose routing model 142 then determines if the captured routing is operationally suitable. In some examples, if the hose routing model 142 determines that the captured routing is suboptimal or otherwise unsuitable, the hose routing model 142 can generate a digital image that demonstrates how to improve the routing while still maintaining the proposed locations of the ends of the hose.

[0124] In some examples, the hose routing model 144 is configured to overlay an image of an improved routing scheme over the digital image (e.g., frame of a video) of the proposed routing that was uploaded. For instance, the hose routing model 142 can cause a routing suggestion diagram 212 to be generated on the graphical interface 208. The diagram 212 shows a digital image of a suggested routing over a digital image of the proposed routing with indicia indicating any flaw(s) in the proposed or existing routing (e.g., too small of a bend radius for the hose in question).

[0125] By performing pixel cluster analysis or other image analysis as described herein on the digital image of the proposed or existing routing, the hose routing model 144 can recognize physical features of the routing, such as the hose type and other information about the hose (e.g., STAMPED data), outer diameter and length of the hose in question for the proposed routing, torsion of the hose for the proposed routing, and one or more bend radii of the hose for the proposed routing. Based on recognition of one of more of these physical and / or operational features, the hose routing model 144 can then perform its rerouting suggestion function to, e.g., generate the visual suggestion 212 and the like. The suggested rerouting can be optimized to maximize one or more variables such, as e.g., the predicted useful life the hose assembly.

[0126] By analyzing photos and / or videos, the system 100 is configured to detect vital indicators of fatigue of the hoses imaged in the videos and / or images and provide alerts and / or recommendations via the client device 114 about replacing the same due to the likelihood of failure.

[0127] In some examples, the system 100 is configured to track the life of a hose using additional sensor information combined with the visual analysis from video orother image input. Such sensors can include sensors that can predict changes in, e.g., ambient conditions at the location of the hose assembly, pressure, temperature, and viscosity changes in the fluid conveyed through the hose assemblies, and vibrations of the hose assembly. Thus, for example, the additional sensors can include one or more of a temperature sensor, a pressure sensor, a humidity sensor, a vibration sensor, and the like. These conditions, together with visual data showing, e.g., bend, torsion, and visual signs of fatigue on the hose assembly, can all impact the predicted usable life of a hose assembly of a known classification. For instance, if the hose assembly is exposed to high vibrations in a warm and humid environment, that hose assembly will be predicted by the hose routing model 144 to fail sooner than if those conditions were not present.

[0128] The sensed conditions and images of hose assembly usage can allow the hose routing model 144 to predict ordering time for a hose or another component of the hose assembly.

[0129] The hose routing model 144 can also generate usage data that is provided to the hose assembly part manufacturer (e.g., automatically), which can inform the manufacturer about performance of particular hose assembly components in different conditions after various numbers of cycles, which can assist the manufacturer in modifying designs of hoses and hose assemblies for improved performance.

[0130] FIG. 11 depicts example methods 300 that can be performed using the system of FIG. 1. Illustrated steps of one or more of the methods 300 need not performed in the illustrated sequence in order to perform a method of the present disclosure. In addition, not all of the steps of each of the methods 300 need be performed in order to perform a method of the present disclosure.

[0131] Referring to FIG. 11, at a step 302 the client application 124 is opened on the client device 114. In some examples, login credentials are provided to access the functionality of the client application 124.

[0132] At a step 304, a pathway selection is received via the application 124. The pathway selection can be, e.g., a selection of one of the buttons of the graphical interface 170 of FIG. 4.

[0133] If at the step 304 the part number pathway is selected, the method proceeds to the step 314, in which a part number of a hose or other hose assembly component is entered and transmitted to the server 110 such that the part number module 134 captures the entered part number. The method then advances to the step 324 at whichthe part number module 134 determines if there is an equivalent component corresponding to the entered part number.

[0134] If the part number module 134 does not identify an equivalent component, the method can return to the step 304 or to the step 314 to prompt the user to enter another part number or to select another pathway. If at the step 324 the part number module 134 does identify an equivalent component, the method proceeds to the step 332 at which information about the equivalent component is displayed on the display device. The method can then advance to the step 346 at which the client application 124 can facilitate, using the eCommerce module 136, purchase of the equivalent component, as described herein.

[0135] If at the step 304 the STAMPED process pathway is selected, the method proceeds to the step 312, in which STAMPED data for a hose is entered and transmitted to the server 110 such that the STAMPED module 132 captures the entered STAMPED data. The method then advances to the step 322 at which the STAMPED module 132 determines if there is an equivalent hose or hose assembly corresponding to the entered STAMPED data.

[0136] If the STAMPED module 132 does not identify an equivalent hose assembly, the method can return to the step 304 or to the step 312 to prompt the user to enter other STAMPED data or to select another pathway. If at the step 322 the STAMPED module 132 does identify an equivalent component, the method proceeds to the step 332 at which information about the equivalent component is displayed on the display device. The method can then advance to the step 346 at which the client application can facilitate, using the eCommerce module 136, purchase of the equivalent hose or hose assembly, as described herein.

[0137] If at the step 304 the layline scan process pathway is selected, the method proceeds to the step 306, in which a digital image or video is captured as described above. The method then advances to the step 316 at which the layline mapping model 140 determines if the minimum needed features have been captured in the provided digital image or video.

[0138] In some examples, at the step 316, the layline mapping model 140 determines if at least a brand logo or brand name and a hose part number are recognizable from the provided digital image or video.

[0139] If at the step 316, the layline mapping model 140 determines that the minimum features are not recognizable in the provided image or video, then the methodcan return to the step 304 or to the step 306 to prompt the user to enter another layline image or to select another pathway.

[0140] If at the step 316 the layline mapping model 140 determines that the minimum features are recognizable in the provided image or video, then the method advances to the step 326 at which the layline mapping model 140 identifies physical and operational characteristics of the imaged hose.

[0141] The method then advances to the step 334 at which the layline mapping model 140 determines if there is an operationally equivalent hose to the identified hose. If no such equivalent is identified, the user is so notified and the method advances to the step 336 at which the layline training data 122 and the layline mapping model algorithm(s) are updated to integrate the new layline information for future equivalents predictions.

[0142] If an equivalent is identified at the step 334, the method advances to the step 332 at which information about the equivalent hose is displayed on the display device. The method can then advance to the step 346 at which the client application 124 can facilitate, using the eCommerce module 136, purchase of the equivalent hose or hose assembly, as described herein.

[0143] If at the step 304 the end fitting scan process pathway is selected, the method proceeds to the step 308, in which a digital image or video is captured as described above. The method then advances to the step 318 at which the end fitting mapping model 142 determines if the minimum needed features have been captured in the provided digital image.

[0144] In some examples, at the step 318, the end fitting mapping model 142 determines if at least a flare, a flare dimension, and a thread dimension are recognizable from the provided digital image.

[0145] If at the step 318, the end fitting mapping model 142 determines that the minimum features are not recognizable in the provided image or video, then the method can return to the step 304 or to the step 308 to prompt the user to enter another end fitting image or video or to select another pathway.

[0146] If at the step 318 the end fitting mapping model 142 determines that the minimum features are recognizable in the provided image or video, then the method advances to the step 328 at which the end fitting mapping model 142 identifies physical and operational characteristics of the imaged end fitting.

[0147] The method then advances to the step 338 at which the end fitting mapping model 142 determines if there is an operationally equivalent end fitting to the identified end fitting. If no such equivalent is identified, the user is so notified and the method advances to the step 336 at which the end fitting training data 120 and the end fitting mapping model algorithm(s) are updated to integrate the new end fitting information for future equivalents predictions.

[0148] If an equivalent is identified at the step 336, the method advances to the step 332 at which information about the equivalent end fitting is displayed on the display device. The method can then advance to the step 346 at which the client application 124 can facilitate, using the eCommerce module 136, purchase of the equivalent end fitting, as described herein.

[0149] If at the step 304 the hose routing scan process pathway is selected, the method proceeds to the step 310, in which a digital image or video is captured as described above. The method then advances to the step 320 at which the hose routing model 144 determines if the minimum needed features have been captured in the provided digital image or video.

[0150] In some examples, at the step 320 the hose routing model 144 determines if at least one critical routing area of a hose is recognizable from the provided digital image or video. An example of a critical routing area is the first inch, or the first two inches, or the first three inches, or the first four inches, or the first five inches, of axial length of a hose extending from an end fitting that is secured to the hose.

[0151] If at the step 320, the hose routing model 144 determines that the minimum features are not recognizable in the provided image or video, then the method can return to the step 304 or to the step 310 to prompt the user to enter another hose routing image or to select another pathway.

[0152] If at the step 320 the hose routing model 144 determines that the minimum features are recognizable in the provided image or video, then the method advances to the step 330 at which the hose routing model 144 measures one or more dimensions in the at least one critical routing area and determines suitable hose routing for the hose in question, e.g., a hose routing that falls within predefined suitability parameters for the hose in question.

[0153] The method then advances to the step 342 at which the hose routing model 144 determines if the imaged hose routing is suitable by comparing it to the determined suitable routing. If the imaged hose routing is found to be suitable, the user is sonotified at the step 340 and the method advances to the step 336 at which the hose routing training data 118 and the hose routing model algorithm(s) are updated to integrate the new routing for future routing suitability predictions.

[0154] If at the step 342 the hose routing model 144 determines that the imaged routing is unsuitable, the method advances to the step 344 at which the determined suitable routing is displayed (e.g., overlayed onto the imaged routing).

[0155] The embodiments described herein may be employed using software, hardware, or a combination of software and hardware to implement and perform the systems and methods disclosed herein. Although specific devices have been recited throughout the disclosure as performing specific functions, one of skill in the art will appreciate that these devices are provided for illustrative purposes, and other devices may be employed to perform the functionality disclosed herein without departing from the scope of the disclosure. In addition, some aspects of the present disclosure are described above with reference to block diagrams and / or operational illustrations of systems and methods according to aspects of this disclosure. The functions, operations, and / or acts noted in the blocks may occur out of the order that is shown in any respective flowchart. For example, two blocks shown in succession may in fact be executed or performed substantially concurrently or in reverse order, depending on the functionality and implementation involved.

[0156] This disclosure describes some embodiments of the present technology with reference to the accompanying drawings, in which only some of the possible embodiments were shown. Other aspects may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments were provided so that this disclosure was thorough and complete and fully conveyed the scope of the possible embodiments to those skilled in the art.

[0157] Although specific embodiments are described herein, the scope of the technology is not limited to those specific embodiments. Moreover, while different examples and embodiments may be described separately, such embodiments and examples may be combined with one another in implementing the technology described herein. One skilled in the art will recognize other embodiments or improvements that are within the scope and spirit of the present technology. Therefore, the specific structure, acts, or media are disclosed only as illustrative embodiments. The scope of the technology is defined by the following claims and any equivalents therein.

Claims

CLAIMSWhat is claimed is:

1. A method of mapping a hose assembly or a hose assembly component, comprising:using a server to:generate, on a display device of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of a layline of the hose assembly or the hose assembly component; andprocess a digital image transmitted from the electronic device and received by the server in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the layline,wherein when the server determines that the digital image includes at least the minimum set of data of the layline, the server is configured to map the hose assembly component or a hose of the hose assembly to a different hose of equivalent mechanical specifications to the hose assembly component or the hose; andwherein when the server determines that the digital image does not include at least the minimum set of data of the layline, the server is configured to reject the digital image.

2. The method of claim 1, wherein the natural language processing model is a machine learning model that includes a named entity recognition (NER) model.

3. The method of claim 1, further comprising using the server to generate at least one other graphical interface on the display device of the electronic device, the at least one other graphical interface including information about the different hose.

4. The method of claim 3, wherein the at least one other graphical interface includes one or more data entry fields configured to receive payment information to order the different hose.

5. The method of claim 1, further comprising determining, by the server, that thedigital image includes the hose, including recognizing pixels of the digital image as corresponding to boundaries of the hose.

6. The method of claim 5, wherein the server is configured to identify whether the digital image includes at least the minimum set of data positioned between the boundaries of the hose.

7. The method of claim 1, wherein when the server determines that the digital image includes at least the minimum set of data of the layline, the method further comprising using the server to generate at least one other graphical interface on the display device, the at least one other graphical interface including another prompt to identify an end fitting for fitting onto an end of the different hose.

8. A method of mapping a hose assembly, comprising:using a server to:generate, on a display device of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a first prompt to transmit a first image of a first portion of the hose assembly and a second prompt to transmit a second image of a second portion of the hose assembly, the second portion being different from the first portion; anddetermine whether a digital image transmitted from the electronic device and received by the server includes an image of the first portion of the hose assembly or of the second portion of the hose assembly,wherein when the digital image is received by the server in response to the first prompt and the server determines that the digital image does not include the first image of the first portion, the server is configured to reject the digital image;wherein when the digital image is received by the server in response to the first prompt and the server determines that the digital image does include the first image of the first portion, the server is configured to accept the digital image;wherein when the digital image is received by the server in response to the second prompt and the server determines that the digital image does not include the second image of the second portion, the server is configured to reject the digital image; andwherein when the digital image is received by the server in response to the second prompt and the server determines that the digital image does include the second image of the second portion, the server is configured to accept the digital image.

9. The method of claim 8,wherein the first portion of the hose assembly includes a layline of a hose; and wherein the second portion of the hose assembly includes an end fitting of the hose.

10. The method of claim 8, wherein one of the first portion of the hose assembly and the second portion of the hose assembly includes a physical routing path of a hose.

11. A method of mapping a hose assembly or a hose assembly component, comprising:using a server to:generate, on a display device of an electronic device operatively linked to the server via a network, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of the hose assembly component or an end fitting of the hose assembly; andprocess a digital image transmitted from the electronic device and received by the server in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the hose assembly component or the end fitting, wherein when the server determines that the digital image includes at least the minimum set of data of the hose assembly component or the end fitting, the server is configured to map the hose assembly component or the end fitting to a different end fitting of equivalent mechanical specifications to the hose assembly component or the end fitting;wherein when the server determines that the digital image does not include at least the minimum set of data of the hose assembly component or the end fitting, the server is configured to reject the digital image; andwherein the minimum set of data includes image data corresponding to a thread of the hose assembly component or the end fitting and image data corresponding to a flare of the hose assembly component or the end fitting.

12. A system for mapping hose assemblies, comprising:one or more processors; andnon-transitory computer readable storage storing instructions which, when executed by the one or more processors, cause the system to:generate, on a display device of an electronic device, a graphical interface, the graphical interface including a first prompt to transmit a first image of a first component of a hose assembly and a second prompt to transmit a second image of a second component of the hose assembly, the first component being different from the second component;in response to receipt of a selection of the first prompt and receipt of the first image:feed the first image to a first machine learning model and not to a second machine learning model; anddetermine, using the first machine learning model, if the first image includes a first minimum set of data corresponding to the first component; andin response to receipt of a selection of the second prompt and receipt of the second image:feed the second image to the second machine learning model and not to the first machine learning model; anddetermine, using the second machine learning model, if the second image includes a second minimum set of data corresponding to the second component.

13. The system of claim 12, wherein the first minimum set of data is different from the second minimum set of data.

14. The system of any of claims 12-13, wherein the first minimum set of data corresponds to a layline of a hose of the hose assembly.

15. The system of any of claims 12-14, wherein the second minimum set of data corresponds to an end fitting of the hose assembly.

16. The system of any of claims 12-15, wherein the second minimum set of data corresponds to a physical routing path of a hose of the hose assembly.

17. The system of any of claims 12-15,wherein the graphical interface includes a third prompt to transmit a third image of a third component of the hose assembly, the third component being different from the second component and different from the first component, andwherein the instructions, when executed by the one or more processors, cause the system to:in response to receipt of a selection of the third prompt and receipt of the third image:feed the third image to a third machine learning model and not to the second machine learning model or to the first machine learning model; anddetermine, using the third machine learning model, if the third image includes a third minimum set of data corresponding to the third component.

18. The system of claim 17, wherein third minimum set of data is different from the second minimum set of data and different from the first minimum set of data.

19. The system of any of claims 17-18,wherein the first minimum set of data corresponds to a layline of a hose of the hose assembly;wherein the second minimum set of data corresponds to an end fitting of the hose assembly; andwherein the third minimum set of data corresponds to a physical routing path of the hose of the hose assembly.

20. The system of any of claims 12-19, wherein the instructions, when executed by the one or more processors, cause the system to:map, using the first machine learning model and based on the first image, a hose of the hose assembly to a different hose of equivalent mechanical specifications to thehose; andmap, using the second machine learning model and based on the first image and the second image, an end fitting of the hose of the hose assembly to a different end fitting of different mechanical specification to the end fitting.

21. The method of claim 11, wherein the digital image is a frame of a video transmitted from the electronic device and received by the server.

22. The method of claim 21,wherein the video includes a plurality of frames; andwherein the server determines that the minimum set of data of the hose assembly component or the end fitting is included in at least two of the plurality of frames.

23. The method of claim 21, wherein the minimum set of data includes an inner diameter and an outer diameter.

24. The method of claim 11, further comprising:using the server to:calculate an angle of the flare based on the minimum set of data; and determine whether the end fitting is male or female based on the minimum set of data.

25. A method of classifying an end fitting, comprising:extracting frames from a video;providing the frames to a machine learning model;calculating, by the machine learning model and for each of the frames, an attribute score; andclassifying an attribute of the fitting based on each attribute score.

26. The method of claim 25, wherein the attribute is classified as male or female.

27. The method of claim 25, wherein the attribute is a size of the end fitting.

28. The method of claim 25, wherein the attribute is an angle of a flare of the end fitting.

29. The method of claim 25, further comprising:detecting, by the machine learning model, a ridge in the frames; and calculating, by the machine learning model, a diameter of a thread corresponding to the ridge.

30. The method of claim 25, further comprising using the machine learning model to:detect an inner diameter of the end fitting in the frames;detect an outer diameter of the end fitting in the frames;detect a contoured surface of the end fitting in the frames, the contoured surface defining a flare of the end fitting and extending from the inner diameter to the outer diameter;calculate ratio of the inner diameter and the outer diameter; andbased on the ratio, determine an angle of the contoured surface.

31. The method of claim 25, further comprising using the machine learning model to:detect a plurality brightness peaks in the frames; anddetermine a size of the end fitting based on the relative positions of the brightness peaks.

32. The method of any of claims 25-31, further comprising using the machine learning model to calculate relative depths of attributes of the end fitting in the frames.

33. The method of claim 25, wherein the attribute is a shape of a portion of the end fitting.

34. The method of claim 25, further comprising using the machine learning model to detect in the frames each of a contour, an edge, and a ridge.

35. The method of any of claims 25-34, further comprising using the machinelearning model to map the end fitting to a different end fitting of equivalent mechanical specifications to the end fitting.

36. The method of any of claims 25-35, wherein an entirety of the method is performed by a single electronic device.

37. The method of any of claims 25-36, further comprising capturing the video with a camera of an electronic device,wherein the machine learning model is stored on the electronic device.

38. The method of any of claims 25-37, wherein the end fitting is an adapter capable of sealingly connecting two other end fittings.

39. The method of any of claims 25-37, wherein the end fitting is configured to be mounted to an end of a hose to sealingly connect the hose to other equipment.

40. The system of claim 12,wherein the first image and the second image are captured by a camera of the electronic device; andwherein the first machine learning model and the second machine learning model are stored on the electronic device.

41. A method of mapping hose assembly component, comprising:generating, on a display device of an electronic device, at least one graphical interface, the at least one graphical interface including a prompt to transmit an image of the hose assembly component or an end fitting of the hose assembly; and processing a digital image captured by a camera of the electronic device in response to the prompt with a natural language processing model to determine whether the digital image includes at least a minimum set of data of the hose assembly component or the end fitting,wherein when the electronic device determines that the digital image includes at least the minimum set of data of the hose assembly component or the end fitting, the electronic device is configured to map the hose assembly component or the end fitting to a different end fitting of equivalent mechanical specifications to the hose assemblycomponent or the end fitting;wherein when the electronic device determines that the digital image does not include at least the minimum set of data of the hose assembly component or the end fitting, the electronic device is configured to reject the digital image; andwherein the minimum set of data includes image data corresponding to a thread of the hose assembly component or the end fitting and image data corresponding to a flare of the hose assembly component or the end fitting.

42. The method of claim 41, wherein the digital image is a frame of a video.

43. The method of claim 42,wherein the video includes a plurality of frames; andwherein the electronic device determines that the minimum set of data of the hose assembly component or the end fitting is included in at least two of the plurality of frames.

44. The method of claim 41, wherein the minimum set of data includes an inner diameter and an outer diameter.

45. The method of claim 41, further comprising:calculating an angle of the flare based on the minimum set of data; and determining whether the end fitting is male or female based on the minimum set of data.

Images