Recommending spare parts for an entrance system

Abstract

A computer-implemented method (100) for providing spare part recommendations (70) for an entrance system (10) is provided. The method (100) comprises receiving (110) a set of imagery (60) of the entrance system (10); applying (120) a feature extraction method (62) to the set of imagery (60), the feature extraction method (62) providing first estimation outputs (63) of an entrance system type estimation; providing (130) the set of imagery (60) as input to a machine learning model (64) trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model (64) providing second estimation outputs (65) of an entrance system type estimation; identifying (140) an entrance system type (68) of the entrance system (10) based on a combination of the first estimation outputs (63) and the second estimation outputs (65); and providing (150) one or more spare part recommendations (70) matching the identified entrance system type (68).

Description

[0001] RECOMMENDING SPARE PARTS FOR AN ENTRANCE SYSTEM

[0002] TECHNICAL FIELD

[0003] The present invention generally relates to the field of entrance systems. More specifically, the present invention relates to computer-implemented methods of recommending spare parts for entrance systems, as well as to an associated computerized system, computer program product and non-transitory computer readable storage medium.

[0004] BACKGROUND

[0005] Entrance system installations having automatic door operators are frequently used for providing automatic opening and / or closing of one or more movable door members in order to facilitate entrance and exit to buildings, rooms and other areas. Servicing and installing entrance systems pose significant challenges, such as challenges relating to the extensive knowledge required to identify the specific type of door involved. Entrance systems come in different forms, including automatic overhead sectional doors, sliding doors, swing doors, and revolving doors, each with unique components and mechanisms. The lack of standardization across manufacturers complicates matters, as service technicians must be familiar with the distinct characteristics and requirements of various products. This absence of uniformity makes it challenging for technicians to adopt a standardized approach, necessitating a detailed understanding of each system.

[0006] The risk of ordering incorrect replacement parts is a major concern in the absence of precise knowledge about the particular entrance system. This can lead to delays, additional costs, and customer dissatisfaction, as the correct components must be re-ordered. Furthermore, the potential for incorrect installations poses significant risks. Without accurate identification of the entrance system, there is an increased likelihood of installing the wrong parts, compromising both functionality and security features. Safety concerns become paramount, as incorrect installations may create hazards for people and property around and compromise the overall integrity of the entrance system. It is in view of the above-identified concerns and others that the present inventors have realized that there is room for improvements in this field.

[0007] SUMMARY

[0008] An object of the present invention is therefore to provide one or more improvements in spare part recommendations for entrance systems. A first aspect of the present invention is a computer-implemented method for providing spare part recommendations for an entrance system. The method comprises receiving a set of imagery including at least two different portions of the entrance system; applying a feature extraction method to the set of imagery, the feature extraction method providing first estimation outputs of an entrance system type estimation; providing the set of imagery as input to a machine learning model trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model providing second estimation outputs of an entrance system type estimation; identifying an entrance system type of the entrance system based on a combination of the first estimation outputs and the second estimation outputs; and providing one or more spare part recommendations matching the identified entrance system type.

[0009] The present invention according to the first aspect effectively recognizes the specific components and configurations of entrance systems, and presents one or more recommendations for spare parts that fit the identified type of entrance system. Businesses can thus significantly reduce the potential for erroneous installations, ensuring both optimal functionality and enhanced security features. Moreover, chances of re-ordering incorrect parts are minimized which streamlines replacement processes. In addition, the integration of automated systems for installation verification adds an extra layer of assurance, reducing the likelihood of installing the wrong components and addressing safety concerns. These technical advancements collectively contribute to a more efficient and secure approach, alleviating delays, additional costs, and customer dissatisfaction associated with entrance system maintenance and installation.

[0010] In one or more embodiments, the identifying comprises generating combined estimation outputs as a combination of the first and second estimation outputs, and selecting the entrance system type as an entry with the highest estimation confidence from among entries of the combined estimation outputs. In one or more embodiments, the combined estimation outputs are generated using one or more of: a weighted average of the first and second estimation outputs, a majority voting of the first and second estimation outputs, a softmax function of the first and second estimation outputs, and a Bayesian fusion of the first and second estimation outputs.

[0011] In one or more embodiments, the method further comprises: increasing estimation confidence for entries among the combined estimation outputs that are common in the first and second estimation outputs, and decreasing estimation confidence for entries among the combined estimation outputs that are unique in either one of the first and second estimation outputs.

[0012] In one or more embodiments, the spare part recommendations match portions of the entrance system included in the set of imagery.

[0013] In one or more embodiments, the spare part recommendations match portions of the entrance system not included in the set of imagery.

[0014] In one or more embodiments, the spare part recommendations include a recommendation factor being one or more of an aging property, a compliance property, a material fatigue property, a corrosion property, an obsolescence property, a manufacturing defect property, an environmental condition property, a maintenance neglect property, and an operational stress property.

[0015] In one or more embodiments, applying the feature extraction method comprises identifying unique indicia in the set of imagery being based on one or more of measurements, coordinates, textures, shapes, edges and color values.

[0016] In one or more embodiments, the unique indicia is used as input to train the machine learning model.

[0017] In one or more embodiments, the machine learning model provides the second estimation outputs based on a classification of the unique indicia provided to the machine learning model.

[0018] In one or more embodiments, the method further comprises applying an image preprocessing method to the set of imagery before providing estimation outputs.

[0019] In one or more embodiments, the method further comprises: providing third estimation outputs by applying optical character recognition to the set of imagery, wherein in addition to the first and second and estimation outputs, the identifying is further based on the third estimation outputs.

[0020] In one or more embodiments, the set of imagery includes a front panel portion of a movable door member of the entrance system, a rear panel portion of the movable door member, and a hinge portion of a hinge mechanism of the movable door member.

[0021] In one or more embodiments, the providing of one or more spare part recommendations comprises sending a request for parts to a digital product store, the request involving the entrance system type, wherein the digital product store returns one or more spare part recommendations matching the requested entrance system type.

[0022] In a second aspect of the present invention a computer-implemented method for obtaining spare part recommendations for an entrance system is provided. The method comprises capturing a set of imagery including at least two different portions of the entrance system; transmitting the set of imagery to a backend computing service, the backend computing service being configured to: receive the set of imagery; apply a feature extraction method to the set of imagery, the feature extraction method providing first estimation outputs of an entrance system type estimation; provide the set of imagery as input to a machine learning model trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model providing second estimation outputs of an entrance system type estimation; identify an entrance system type of the entrance system based on a combination of the first estimation outputs and the second estimation outputs; and provide one or more spare part recommendations matching the identified entrance system type; and obtaining the one or more spare part recommendations from the backend computing service.

[0023] In a third aspect of the present invention a computerized recommendation system for an entrance system is provided. The computerized recommendation system comprises a backend computing service configured to receive a set of imagery including at least two different portions of the entrance system; apply a feature extraction method to the set of imagery, the feature extraction method providing first estimation outputs of an entrance system type estimation; provide the set of imagery as input to a machine learning model trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model providing second estimation outputs of an entrance system type estimation; identify an entrance system type of the entrance system based on a combination of the first estimation outputs and the second estimation outputs; and provide one or more spare part recommendations matching the identified entrance system type.

[0024] In a fourth aspect of the present invention a computer program product is provided. The computer program product comprises computer code for performing the method according to the first or second aspect when the computer program code is executed by a processing device.

[0025] In a fifth aspect of the present invention a non-transitory computer readable storage medium is provided. The non-transitory computer readable storage medium has stored thereon a computer program comprising computer program code for performing the method according to the first or second aspect when the computer program code is executed by a processing device.

[0026] The advantages discussed in relation to the first aspect may be envisaged for either one of the second, third, fourth or fifth aspect as well. The provision of such computer-implemented methods, computerized system, computer program product and non-transitory computer readable storage medium will solve or at least mitigate one or more of the problems or drawbacks identified in the above, as will be clear from the following detailed description section and the drawings.

[0027] Embodiments of the invention are defined by the appended dependent claims and are further explained in the detailed description section as well as in the drawings.

[0028] It should be emphasized that the term “comprises / comprising” when used in this specification is taken to specify the presence of stated features, integers, steps, or components, but does not preclude the presence or addition of one or more other features, integers, steps, components, or groups thereof.

[0029] All terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a / an / the [element, device, component, means, step, etc.]" are to be interpreted openly as referring to at least one instance of the element, device, component, means, step, etc., unless explicitly stated otherwise.

[0030] The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Objects, features and advantages of embodiments of the invention will appear from the following detailed description, reference being made to the accompanying drawings.

[0032] FIG. 1A is a schematic block diagram of an exemplary entrance system for which spare parts can be recommended generally according to the present invention.

[0033] FIG. IB is a more detailed example of an automatic door operator shown in FIG. 1 A and associated units.

[0034] FIG. 2A is a side view of an entrance system embodied as an overhead door system.

[0035] FIG. 2B is a perspective view of the overhead door system shown in FIG. 2A.

[0036] FIG. 3 depicts some general methods for recommending spare parts for an entrance system according to various examples.

[0037] FIG. 4 is a schematic flowchart diagram depicting exemplary steps of a computer-implemented method for obtaining spare part recommendations for an entrance system.

[0038] FIG. 5 is a schematic block diagram depicting exemplary software modules involved in a computer-implemented method for recommending spare parts for an entrance system.

[0039] FIGs. 6A-6E are exemplary illustrations of hinge portions of an entrance system from which unique indicia combinations can be extracted in various examples.

[0040] FIG. 7 is an exemplary illustration of how two estimation outputs can be combined for purposes of providing spare part recommendations for an entrance system according to some embodiments.

[0041] FIG. 8 is a flowchart diagram illustrating a computer-implemented method implemented by computer backend functionality for recommending spare parts for an entrance system generally according to the present invention.

[0042] FIG. 9 is a flowchart diagram illustrating a computer-implemented method implemented by computer frontend functionality for obtaining spare part recommendations for an entrance system generally according to the present invention. FIG. 10 is a schematic illustration of a non-transitory computer-readable storage medium in one exemplary embodiment, capable of storing a computer program product.

[0043] DETAILED DESCRIPTION OF EMBODIMENTS

[0044] Embodiments of the invention will now be described with reference to the accompanying drawings. The invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. The terminology used in the detailed description of the particular embodiments illustrated in the accompanying drawings is not intended to be limiting of the invention. In the drawings, like numbers refer to like elements.

[0045] FIG. 1A is a schematic block diagram illustrating an entrance system 10 to which the inventive aspects of the present invention may be applied. The entrance system 10 comprises one or more movable door members 12-1...12-n, and an automatic door operator 20 for causing movements of the door members 12-1...12-n between closed and open positions. In FIG. 1, a transmission mechanism 13 conveys mechanical power from the automatic door operator 20 to the movable door members 12-1...12-n.

[0046] As can be seen in FIG. 1 A, a control arrangement 21 is provided for the entrance system 10. The control arrangement 21 comprises a controller 22, which may be part of the automatic door operator 20, but which may be a separate device in other embodiments. The control arrangement 21 also comprises a plurality of sensor units S-

[0047] 1...5-n. Each sensor unit S-l...S-n may generally be connected to the controller 22 by wired connections, wireless connections, or any combination thereof.

[0048] The embodiment of the automatic door operator 20 shown in FIG. IB will now be described in more detail. The automatic door operator 20 may typically be arranged in conjunction with a frame or other structure which supports the door members 12-

[0049] 1...12-n for movements between closed and open positions, often as a concealed overhead installation in or at the frame or support structure.

[0050] In addition to the aforementioned controller 22, the automatic door operator 20 comprises a motor 24, typically an electrical motor, being connected to an internal transmission (or gearbox) 25. An output shaft of the transmission 25 rotates upon activation of the motor 24 and is connected to the external transmission mechanism 13. The external transmission mechanism 13 translates the motion of the output shaft of an internal transmission 25 into an opening or a closing motion of one or more of the door members 12-1...12-n with respect to the frame or support structure.

[0051] The automatic door operator 20 has a power unit 28b that supplies power to the electric motor 24, controller 22 and other components of the automatic door operator 20 as appropriate. The power unit 28b typically comprises an AC / DC converter, such as a switch mode power supply (SMPS), having an input end coupled to AC mains 28a and an output end for supplying internal DC power to the electric motor 24, controller 22, etc.

[0052] In addition to the power unit 28b, the automatic door operator 20 furthermore comprises a battery 29 that may supply power to the electric motor 24, etc., for instance in an evacuation operating mode of the automatic door operator 20, or in times of AC mains power shortage. In the disclosed embodiment, the battery 29 is coupled for charging by the power unit 28b. In other embodiments, the battery 29 may be charged by other means, such as external battery charging equipment. Preferably, therefore, the battery 29 is a rechargeable battery made from, for instance, lithium-ion (Li-ion), lithium-ion polymer (Li-ion polymer), nickel-metal hydride (NiMH), nickel-cadmium (NiCd) or lead-acid technology.

[0053] The controller 22 is arranged for performing different functions of the automatic door operator 20, typically in different operational modes (states) of the entrance system 10, using inter alia sensor input data from the plurality of sensor units S-l...S-n. Hence, the controller 22 is operatively connected with the plurality of sensor units S-l...S-n. At least some of the different functions performable by the controller 22 have the purpose of causing desired movements of the door members 12-1...12-n. To this end, the controller 22 has at least one control output connected to the motor 24 for controlling the actuation thereof.

[0054] The controller 22 may be implemented in any known controller technology, including but not limited to a microcontroller, processor (e.g. PLC, CPU, DSP), FPGA, ASIC or any other suitable digital and / or analog circuitry capable of performing the intended functionality. The controller 22 also has an associated memory 23. The memory 23 may be implemented in any known memory technology, including but not limited to E(E)PROM, S(D)RAM or flash memory, or any combination thereof. In some embodiments, the memory 23 may be wholly or partly integrated with or internal to the controller 22. The memory 23 may store program instructions for execution by the controller 22, as well as temporary and permanent data used by the controller 22. The entrance system 10 may have a plurality of control parameters CPl-CPn, the values of which are set by an authorized user (e.g. installer or service person) upon configuration of the entrance system 10. The control parameters CPl-CPn will be used by the automatic door operator 20 for controlling movements of the door members 12- 1...12-n. In the disclosed embodiment, at least some of the control parameters CPl-CPn are stored in the memory 23 of the automatic door operator 20.

[0055] In the embodiment shown in FIG. IB, the entrance system 10 has a communication bus 27. Some or all of the plurality of sensor units S-l...S-n are connected to the communication bus 27, and so is the automatic door operator 20. In the disclosed embodiment, the controller 22 and the memory 23 of the automatic door operator 20 are connected to the communication bus 27; in other embodiments this may be the case for other devices or components of the automatic door operator 20. In still other embodiments, the outputs of the sensor units S-l...S-n may be directly connected to respective data inputs of the controller 22.

[0056] The automatic door operator 20 in FIG. IB is enabled for external data communication 26a by means of a data communication interface 26b which is furthermore connected to the communication bus 27 at 26b. The external data communication 26a is typically made with another communication device or system over a data communication network 26c, such as a wide area network (WAN) or local area network (LAN). Accordingly, the data communication network 26c may comply with any commercially available mobile telecommunications standard, including but not limited to GSM, UMTS, LTE, 5G, D-AMPS, CDMA2000, FOMA and TD-SCDMA. Alternatively or additionally, the data communication network 26c may comply with one or more short-range wireless data communication standards such as Bluetooth®, BLE, WiFi (e.g. IEEE 802.11, wireless LAN), Near Field Communication (NFC), RFID (Radio Frequency Identification) or Infrared Data Association (IrDA). In some embodiments, the communication may be wired, such as TCP / IP over Ethernet. An exemplary type of entrance system 10, more specifically an overhead door system (sometimes referred to as an industrial door system) will now be further described with reference to FIGs. 2A-B in different views. It is presently believed that the present invention is particularly well suited to be applied to this type of entrance system, however without limitation.

[0057] In FIG. 2A, an overhead door system 10 is shown in a side view, while FIG. 2B shows the overhead door system 10 in a perspective view. The overhead door system 10 has a movable door member 12 with a door leaf 11 for covering a door opening in a wall, running in a side frame. The door leaf 11 has a front panel portion 11-1 directed towards a front side of the door leaf 11 and a rear panel portion 11-2 directed toward a rear side of the door leaf 11.

[0058] In this example the door leaf 11 covers a door opening surrounded doublesided by a side frame 16, the wall and by a ceiling above. The side frame has a doublesided vertical track 13, provided as a guiding element on the left and on the right side of the door leaf 11, respectively. Further, the side frame 16 has a double-sided horizontal track 14. The horizontal track 14 runs transversely, in this execution example perpendicular to the vertical track 13, extending rearward to the door leaf 11. A connecting portion 15 with a curved shape connects the vertical track 16 and the horizontal track 14, providing a guideway. The horizontal tracks 14 of the side frame 16 are fixed at a rear side under the ceiling. At a front side the horizontal tracks 14 are fixed at the vertical tracks 13 of the side frame 16, respectively. Accordingly, the overhead door system 10 comprises a pair of vertical tracks 13 and a pair of horizontal tracks 14 connected by a pair of connecting track portions 15. Guiding arrangements may be attached to the door leaf 11 for interfacing with the tracks.

[0059] The door leaf 11 is movable between a vertical closed position and a horizontal, at least partly opened or overhead position within the side frame. Hence, the door leaf 11 may be movable between a vertical closed position and a horizontal position and / or between a vertical closed position and an opened position and / or between a vertical closed position and an overhead position. An overhead position herein refers to a position where the door leaf 11 is lifted up from the ground. In the overhead position, the door leaf 11 may for example be vertically arranged or inclined. The overhead sectional door comprises a plurality of door panel sections 17a-f. The door panel sections 17a-f are arranged to be moved a distance in an upward or downward direction with respect to the ground level G. The door panel sections 17a-f are preferably connected to each other by hinge mechanisms 18 including at least one hinge. FIGs. 2A-2B illustrate an embodiment where five door panels in total 17a-f have been arranged to the doorframe, however, as understood by a person skilled in the art the number of door panels may vary.

[0060] In view of the aforementioned explanations, entrance systems and related components, modules and devices are subjected to different variants, alternatives, versions, options, modifications, etc. These variants can be minor, i.e., having little or no effect on the functionality of the components of the entrance system. They can also be significant, i.e., potentially making various components incompatible with one another for purposes of providing a safe operation. As discussed in the background section, this is due to the lack of standardization across entrance system manufacturers. Moreover, during the lifetime of an entrance system, a certain entrance system installation is oftentimes subjected to a plurality of changes for one or more components, thus increasing the number of possible options on the market. This complicates matters, as authorized users must be familiar with the distinct characteristics and requirements of various products in order to install or service an existing entrance system installation. This absence of uniformity makes it challenging for technicians to adopt a standardized approach for ordering and installing correct spare parts since it necessitates a detailed understanding of each system, which is practically an impossible manual task. The approach presented in this disclosure therefore relates to enabling an authorized user to obtain spare part recommendations for an installable or serviceable entrance system installation.

[0061] With further reference to FIG. 3, an exemplary visualization of providing spare part recommendation(s) is shown. In this example an overhead door system 10 is considered, although other examples may involve any type of entrance system as discussed herein, including but not limited to revolving door systems, swing door systems, or sliding door systems, to name a few. The overhead door system 10 includes a movable door member 12 having a door leaf 11. The door leaf 11 includes a front panel portion 11-1 and a rear panel portion 11-2. For visualization purposes, the front and rear panel portions 11-1, 11-2 are shown as two separate units, although it should be understood that these are parts of the same door leaf 11 but on opposite sides thereof. In addition, a hinge mechanism 18 is shown (zoomed in at a hinge portion 18-1) as attached to the movable door member 12.

[0062] An authorized user 30 is operating a client device 310. The authorized user 30 may be a service personnel, technician, operator, or the like. The client device 310 may be a smartphone, smart tablet, PDA, laptop, smartwatch, AR glasses, or the like. The client device 310 includes an image capturing device capable of capturing one or more images of the surroundings. The image capturing device may be integrated with the client device 310, or alternatively provided external to the client device 310 and capable of communicating the image(s) with the client device 310 using any known communication standards. The client device 310 is an interactive device, meaning that the authorized user 30 can interact with the client device 310 for instructing it to capture one or more images of the surroundings. The client device 310 comprises a display screen configured to present visual information to the authorized user 30. Additionally or alternatively, the client device 310 may be configured to provide other feedback, such as sound, audio, haptics, etc., through suitable devices integrated with or external to the client device 310.

[0063] The client device 310 is configured to communicate with a backend computing service 320, here depicted as a cloud-based server. The communication may be effected using any known communication approaches known in the art, including but not limited to short-range communication interfaces such as IEEE 802.11, IEEE 802.15, ZigBee, WirelessHART, WiFi, Bluetooth®, BLE, RFID, WLAN, MQTT loT, CoAP, DDS, NFC, AMQP, LoRaWAN, Z-Wave, Sigfox, Thread, EnOcean, mesh communication, other forms of proximity-based device-to-device radio communication signal such as LTE Direct, or long-range communication interfaces such as W-CDMA / HSPA, GSM, UTRAN, LTE or Starlink.

[0064] The cloud-based server may be implemented using any commonly known cloud-computing platform technologies, such as e.g. Amazon Web Services, Google Cloud Platform, Microsoft Azure, DigitalOcean, Oracle Cloud Infrastructure, IBM Bluemix or Alibaba Cloud. The cloud-based server may be included in a distributed cloud network that is widely and publically available, or alternatively limited to an enterprise. Alternatively, the server may in some embodiments be locally managed as e.g. a centralized server unit.

[0065] The cloud-based server may be configured to maintain a storage device, being included with or external to the cloud-based server. Connection to cloud-based storage means may be established using DBaaS (Database-as-a-service). For instance, cloudbased storage means may be deployed as a SQL data model such as MySQL, PostgreSQL or Oracle RDBMS. Alternatively, deployments based on NoSQL data models such as MongoDB, Amazon DynamoDB, Hadoop or Apache Cassandra may be used. DBaaS technologies are typically included as a service in the associated cloudcomputing platform.

[0066] Server configurations other than cloud-based may be realized in other examples, for instance based on any type of client-server or peer-to-peer (P2P) architecture. Server configurations may thus involve any combination of e.g. web servers, database servers, email servers, web proxy servers, DNS servers, FTP servers, file servers, DHCP servers, to name a few.

[0067] The present inventors have realized that the objectives described herein can be achieved by way of providing a set of imagery including at least two different portions of the overhead door system 10. A set of imagery shall be understood as including one or more images. Therefore, in some examples one single image in the set of imagery may include a plurality of portions of the overhead door system 10. For example, a picture taken from afar may show both a hinge portion 18-1 and a front panel portion 11-1 of the movable door member 12. In other examples, the set of imagery includes two or more images, each image in the set of imagery including at least one portion of the overhead door system 10. No restrictions shall thus be put in this context. A “portion” shall be understood as an entire component, such as a hinge, or at least parts thereof, such as an upper / lower hinge plate including one or more sets of screw holes or openings.

[0068] In the particular example of FIG. 3, the set of imagery includes three images, each image picturing a respective portion of the overhead door system 10. More specifically, the set of imagery includes a first image of a front panel portion 11-1, a second image of a rear panel portion 11-2, and a third image of a hinge portion 18-1. While this is just an exemplary set of imagery, the present inventors have conducted studies that indicate that these three particular portions can in some cases be sufficient for establishing the identity of the entrance system type using the approaches taught herein. This is due to these three units oftentimes being associated with unique appearances, or at least a distinctive appearance that is associated with fewer possible alternatives compared to other portions of the entrance system. Combined with one another, most entrance systems on the market can currently be accurately identified. For the panel portions 11-1, 11-2, said distinctive appearance can relate to patterns, colors, textures, carvings / engravings, glass inserts, hardware, a number of panels, shapes, material contrast, and the like. For the hinge portion 18-1, said distinctive appearance can relate to designs, shapes, finishes, coatings, materials, number of holes / plates, type of holes / plates, hole arrangements, plate configurations, sharp edges / round edges, ball bearings, and the like.

[0069] In other examples, similar or alternative unique or at least distinctive appearances can be envisaged for any of the other units / components / modules / etc., for which imagery can be captured. Purely by way of non-limiting examples, imagery can be captured for automatic door operators including its components as discussed in relation to FIGs 1 A-B, door components as discussed in relation to FIGs. 2A-B, handles, knobs, locksets, thresholds, windows, lights, door surroundings, architectural moldings, tracks, ventilation openings, emergency exit features, weather stripping and seals, and the like, each having a set involving one of more distinctive features.

[0070] The approaches herein may be discussed from a frontend perspective, i.e., from a perspective of the client device 310, or from a backend perspective, i.e., from a perspective of the backend computing service 320. Backend contexts refer to the serverside of an application program, where data is processed, stored, and managed, typically handling business logic, database operations, and server-side functionalities. Frontend contexts pertain to the client-side of an application program, involving the presentation layer and user interface elements that users interact with directly in their browsers. The frontend perspective will firstly be discussed, with further reference to FIG. 4.

[0071] In FIG. 4, a client device 310 is shown. The client device 310 is configured to obtain spare part recommendations for an entrance system, such as the entrance system 10 as discussed herein. The process of obtaining spare part recommendations according to the teachings herein shall be understood as a continued and / or guided human- machine interaction process. The user operating the client device 310, such as the authorized user 30 as discussed herein, will be assisted in performing the technical task of finding accurate replacement options for parts of the entrance system. This particular computer program is named “PartFinder”, which in this particular example involves a three-step continued and / or guided human-machine interaction process, followed by a computer backend determination of an entrance system type. Other examples may include any number of steps in the guided interaction process, as it is generally based upon how many images that are to be included in the set of imagery.

[0072] In a first step of the guided interaction process, the client device 310 instructs the authorized user to take a picture of a front panel portion 11-1 of the entrance system, more specifically of a door leaf of a movable door member. The client device 310 may continuously guide the user into capturing a sufficiently accurate representation of the front panel portion 11-1. The client device 310 may identify any potential issues with the captured image, such as poor lighting, wrong focus, too zoomed in or zoomed out, blurry view, and the like. The client device 310 may request that the user captures one or more additional images until the quality is sufficient for subsequent analysis, or from different perspectives, or from different distances to the front panel portion 11-1, etc.

[0073] In response to the client device 310 accepting the captured image(ry) with respect to one or more of the above considerations, a second step of the guided interaction process involves capturing a picture of a rear panel portion 11-2. This may involve similar considerations as for the first step.

[0074] In response to the client device 310 accepting the captured image(ry) with respect to one or more of the above considerations, a third step of the guided interaction process involves capturing a picture of a hinge portion 18-1. This may involve similar considerations as for the first and / or second steps.

[0075] After the completion of the guided interaction process, the remaining steps of obtaining spare part recommendations for an entrance system are functionally carried out by a backend computing service, such as the backend computing service 320 as discussed herein. The entrance system is identified, in this case as type “es-123”, and a plurality of spare parts (“htype2”, “fptype7”, “rptypel”) are provided as recommendations to the client device 310. These may be obtained in response to a request to a digital product store that returns recommendations of one or more spare parts matching the requested entrance system type. The digital product store may for example be a web shop, Product Information Management (PIM) system or Enterprise Resource Planning (ERP) system. It shall thus be understood that these particular spare parts have been tailored for the existing type of entrance system installation, thus effectively allowing the authorized user to service and / or install appropriate spare parts to the entrance system based on the returned recommendations. The client device 310 may feature a functionality to automatically order these parts (“Place order”), and optionally have them delivered to a location of the entrance system. This may be reported automatically from the client device 310 to the digital product store.

[0076] The backend perspective, i.e., the identification of the type of entrance system, will now be discussed with further reference to FIG. 5.

[0077] FIG. 5 shows the client device 310 operating in frontend contexts. The client device 310 transmits a set of imagery 60 to the backend computing service 320 operating in backend contexts. The backend computing service 320 is configured to provide one or more spare part recommendations 70 for an entrance system, such as the entrance system 10. The different blocks of software modules or data shown in FIG. 5 with uniform lines shall be interpreted as preferred embodiments, while the blocks with dashed lines shall be interpreted as optional embodiments. The backend computing service 320 may be included in a computerized recommendation system 300. In some examples, the client device 310 is also included in the computerized recommendation system 300.

[0078] The set of imagery 60 is received by the backend computing service 320. This may be done through any conventional communication methods as discussed in relation to FIG. 3. The set of imagery 60 may be received automatically in response to the client device 310 affirming that an accurate representation of the entrance system has been achieved by the images in the set of imagery 60. This occurs in the example of FIG. 3 after the third and last image has been affirmed.

[0079] The set of imagery 60 is then processed in two different ways, namely by a feature extraction method 62 and by a machine learning model 64, respectively. In the following disclosure these two approaches individually provide estimation outputs 63, 65 indicating estimations of types of entrance systems generated by the respective processes of estimation. Generally, each approach approximates values based on information in the set of imagery 60 to predict the likelihood of said imagery 60 representing a known type of entrance system. By combining the estimation outputs 63, 65 for each model 62, 64, an entrance system type 68 can be identified. The spare part recommendations 70 can then be provided for the specific entrance system type 68. The present inventors have achieved a surprisingly high accuracy in determining the correct entrance system type 68 by such a combination of approaches. Accordingly, one or more of the technical advantages mentioned in the summary section can be achieved.

[0080] The first processing of the set of imagery 60 is done by the feature extraction method 62. The feature extraction method 62 transforms raw pixel data of the set of imagery 60 into meaningful and informative predictions, in the form of first estimation outputs 63. The process of feature extraction 62 may involve a plurality of sub-steps, some of which will now be described.

[0081] Before extracting features, a preprocessing method may be applied to the set of imagery 60 to enhance their quality and consistency. The preprocessing method may involve resizing, rotating, translating, shearing, reducing noise, enhancing contrast, enhancing brightness, correcting colors, segmenting, normalizing, removing any irrelevant background clutter, etc.

[0082] The next sub-step may be to represent the images as features for purposes of being able to capture unique indicia of objects in the images. The unique indicia may be based on measurements, coordinates, textures, shapes, edges or pixel values of the set of imagery 60. Various techniques can be employed for feature representation, including color-based features (color distribution of the pixels, such as average color, hue, saturation, and color moments, etc.), texture-based features (spatial arrangement and patterns of pixels, such as edge detection, texture filters, and local binary patterns (LBP), etc.), shape-based features (geometric properties of objects, such as size, orientation, aspect ratio, bounding box dimensions, etc.), and the like.

[0083] The next sub-step may be to perform dimensionality reduction. The sub-step involves a reduction of the number unique indicia while preserving the essential information. This is often done to improve computational efficiency and prevent overfitting.

[0084] The next sub-step may be to identify, classify or recognize the unique indicia. The unique indicia, either in their original form or after dimensionality reduction, are then fed into a classification or recognition algorithm. This algorithm compares the unique indicia to a set of known object classes or patterns and assigns a probability or score for each class, indicating how likely it is that the image represents that particular object type. The probability or score in FIG. 5 is represented by the first estimation outputs 63. The classification or recognition algorithm may vary in complexity. In some examples it may be a piece-wise comparison with a dataset of stored images of entrance system portions for different predetermined entrance system types. In other examples the unique indicia may be fed into the machine learning model 64 and used as classification input. It may additionally or alternatively be used as training input for training the machine learning model 64, which will be discussed in more detail soon.

[0085] FIGs. 6A-E show exemplary unique indicia identifiable by the feature extraction method 62. In these examples five different types of hinge portions 18-1, 18-2, 18-3, 18-4, 18-5 are shown. The feature extraction method 62 has successfully identified five types of indicia for each image of the hinge portions 18-1, 18-2, 18-3, 18-4, 18-5, which are (1) end-holes upper plate, (2) end-holes bottom plate, (3) top holes, (4) top blade, and (5) other. The hinge portions 18-1, 18-2, 18-3, 18-4, 18-5 are respectively uniquely identifiable by a combination of these indicia. The hinge portion 18-1 of FIG. 6A has the unique indicia combination “2-2-2-arrow-null”. The hinge portion 18-2 of FIG. 6B has the unique indicia combination “2-2-2-cone rim-sides”. The hinge portion 18-3 of FIG. 6C has the unique indicia combination “2-2-0-null-rimmed sides”. The hinge portion 18-4 of FIG. 6D has the unique indicia combination “3-3-2-null-thin top near bottom”. The hinge portion 18-5 of FIG. 6E has the unique indicia combination “3-2-0- straight rim-null”. Clearly, other unique indicia combinations can be envisaged, and used together with other identifiable portions of the entrance system, for purposes of determining the entrance system type.

[0086] Returning to FIG. 5, the second processing of the set of imagery 60 is done by the machine learning model 64. Similar to the feature extraction method 62, the machine learning model 64 transforms raw pixel data of the set of imagery 60 into meaningful and informative predictions, in the form of second estimation outputs 65. Since it is a different type of algorithm, the predictions are provided differently.

[0087] The machine learning model 64 is trained on a dataset involving sets of imagery of predetermined entrance systems. The dataset may be continuously updated when new combinations of units are used in an entrance system type, and / or refined when certain combinations of units are changing. The predetermined entrance system in the dataset may be labelled, where each entry includes both image data (and optionally text), and the corresponding ground truth label indicating the correct classification or annotation for each entry. The machine learning model 64 learns from this dataset during a training process, thereby adjusting its parameters, weights, layers, etc., to accurately predict labels for an entrance system type represented by the set of imagery 60.

[0088] The machine learning model 64 may be based on any suitable estimation model known in the art, including but not limited to convolutional neural networks, support vector machines, random forests, K-nearest neighbors, logistic regression, naive Bayes, inception networks, residual networks, gradient boosting machines, or the like.

[0089] As discussed above, the machine learning model 64 may receive inputs not only including the set of imagery 60, but also from unique indicia identified by the feature extraction method 62. These may be provided as training input (e.g. in the dataset involving sets of imagery of predetermined entrance systems as discussed above) for purposes of improving the performance of the machine learning model 64, and / or as classification input for purposes of identifying features that have been extracted by the feature extraction method 62. By combining these methods, the weight of coefficients (for example convolution coefficients for the machine learning model 64 being a convolutional neural network) may be optimized. To this end, negative learning by the machine learning model 64 is prevented, which improves the classification performance and thus the likelihood that the second estimation outputs 65 are more accurate.

[0090] The backend computing service 320 is further configured to identify the entrance system type 68 of the entrance system depicted by the set of imagery 60. This is done based on a combination of the first and second estimation outputs 63, 65. By determining the entrance system type 68 based on a combination of the first and second estimation outputs 63, 65, the likelihood of obtaining a correct identification is increased.

[0091] In some examples, the identification may involve generating combined estimation outputs 66 as a combination of the first and second estimation outputs 63, 65. The entrance system type 68 is then selected as an entry in the combined estimation output 66 with the highest estimation confidence. The highest estimation confidence of an entry in a dataset in comparison with other entries in the dataset refers to the maximum level of certainty or confidence associated with a particular data point within said dataset, relative to the confidence levels of all other entries in that same dataset. This is a quantitative metric indicating the degree of reliability or certainty assigned to a specific estimation or prediction, with the highest estimation confidence signifying the most assured or confident prediction among all entries in the dataset. One such example is shown in FIG. 7, which will now be described with reference thereto.

[0092] FIG. 7 illustrates the first estimation outputs 63, second estimation outputs 65, and combined estimation outputs 66. In this example the estimation outputs 63, 65, 66 are illustrated in the form of matrices with 3x3 entries (ci-C3; n- ), although this is not a necessity. The estimation outputs 63, 65, 66 may be stored and represented in any suitable form including one or more entries, e.g. lists, arrays, sets, maps, or the like, with any suitable size. Moreover, in this example, the probability values corresponding to the estimation confidence in the estimation outputs 63, 65, 66 are herein represented as arbitrary values between 0.00 and 1.00. This is however just an example as they may be represented in any suitable way, including but not limited to numbers, percentages, Boolean indicators, estimation groups (“LESS LIKELY”, “MORE LIKELY”, etc.), and so forth.

[0093] The first and second estimation outputs 63, 65 include nine entries each, each entry corresponding to a particular estimation probability of an entrance system type based on the provided set of imagery. If either one of these were to be considered individually, an incorrect identification would be obtained. The first estimation outputs 63 suggests that entry C2, is the correct type, while the second estimation outputs 65 indicate that the entry C2, n is the correct type. However, when combining these in the combined estimation outputs 65 both of these entries have a zero percent chance of being the correct type. This is due to the fact that the entry C2, n of the second estimation outputs 65 indicates a zero percent chance of being correct, and the entry C2, ri of the first estimation outputs 63 indicates a zero percent chance of being correct. When combining these using piece-wise element multiplication as done in this example, the combined estimation outputs 66 will thus indicate a zero percent chance for both of these entries. Instead, it is entry C3, r2 that involves the highest probability of a correct entrance system type in the combined estimation outputs 66, as both the first and second estimation outputs 63, 65 have respectively indicated a high chance of success for the type of this entry (0.99 vs. 0.97). To this end, the entry C3, r2 involves the highest estimation confidence from among entries of the combined estimation outputs 66, and is therefore selected as the most likely hit as the entrance system type 68.

[0094] While the illustration of FIG. 7 utilizes piece- wise element multiplication of elements for calculating the combined estimation outputs 66, this shall not be seen as limiting to the scope. Other suitable methods for generating the combined estimation outputs 66 by combining the first and second estimation outputs 63, 65 may be envisaged, hi other examples, the combined estimation outputs 66 may be generated using one or more of a weighted average, a majority voting, a softmax function, and a Bayesian fusion of the first and second estimation outputs 63, 65. In any event, the entry in the combined estimation outputs 66 with the highest estimation confidence compared to other entries therein is preferably selected as the entrance system type 68.

[0095] Regardless of what type of method is used to generate the combined estimation outputs 66, this may in some examples involve increasing estimation confidence for entries that are common in the first and second estimation outputs 63, 65. At the same time, estimation confidence for entries that are unique in either one of the first and second estimation outputs 63, 65 may be reduced. This bolsters the concept of trusting the predictions of the combined estimation outputs 66 more than the individual estimation outputs 63, 65, as a prediction is more likely correct if it is found in two separate models rather than one single model.

[0096] Returning to FIG. 5, another optional step is shown where the entrance system type identification in some examples are further based on third estimation outputs 61. The identification is in these examples thus based on the first 63, second 65, and third 61 estimation outputs. The third estimation outputs are obtained by applying optical character recognition (OCR) to the set of imagery 60. OCR identifies and transforms text embedded in the set of imagery 60 into an analyzable format preferably involving exact characters. The text may be provided directly on the components of the entrance system, such as engraved, imprinted or written on the components. The text may be alternatively or additionally be provided as a readable label, such as printed text, handwritten text, numerals, alphanumeric combinations, barcodes, QR codes, signs, or the like. The third estimation outputs may in some examples be given higher weights compared to the first and second estimation outputs 63, 65 as it is capable of directly identifying a correct entrance system type if the embedded text is of a sufficiently high quality.

[0097] The entrance system type 68 has now been identified. The backend computing service 320 may accordingly provide one or more spare part recommendations 70 that match(es) the identified entrance system type 68. This may be done by transmitting a message back to the client device 310, the message involving a set of recommendations 70 for spare parts. This may be presented on display screen of the client device 310, for example as shown and explained with reference to FIG. 4.

[0098] In some examples, the spare part recommendations 70 correspond to the portions of the entrance system that were photographed and provided in the set of imagery 60. This allows for a direct replacement of the photographed portions of the entrance system.

[0099] In some examples, the spare part recommendations 70 correspond to portions of the entrance system that match the identified entrance system type 68, but were not necessarily included in the set of imagery 60 provided to the backend computing service 320. This can be done due to the fact that certain parts of an entrance system are only compatible with other certain parts, or other reasons not necessarily pertaining to compatibility. Therefore, if a particular entrance system type 68 is accurately identified based on imagery 60 of one or more first components of an entrance system, one or more second components different from said one or more first components can be included in the spare part recommendations 70. This enables a servicing / installation of a larger portion of the entrance system than what was originally photographed and supplied to the backend computing service 320 via the set of imagery 60.

[0100] In some examples, recommendation factors can be included in the spare part recommendations 70. These may include one or more of an aging property (e.g. how old a certain component is), a material fatigue property (e.g. whether there is a decline in structural integrity due to repeated stress / load-bearing activity), a corrosion property (e.g. if there is an exposure to moisture / air / corrosive substances), an obsolescence property (e.g. due to advancements in technology rendering certain components obsolete over time), a manufacturing defect property (e.g. inherent defects or weaknesses from the manufacturing process of components), an environmental condition property (e.g. extreme temperatures, humidity, exposure to contaminants), a maintenance neglect property (e.g. inadequate or irregular maintenance practices contributing to deterioration of components), and an operational stress property (e.g. components being subjected to heavy loads, high speeds or intense operating conditions so that wear is accelerated). In view of the above and to further exemplify this, the spare part recommendations 70 may realize that certain components of that entrance system type 68 no longer comply with compliance or safety standards due to regulatory requirement changes and recommend said certain components to be replaced. Alternatively, the spare part recommendations 70 may include time stamp data indicating a time when the entrance system was previously serviced / installed, and accordingly a recommendation to replace certain outdated components. Yet alternatively, the spare part recommendations 70 may include locational data of certain regional or other standards that shall be complied with.

[0101] In some examples, the entrance system type 68 may be provided as input to a digital product store 80 which returns the one or more spare part recommendations 70 matching the provided entrance system type 68 input. The digital product store 80 may be one of a web shop 81, a PIM system 82 or an ERP system 83. The backend computing service 320 may thus automatically provide in the spare part recommendations 70 a quick and efficient way for the authorized user to place orders directly with a digital product store 80 through the client device 310.

[0102] With further reference to FIG. 8, a computer-implemented method 100 for providing spare part recommendations for an entrance system is shown. The method 100 may be implemented from a backend perspective, for example the perspective of the backend computing service 320 as discussed herein. The method 100 comprises receiving 110 a set of imagery including at least two different portions of the entrance system. The method 100 further comprises applying 120 a feature extraction method to the set of imagery, the feature extraction method providing first estimation outputs of an entrance system type estimation. The method 100 further comprises providing 130 the set of imagery as input to a machine learning model trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model providing second estimation outputs of an entrance system type estimation. The method 100 further comprises identifying 140 an entrance system type of the entrance system based on a combination of the first estimation outputs and the second estimation outputs. The method 100 further comprises providing 150 one or more spare part recommendations matching the identified entrance system type.

[0103] With further reference to FIG. 9, a computer-implemented method 200 for obtaining spare part recommendations for an entrance system is shown. The method 200 may be implemented from a frontend perspective, for example the perspective of the client device 310 as discussed herein. The method 200 comprises capturing 210 a set of imagery including at least two different portions of the entrance system. The method 200 comprises further transmitting 220 the set of imagery to a backend computing service. The backend computing service is configured to receive the set of imagery, and apply a feature extraction method to the set of imagery, the feature extraction method providing first estimation outputs of an entrance system type estimation. The backend computing service is further configured to provide the set of imagery as input to a machine learning model trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model providing second estimation outputs of an entrance system type estimation. The backend computing service is further configured to identify an entrance system type of the entrance system based on a combination of the first estimation outputs and the second estimation outputs, and provide one or more spare part recommendations matching the identified entrance system. The method 200 further comprises obtaining 230 the one or more spare part recommendations from the backend computing service.

[0104] FIG. 10 is a schematic illustration of a non-transitory computer-readable storage medium 800 in one exemplary embodiment, capable of storing a computer program product 810. The non-transitory computer-readable storage medium 800 in the disclosed embodiment is a memory stick, such as a Universal Serial Bus (USB) stick; the non-transitory computer-readable storage medium 800 may however be embodied in various other ways instead, as is well known per se to the skilled person. The USB stick 800 comprises a housing 830 having an interface, such as a connector 840, and a memory chip 820. In the disclosed embodiment, the memory chip 820 is a flash memory, i.e. a non-volatile data storage that can be electrically erased and reprogrammed. The memory chip 820 stores the computer program product 810 which is programmed with computer program code (instructions) that when loaded into and executed by a processing device, such as a CPU, will perform either the method 100 or the method 200 as described herein, and optionally any of their respective embodiments or examples as described herein. The USB stick 800 is arranged to be connected to and read by a reading device for loading the instructions into the processing device. The processing device may, for instance, be comprised in the client device 310 or in the backend computing functionality 320.

[0105] It should be noted that a computer-readable medium can also be other mediums such as compact discs, digital video discs, hard drives or other memory technologies commonly used. The computer program code (instructions) can also be downloaded from the computer-readable medium via a wireless interface to be loaded into the processing device.

[0106] The invention has been described above in detail with reference to embodiments thereof. However, as is readily understood by those skilled in the art, other embodiments are equally possible within the scope of the present invention, as defined by the appended claims.

Claims

CLAIMS1. A computer-implemented method (100) for providing spare part recommendations (70) for an entrance system (10), the method (100) comprising: receiving (110) a set of imagery (60) including at least two different portions of the entrance system (10); applying (120) a feature extraction method (62) to the set of imagery (60), the feature extraction method (62) providing first estimation outputs (63) of an entrance system type estimation; providing (130) the set of imagery (60) as input to a machine learning model (64) trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model (64) providing second estimation outputs (65) of an entrance system type estimation; identifying (140) an entrance system type (68) of the entrance system (10) based on a combination of the first estimation outputs (63) and the second estimation outputs (65); and providing (150) one or more spare part recommendations (70) matching the identified entrance system type (68).

2. The method (100) of claim 1, wherein the identifying (140) comprises generating combined estimation outputs (66) as a combination of the first and second estimation outputs (63, 65), and selecting the entrance system type (68) as an entry with the highest estimation confidence from among entries of the combined estimation outputs (66).

3. The method (100) of claim 2, wherein the combined estimation outputs (66) are generated using one or more of: a weighted average of the first and second estimation outputs (63, 65), a majority voting of the first and second estimation outputs (63, 65), a softmax function of the first and second estimation outputs (63, 65), and a Bayesian fusion of the first and second estimation outputs (63, 65).

4. The method (100) of claim 3, further comprising: increasing estimation confidence for entries among the combined estimation outputs (66) that are common in the first and second estimation outputs (63, 65), and decreasing estimation confidence for entries among the combined estimation outputs (66) that are unique in either one of the first and second estimation outputs (63, 65).

5. The method (100) of any preceding claim, wherein the spare part recommendations (70) match portions of the entrance system (10) included in the set of imagery (60).

6. The method (100) of any preceding claim, wherein the spare part recommendations (70) match portions of the entrance system (10) not included in the set of imagery (60).

7. The method (100) of any preceding claim, wherein the spare part recommendations (70) include a recommendation factor being one or more of an aging property, a compliance property, a material fatigue property, a corrosion property, an obsolescence property, a manufacturing defect property, an environmental condition property, a maintenance neglect property, and an operational stress property8. The method (100) of any preceding claim, wherein applying the feature extraction method (62) comprises identifying unique indicia in the set of imagery (60) being based on one or more of measurements, coordinates, textures, shapes, edges and color values.

9. The method (100) of claim 8, wherein the unique indicia is used as input to train the machine learning model (64).

10. The method (100) of any of claims 8-9, wherein the machine learning model (64) provides the second estimation outputs (65) based on a classification of the unique indicia provided to the machine learning model (64).

11. The method (100) of any preceding claim, further comprising applying an image preprocessing method to the set of imagery (60) before providing estimation outputs (63, 65).

12. The method (100) of any preceding claim, further comprising: providing third estimation outputs by applying optical character recognition to the set of imagery (60), wherein in addition to the first and second and estimation outputs (63, 65), the identifying (140) is further based the third estimation outputs.

13. The method (100) of any preceding claim, wherein the set of imagery includes a front panel portion (11-1) of a movable door member (12) of the entrance system (10), a rear panel portion (11-2) of the movable door member (12), and a hinge portion (18-1) of a hinge mechanism (18) of the movable door member (16).

14. The method (100) of any preceding claim, wherein the providing (150) of one or more spare part recommendations (70) comprises sending a request for parts to a digital product store (80), the request involving the entrance system type (66), wherein the digital product store (80) returns one or more spare part recommendations (70) matching the requested entrance system type (66).

15. A computer-implemented method (200) for obtaining spare part recommendations for an entrance system (10), the method (200) comprising: capturing (210) a set of imagery (60) including at least two different portions of the entrance system (10); transmitting (220) the set of imagery (60) to a backend computing service, the backend computing service being configured to:- receive the set of imagery (60); apply a feature extraction method (62) to the set of imagery (60), the feature extraction method (62) providing first estimation outputs (63) of an entrance system type estimation;- provide the set of imagery (60) as input to a machine learning model (64) trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model (64) providing second estimation outputs (65) of an entrance system type estimation;- identify an entrance system type (68) of the entrance system (10) based on a combination of the first estimation outputs (63) and the second estimation outputs (65); and- provide one or more spare part recommendations (70) matching the identified entrance system type; and obtaining (230) the one or more spare part recommendations (70) from the backend computing service.

16. A computerized recommendation system (300) for an entrance system (10), the system (300) comprising a backend computing service (320) configured to: receive a set of imagery (60) including at least two different portions of the entrance system (10); apply a feature extraction method (62) to the set of imagery (60), the feature extraction method (62) providing first estimation outputs (63) of an entrance system type estimation; provide the set of imagery (60) as input to a machine learning model (64) trained on a dataset involving sets of imagery of predetermined entrance systems, the machine learning model (64) providing second estimation outputs (65) of an entrance system type estimation; identify an entrance system type (68) of the entrance system (10) based on a combination of the first estimation outputs (63) and the second estimation outputs (65); and provide one or more spare part recommendations (70) matching the identified entrance system type.

17. A computer program product comprising computer code for performing the method according to claim 1 or 15 when the computer program code is executed by a processing device.

18. A non-transitory computer readable storage medium having stored thereon a computer program comprising computer program code for performing the method according to claim 1 or 15 when the computer program code is executed by a processing device.