Ship production management system

Abstract

The present invention relates to a ship production management system comprising: an imaging unit for acquiring images of a plurality of zones in which blocks constituting a ship are loaded; an image processing unit for matching the images with the plurality of zones; and an inference unit for extracting information of the blocks included in the images, wherein the imaging unit includes a flight vehicle for collecting imaging data while moving along the plurality of zones, the image processing unit generates aligned data by aligning the imaging data and acquires data for the respective zones by tiling the aligned data according to a preset zone, and the inference unit extracts information of the blocks included in the zone-specific data by using a trained inference model.

Description

Ship production management system

[0001] The present invention relates to a ship production management system.

[0002] Ships navigating the oceans loaded with various types of cargo are constructed by assembling multiple blocks due to their sheer size. Specifically, a ship is completed through production stages such as sub-assembly, medium assembly, large assembly, PE, and loading.

[0003] To manufacture ships, spaces capable of carrying out each production stage must be provided. In other words, a shipyard is equipped with areas such as a sub-assembly plant, a large-scale assembly plant, a PE (Pre-assembly) area, and a dock where loading takes place.

[0004] At this stage, components or blocks are loaded in each section of the shipyard, and the blocks are transported according to the manufacturing process. When producing multiple ships simultaneously, the loading and transport of blocks become highly complex; therefore, effectively managing this process has a direct impact on shortening construction time and increasing production efficiency.

[0005] In the case of conventional shipyards, although some automation has been applied to production management, there are still many aspects that rely on manual labor, which limits the improvement of production efficiency.

[0006] Therefore, research and development aimed at incorporating various automation technologies into shipyards to improve work efficiency and safety in the working environment are actively underway, but practical application has not yet been sufficiently achieved.

[0007] The present invention was created to solve the problems of the prior art described above. The objective of the present invention is to provide a ship production management system that can significantly shorten the construction period by improving the efficiency of material management and maximizing production efficiency through the application of automation based on images for the loading, transfer, and installation of blocks within a shipyard.

[0008] Furthermore, the objective of the present invention is to provide a ship production management system that prevents unnecessary cost consumption by minimizing errors that may occur in material management and statistics, by matching design data with actual data regarding blocks, etc., stored in a shipyard and immediately identifying and correcting errors.

[0009] A ship production management system according to one aspect of the present invention comprises: a shooting unit that acquires images for a plurality of zones for loading blocks forming a ship; an image processing unit that matches the images with the plurality of zones; and an inference unit that extracts information of the blocks included in the images, wherein the shooting unit includes an aircraft that collects shooting data while moving along the plurality of zones, the image processing unit aligns the shooting data to generate aligned data and tiles the aligned data according to preset zones to acquire data for a plurality of zones, and the inference unit extracts information of the blocks included in the data for each zone using a learned inference model.

[0010] Specifically, the inference unit can extract information of the block through the inference model based on the matching data and the multiple zone-specific data.

[0011] Specifically, the system further includes a matching unit that matches inference information extracted by the inference unit with GIS information, wherein the matching unit may include: a GIS input unit that retrieves GIS information of a block for the area; a layout modification unit that matches the inference information with the GIS information and updates the GIS information according to whether the matching is successful; and a new generation unit that receives new information for the inference information when there is no GIS information corresponding to the inference information.

[0012] Specifically, the above-mentioned placement modification unit may delete the GIS information, correct the location of the GIS information, or replace the GIS information with the inference information, depending on the matching situation between the inference information and the GIS information.

[0013] Specifically, when the performance of the inference model of the inference unit deteriorates, a learning unit may be further included to retrain and share the inference model.

[0014] Specifically, the inference unit may include: a first inference unit that infers information of the block using a plurality of library models for the area-specific data; a second inference unit that infers information of the block for at least some areas identified based on polygon coordinates; and an inference result output unit that reflects the inference result in the image.

[0015] Specifically, it further includes a stacking management unit that manages the stacking status of blocks for the said area, and the stacking management unit can calculate the occupied area of ​​said blocks based on inference information extracted by the inference unit for a number of land parcels belonging to said area.

[0016] Specifically, the stacking management unit calculates the occupied area based on the number of lot numbers over which the shape of the block corresponding to the inference information overlaps, or can calculate the occupied area based on the bounding box of the block corresponding to the inference information.

[0017] The ship production management system according to the present invention can secure images using drones, etc., regarding various production processes such as loading, transferring, and mounting of blocks in each area within a shipyard, and can implement automation of the production process based on the secured images. Through this, the present invention can effectively manage block inventory and fully secure effects such as shortening construction time and increasing production efficiency.

[0018] In addition, the ship production management system according to the present invention can compare design data related to block stacking at a shipyard with actual stacking data confirmed from an image, and analyze errors based on the comparison results to accurately verify the stacking status. Through this, the present invention can minimize production costs by enabling error-free management of the block stacking situation.

[0019] FIG. 1 is a block diagram of a ship production management system according to a first embodiment of the present invention.

[0020] FIG. 2 is a partial block diagram of a ship production management system according to a first embodiment of the present invention.

[0021] FIG. 3 is a partial block diagram of a ship production management system according to a first embodiment of the present invention.

[0022] FIG. 4 is a conceptual diagram illustrating a shot in a ship production management system according to the first embodiment of the present invention.

[0023] FIG. 5 is a conceptual diagram illustrating image processing in a ship production management system according to the first embodiment of the present invention.

[0024] FIG. 6 is a conceptual diagram illustrating inference in a ship production management system according to the first embodiment of the present invention.

[0025] FIG. 7 is a conceptual diagram illustrating inference in a ship production management system according to the first embodiment of the present invention.

[0026] FIG. 8 is a conceptual diagram illustrating matching in a ship production management system according to the first embodiment of the present invention.

[0027] FIG. 9 is a conceptual diagram illustrating matching in a ship production management system according to the first embodiment of the present invention.

[0028] FIG. 10 is a conceptual diagram illustrating matching in a ship production management system according to the first embodiment of the present invention.

[0029] FIG. 11 is a conceptual diagram illustrating matching in a ship production management system according to the first embodiment of the present invention.

[0030] FIG. 12 is a conceptual diagram illustrating matching related to pre-loading in a ship production management system according to the first embodiment of the present invention.

[0031] FIG. 13 is a conceptual diagram illustrating matching related to pre-loading in a ship production management system according to the first embodiment of the present invention.

[0032] FIG. 14 is a conceptual diagram of a ship production management system according to a first embodiment of the present invention.

[0033] FIG. 15 is a conceptual diagram illustrating a layer generated by a ship production management system according to a first embodiment of the present invention.

[0034] FIG. 16 is a conceptual diagram illustrating the lot number of a ship production management system according to the first embodiment of the present invention.

[0035] FIG. 17 is a conceptual diagram illustrating the numbering of a ship production management system according to the first embodiment of the present invention.

[0036] FIG. 18 is a conceptual diagram illustrating the occupancy management of a ship production management system according to the first embodiment of the present invention.

[0037] FIG. 19 is a conceptual diagram illustrating the current status of occupancy of a ship production management system according to the first embodiment of the present invention.

[0038] FIG. 20 is a conceptual diagram illustrating the current status of occupancy of a ship production management system according to the first embodiment of the present invention.

[0039] FIG. 21 is a conceptual diagram illustrating changes in stacking of a ship production management system according to the first embodiment of the present invention.

[0040] FIG. 22 is a conceptual diagram illustrating changes in stacking of a ship production management system according to the first embodiment of the present invention.

[0041] FIG. 23 is a conceptual diagram illustrating a passageway area in a ship production management system according to the first embodiment of the present invention.

[0042] FIG. 24 is a conceptual diagram illustrating the management of equipment installation in a ship production management system according to a second embodiment of the present invention.

[0043] FIG. 25 is a conceptual diagram illustrating the management of equipment installation in a ship production management system according to a second embodiment of the present invention.

[0044] FIG. 26 is a block diagram of a ship production management system according to a third embodiment of the present invention.

[0045] FIG. 27 is a conceptual diagram illustrating the indoor status in a ship production management system according to the fifth embodiment of the present invention.

[0046] The objects, specific advantages, and novel features of the present invention will become more apparent from the following detailed description and preferred embodiments in conjunction with the accompanying drawings. It should be noted that in assigning reference numerals to the components of each drawing in this specification, identical components are assigned the same number whenever possible, even if they are shown in different drawings. Furthermore, in describing the present invention, detailed descriptions of related prior art are omitted if it is determined that such detailed descriptions would unnecessarily obscure the essence of the invention.

[0047] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0048]

[0049] FIG. 1 is a block diagram of a ship production management system according to a first embodiment of the present invention, and FIG. 2 and FIG. 3 are partial block diagrams of a ship production management system according to a first embodiment of the present invention. FIG. 4 is a conceptual diagram for explaining shooting in a ship production management system according to a first embodiment of the present invention, and FIG. 5 is a conceptual diagram for explaining image processing in a ship production management system according to a first embodiment of the present invention. FIG. 6 and FIG. 7 are conceptual diagrams for explaining inference in a ship production management system according to a first embodiment of the present invention.

[0050] FIGS. 8 to 11 are conceptual diagrams for explaining matching in a ship production management system according to a first embodiment of the present invention, and FIGS. 12 and 13 are conceptual diagrams for explaining matching related to pre-installation in a ship production management system according to a first embodiment of the present invention.

[0051]

[0052] Referring to FIGS. 1 to 13, the ship production management system (1) according to the first embodiment of the present invention is intended to efficiently improve the overall production work carried out in the yard of a shipyard for building ships.

[0053] However, the present invention is not limited to applications to shipyards that build ships, but can be applied to various factories that produce finished products using multiple blocks. In this case, the various factories may have a structure in which the blocks are exposed to the outside and images of the blocks can be obtained through the shooting unit (10) described later. Of course, even if the blocks are provided indoors, the present invention can be applied as long as the shooting unit (10) is provided in each indoor space.

[0054]

[0055] A ship production management system (1) according to the first embodiment of the present invention includes a shooting unit (10), an image processing unit (20), an inference unit (30), a matching unit (40), a stacking management unit (50), and a work management unit (60).

[0056]

[0057] The shooting unit (10) acquires an image. The shooting unit (10) can acquire an image including the blocks with respect to a space where the blocks are loaded. The space that is the target of the image acquired by the shooting unit (10) can be defined as a zone, and the shooting unit (10) can acquire multiple images for multiple zones. In addition, the blocks included in the image acquired by the shooting unit (10) may form a ship.

[0058] The imaging unit (10) may include an aircraft (11) such as a drone, and an unmanned method may be preferred. The imaging unit (10) may acquire continuous or discontinuous images of multiple areas where blocks may be loaded, and the multiple images acquired by the imaging unit (10) may be of a part subject to production management. For example, the imaging unit (10) may acquire images of most areas of a shipyard yard, excluding indoor spaces.

[0059] Of course, as explained above, the shooting unit (10) can acquire images of the indoor space through a separate shooting device. In this case, the shooting unit (10) can acquire images of the outdoor space area using an aircraft (11) and acquire images of the indoor space through a separate shooting device. In this case, the images can be aligned with each other after undergoing appropriate processing, and the shooting unit (10) can be guided to manage the indoor space as if it were an outdoor space.

[0060] The aircraft (11) of the shooting unit (10) can collect shooting data while moving along multiple zones. For reference, it should be noted that in this specification, data and images may be used interchangeably. The aircraft (11) can obtain images of the zones of the shipyard as shooting data while flying according to a pre-set schedule. Referring to FIG. 4, one or more aircraft (11) may be provided, and after starting flight from a landing point, it can move to the zone to be photographed. Subsequently, the aircraft (11) can obtain continuous images of the zone.

[0061] The aircraft (11) can automatically perform mission flights based on the station (12). Additionally, the aircraft (11) can fly at a constant altitude to secure image data in order to ensure consistency, but it is also possible to selectively collect image data for areas requiring detailed analysis. The aircraft (11) can be used during both day and night, and may be equipped with a camera with specifications capable of overcoming variables caused by weather.

[0062] The user can set a route and time for the aircraft (11), and the aircraft (11) can perform orthophotos to obtain shooting data. The aircraft (11) can use a built-in battery, and can automatically return to the station (12) depending on the remaining battery level. In addition, all shooting data of the aircraft (11) can be transmitted to the cloud using the station (12).

[0063] That is, the shooting unit (10) can automate operations by remotely specifying the flight path of the aircraft (11) and implementing autonomous flight and automatic charging. In addition, the aircraft (11) can be operated even in adverse weather conditions such as winter, snow, or rain, and flight control can be performed in real time by reflecting weather conditions. Furthermore, flight stability can be ensured by linking cloud transmission based on LTE communication for the shooting data.

[0064] In addition, since the filming unit (10) is configured to allow remote monitoring of the aircraft (11), the aircraft (11) can be quickly moved to the site without separate preparation work when a specific situation occurs. Through this, the filming unit (10) can achieve unmanned operation.

[0065] By operating the station (12), the shooting unit (10) can efficiently manage the remote control and automatic flight of the aircraft (11), and the user can easily control the flight path and scheduling using a terminal or the like. The station (12) of the shooting unit (10) can manage multiple aircraft (11) in an integrated manner and monitor the real-time status (weather, location, altitude, speed, battery, etc.) of the aircraft (11), and can immediately respond to and record history when variables occur in the aircraft (11). In addition, the station (12) is equipped with an internal cloud to meet industrial security requirements and can implement streaming, sharing, and data management of shooting data.

[0066]

[0067] The image processing unit (20) processes the image. The image processing unit (20) can match the image obtained by the shooting unit (10) to multiple areas subject to management of ship production. The image processing unit (20) obtains area-specific data from the shooting data collected by the shooting unit (10). For example, the image processing unit (20) aligns the shooting data to generate aligned data. At this time, the aligned data may be for all of the multiple areas subject to production management, or it may be for at least some of the areas considering the efficiency of data management. That is, one or more aligned data may be generated for the entire shipyard, and aligned data may be generated separately for each aircraft (11).

[0068] The image processing unit (20) can perform tiling on the matched data. Tiling is a task of dividing the matched data, and the division of the matched data can be performed according to pre-set areas.

[0069] Previously, the captured data collected by the aircraft (11) may be divided according to the aircraft's (11) movement path and speed, and may be excessively overlapped depending on the situation. Therefore, the image processing unit (20) may align all the captured data to generate aligned data, and then divide it again to generate zone-specific data corresponding to each zone. Zone-specific data may refer to images for each zone, and zone-specific data for adjacent zones may overlap at least partially with each other.

[0070] The sector-specific data generated by the image processing unit (20) is an image assigned according to the sector, and the sector is a space within the shipyard where blocks can be located that is arbitrarily partitioned. For example, the sector can be arbitrarily divided and formed to have a size greater than a certain size, and can be defined by considering conditions such as work stages (sub-assembly, large assembly, PE, etc.) and separation by passages.

[0071] The image processing unit (20) can implement at least two types of alignment. For example, the image processing unit (20) can form two versions of alignment data by applying fast alignment and precise alignment to the captured data.

[0072] For each of the two versions of matched data, zone-specific data can be generated. For the two versions of zone-specific data, inference and matching described below can all be performed. For example, zone-specific data based on fast matched data may be used first, and zone-specific data based on precise matched data may be used secondarily as needed for certain zones.

[0073] The area-specific data generated by the image processing unit (20) may contain blocks to be inferred by the inference unit (30), which will be described later, with a size that occupies a specific pixel. Additionally, the area-specific data may contain multiple blocks, and at least most of the blocks may not be stacked vertically. Furthermore, depending on the stacking state of the blocks, only a portion of the blocks may be included within the area-specific data. That is, at least one block may be included across two adjacent area-specific data.

[0074] Below, with reference to FIG. 5, image processing and the like from the shooting unit (10) will be explained again. Referring to FIG. 5, shooting data captured by the aircraft (11) can be collected at one or more stations (12), and the shooting data from the stations (12) can be transmitted to a NAS corresponding to a cloud server or an image server. Subsequently, the shooting data can be aligned through the orthogonal operation of the image processing unit (20), and zone-specific data can be generated as the aligned data is tiled based on zone (sector) information (shapefile format, etc.).

[0075] The matching data generated by the image processing unit (20) can be transmitted to and managed by a GIS server. The GIS server may be linked with a DB server and may be a server that manages GIS information used by the matching unit (40) described later. The matching data transmitted to the GIS server may be managed by the previously described version, and may also be managed by the GIS server in conjunction with GIS information. Additionally, considering the characteristics of the shipyard, data regarding the arrangement of each ship (by vessel) built by blocks may also be managed by the GIS server in conjunction with the matching data.

[0076] Since the data by zone corresponds to an image for block inference, it can be referred to as an inference image. The inference image is transmitted to an inference server, and inference can be performed by the inference unit (30) described later.

[0077] Block inference by the GIS information of the GIS server and the inference server can be used for block matching, which will be described later. This will be described later.

[0078]

[0079] The inference unit (30) can extract information about blocks included in an image. The inference unit (30) can extract information about blocks using a learned inference model, and the inference model may be a library model such as a YOLO model, but the model is not limited to this.

[0080] The inference unit (30) can perform inference based on the region-specific data generated by the image processing unit (20). The inference unit (30) can extract information about blocks included in the region-specific data based on a pre-trained inference model.

[0081] The inference unit (30) can utilize both zone-specific data and alignment data. Additionally, the inference unit (30) can also use the two versions of data described above. That is, the inference unit (30) can use various data, such as shooting data obtained by the aircraft (11), in combination to process blocks included in the image.

[0082] A learning unit (31) may be included for learning the inference model of the inference unit (30). The learning unit (31) can enable accurate inference of blocks from region-specific data through the learning of the inference model. Additionally, the learning unit (31) can re-learn and share the inference model if it is determined that the performance of the inference model of the inference unit (30) has deteriorated or if there is a request from a user.

[0083] The learning unit (31) can perform data labeling to train the inference model. Data labeling involves classifying the blocks included in the data into classes, and the classes may be flat blocks, curved blocks, cabin blocks, sub-assemblies, flat plates, etc. Data labeling can be used importantly to filter out noise that does not correspond to a block so that the inference model can more accurately infer blocks from the data by region. In addition, through such training, the inference model can increase the accuracy of block matching by distinguishing blocks with unique shapes, such as cabin blocks.

[0084] The learning unit (31) can resolve errors (omission, misrecognition, separation, etc.) that occur due to a lack of learning data or the processing of confidence values ​​in advance. For example, the omission of detection may be a simple omission or a case where the plate and the trolley cannot be distinguished, and to this end, the learning unit (31) can perform separate learning on the inference model regarding the plate or trolley having a size similar to a block within the image.

[0085] In addition, there may be cases where new shapes or new classes that were not learned occur. Taking this into consideration, the learning unit (31) can perform re-learning such as labeling and adding classes.

[0086] Re-learning by the learning unit (31) and evaluation of the inference model of the learning unit (31) can be continuously monitored, and the inference model shared by the learning unit (31) can be utilized in the first inference unit (32) and second inference unit (33) described later. The learning unit (31) can be defined as a learning server, the image processing unit (20) described above can be defined as an image server, and the first inference unit (32) described later can be defined as an inference server. In addition, the present invention may also include a DB server for user access, application implementation, etc.

[0087] The inference unit (30) includes a first inference unit (32), a second inference unit (33), an inference result output unit (34), etc. The first inference unit (32) infers block information for zone-specific data using multiple library models. As previously explained, the image processing unit (20), which receives an image from the station (12) of the shooting unit (10), can generate zone-specific data through a segmentation operation after changing the format, direction, coordinates, etc. of the image. At this time, the first inference unit (32) can distinguish blocks included in the zone-specific image by performing inference on the zone-specific image using a YOLO model, etc.

[0088] The first inference unit (32) can learn and apply multiple models to ensure robustness. The inference model of the first inference unit (32) can use multiple library models that consider different environmental conditions, such as normal / snow / foggy. Through this, the first inference unit (32) can infer block information contained in the area-specific data.

[0089] The first inference unit (32) can perform inference of the block using a learning model and then proceed with post-processing. The post-processing of the first inference unit (32) may be processing of overlapping regions. As previously explained, the region-specific data may be divided by the image processing unit (20), and the division at this time is image division for the optimization of the YOLO model. However, depending on the stacking state, the block may span across two or more region-specific data.

[0090] Considering this, the first inference unit (32) selects overlapping areas where the data for each zone overlaps with each other and can define a block class for the overlapping area. That is, the first inference unit (32) defines a class for a block that spans the overlapping area of ​​the data for each zone, and the class may be a flat block, a curved block, a cabin block, etc. Among the classes described above, a flat plate, others, etc., are not inferred as a block by a sufficiently trained YOLO model.

[0091] The definition of a class made by the first inference unit (32) can be determined based on the classes identified during block inference of data for each zone for a block that spans across an overlapping area. For example, if a class inferred based on the data for each zone for a block stacked across three zones appears as one first value and two second values, the class for the block in the overlapping area can be defined as the second value.

[0092] The first inference unit (32) defines the block class in the overlapping area, merges the overlapping area and the non-overlapping area, and identifies the abnormal range. Since the block is a large object, the inference area is wide, so an abnormal range may occur. Therefore, the first inference unit (32) can identify the abnormal range and allow re-inference to be performed.

[0093] The second inference unit (33) can infer information about the block regarding the abnormal range identified by the first inference unit (32). Since the method of inference by the second inference unit (33) may be the same or similar as that described in the first inference unit (32), a detailed description is omitted.

[0094] The secondary inference unit (33) can perform re-inference on the abnormal range to ensure sufficient inference is made for the block. Additionally, considering that the secondary inference unit (33) performs inference on the abnormal range that is part of the area-specific data, it can perform operations such as coordinate system transformation and unification so that the inference results can be merged back into an image.

[0095] The inference unit (30) will be described again below with reference to FIGS. 6 and FIGS. 7. Referring to FIGS. 6, first, an image captured by the capturing unit (10) is processed by the image processing unit (20) via an image server. The image processing unit (20) can change the file format of the image and, as shown in FIGS. 7 (A), divide the image into sizes optimized for the YOLO model to generate zone-specific data.

[0096] Afterward, the first inference unit (32) can perform inference on the data by zone to verify the information of the block. Additionally, the first inference unit (32) can select overlapping areas where the data by zone overlaps with each other, define a block class for the overlapping area as shown in (B) of FIG. 7, and merge the overlapping area and the non-overlapping area.

[0097] Additionally, the primary inference unit (32) identifies an abnormal range. Specifically, the abnormal range may refer to an area where the polygon area exceeds a preset range. For example, in an image where overlapping and non-overlapping areas are merged, the primary inference unit (32) may not infer a block if the polygon area is too small or too large. This refers to blocks that were not included in the inference result in (B) of FIG. 7 but are identified in (C) of FIG. 7. Additionally, the polygon area may correspond to the width in pixel units within the image.

[0098] The first inference unit (32) identifies and removes abnormal ranges, where removal means resolving abnormal ranges by performing re-inference on the abnormal ranges through the second inference unit (33). The first inference unit (32) may apply cropping to abnormal ranges so that the second inference unit (33) can re-infer the abnormal ranges.

[0099] The second inference unit (33) can perform re-inference on the cropped abnormal range. That is, the second inference unit (33) can re-infer information of the block using a library model for the area (abnormal range) identified by the first inference unit (32) based on polygon coordinates.

[0100] Subsequently, the second inference unit (33) transforms the coordinate system of the re-inferenced data. Specifically, the second inference unit (33) can transform the coordinate system of the re-inferenced data to match the inference data from the first inference unit (32). Additionally, the second inference unit (33) can unify the coordinate system of the re-inferenced data with the coordinate system of the image, which can be done through the integration and approximation of labels.

[0101] As the inference regarding the abnormal range is fully processed through the above first inference and second re-inference (abnormal range removal), sufficient inference information regarding the blocks included in the image can be extracted.

[0102] The result of the inference completed in this manner can be output by the inference result output unit (34). The inference result output unit (34) can reflect the inference result in an image, and the image output by the inference result output unit (34) is the same as shown in (D) of FIG. 7, and includes both the result of the first inference and the result of the second inference regarding the abnormal range. The inference information included in the inference result can be matched with GIS information by the matching unit (40) to be described later.

[0103]

[0104] The matching unit (40) matches the inference information extracted by the inference unit (30) with the GIS information. The GIS information is a Geographic Information System and may be GIS information of a block for a zone. That is, the GIS information used by the matching unit (40) is information of a block that is stored or entered in advance, and may be information including the block's class, specifications, storage location, etc.

[0105] Since this information is collected by the user rather than extracted from an image, there may be cases where it does not match the actual site. Therefore, the matching unit (40) can match the inference information with the GIS information and then update / modify the GIS information depending on whether the matching is successful.

[0106] As explained earlier, since GIS information is manually entered by the user, there may be errors compared to inference information extracted from images of the site. Also, failure to match between GIS information and inference information means that there is an error in the GIS information. Therefore, the GIS information can be modified based on the inference information according to the matching result of the matching unit (40).

[0107] This matching unit (40) may include a GIS input unit (41), a block selection unit (42), a matching implementation unit (43), a layout modification unit (44), and a new creation unit (45). The GIS input unit (41) can retrieve GIS information of a block for a zone. The GIS input unit (41) can process the GIS information of a block input by a user, etc., into a form that can be matched with inference information. For example, the GIS input unit (41) can extract GIS information into information suitable for the zone to be matched. That is, the GIS input unit (41) can go beyond inputting GIS information and include a function to process GIS information.

[0108] The block selection unit (42) selects a group of block candidates based on GIS information. Since the GIS information and the inference information may not perfectly match, in order to find a corresponding block in the inference information based on the GIS information corresponding to a specific block, it is necessary to select at least one block that appears to correspond to the GIS information.

[0109] Accordingly, the block selection unit (42) can select a block candidate group expected to be matched based on GIS information. The block candidate group may be extracted from GIS information or inference information, and may be extracted based on whether the lot number assigned to each storage space within the area matches or is similar. For example, the block selection unit (42) can select a block candidate group based on GIS information and inference information, which is located within four ranges up, down, left, right, or diagonally based on the lot number. That is, the block selection unit (42) can select a block candidate group by considering the lot number data included in the area.

[0110] The and / or block selection unit (42) can select candidate blocks to be matched based on the shape of the blocks overlapping in GIS information and inference information. If the overlapping area is greater than a certain ratio to the block, the block may be included in the block candidate group.

[0111] The block selection unit (42) can select a block candidate group by utilizing the IoU value calculation of the matching implementation unit (43) described below. However, the preset range of the IoU value used for selecting the block candidate group may be set wider than the preset range used by the matching implementation unit (43), and for example, the preset range of the IoU value used by the block selection unit (42) may be 0.6 to 1.3, preferably 0.7 to 1.25, etc.

[0112] The matching implementation unit (43) performs matching between GIS information and inference information based on a block candidate group selected by the block selection unit (42). First, the matching implementation unit (43) can align the center points of the GIS information and inference information with respect to the block candidate group. Since there may be cases where the coordinates between the blocks of the GIS information and the blocks of the inference information are misaligned (see FIG. 8), the block selection unit (42) can perform coordinate movement of the blocks with respect to the GIS information.

[0113] Additionally, the matching implementation unit (43) can calculate a matching block among block candidates while changing the direction of the block included in the GIS information. Since the block in the GIS information is a block based on a CAD shape and the block in the inference information is based on a real shape, there may be mutual errors in position and angle (see FIG. 9). Therefore, the matching implementation unit (43) can perform a center point matching operation to match the position and a rotation operation to match the angle.

[0114] Specifically, the matching implementation unit (43) can calculate an IoU value for each angle while performing rotation on the GIS information. The IoU (Intersection over Union) value is a value that can check the degree of overlap between spatial objects, and means the value obtained by dividing the area of ​​the overlapping part between two objects by the total combined area of ​​the two objects. If the IoU value is 1, it means that the two objects are completely aligned, and as the IoU value goes from 1 to 0, it means that the degree of misalignment between the two objects increases.

[0115] The matching implementation unit (43) can calculate an IoU value by adjusting the angle after matching the coordinates of the previously selected block candidate group, and can calculate a matching block from the inference information according to the IoU value. For example, the matching implementation unit (43) can calculate an IoU value for each angle while changing the direction of the GIS information, and then calculate a block among the block candidate group that has an IoU value greater than or equal to a preset value (e.g., 0.8 to 0.85) as a matching block of the GIS information.

[0116] However, since such automatic matching work can be performed when there are only partial errors in the location and angle of the GIS information compared to the inferred information, in reality, there may be cases where matching is not achieved due to block matching or other variables.

[0117] Accordingly, the matching implementation unit (43) can implement matching even when the IoU value is less than the preset value, and the matching implementation unit (43) can perform the calculation of a matching block even when the IoU value in the inference information is less than the preset value and is included within the preset range.

[0118] Specifically, the matching implementation unit (43) performs automatic matching when the IoU value is 0.85 or higher, and can check whether matching is possible through a separate calculation when the IoU value is less than 0.85 and 0.6 or higher. For example, the matching implementation unit (43) can perform a Hausdorff calculation when the IoU value is within a preset range (0.7 to 0.83).

[0119] The term "hausdorff calculation" refers to an indicator that quantitatively measures the difference in shape or position between two objects, and is a value that measures how far apart two objects are from each other. Since the hausdorff distance calculation can use a known method, a more detailed explanation is omitted.

[0120] The matching implementation unit (43) can determine automatic matching for a block when the calculated value of hausdorff is less than or equal to a preset value (e.g., around 20). Through this, the matching implementation unit (43) can implement automatic matching for a block candidate group when the IoU value is greater than or equal to a preset value, and automatic matching when the IoU value is less than or equal to a preset value.

[0121] The layout modification unit (44) can update the GIS information based on whether the inference information and the GIS information are successfully matched. The inference information and the GIS information can be matched by the block selection unit (42) which selects block candidates and the matching implementation unit (43) which matches blocks based on IoU values, and the layout modification unit (44) can update the GIS information based on the inference information when the matching is successful.

[0122] Since the inference information is based on images and is information that actually corresponds to the site, the layout correction unit (44) can update the GIS information by assuming that there is an error in the GIS information input by the user, etc., when there is an error between the inference information and the GIS information. For reference, it should be noted that the layout correction unit (44) can also be interpreted as a configuration including a block selection unit (42) and a matching implementation unit (43).

[0123] The layout correction unit (44) can perform modification of GIS information for GIS information where automatic matching failed. For example, cases where automatic matching is not achieved may include cases where a block included in the GIS information is not present at the site, where the information of the block is significantly different, or where there is a significant difference in the location of the block. In this case, the layout correction unit (44) can modify the GIS information by performing methods such as layout deletion, location correction, and shape replacement, and the modification of the GIS information can be performed by replacing it with inferred information.

[0124] The placement correction unit (44) can delete GIS information when it is included in GIS information but there is no block in the inference information, depending on the matching situation between the inference information and the GIS information. Alternatively, it can correct the location of GIS information when the location of the GIS information and the location of the inference information are different, and can replace the GIS information with the inference information when the shape of the block does not match during the matching process.

[0125] Among such operations such as deletion / correction / replacement, correction / replacement can be performed even if automatic matching is successful but some difference is found between the GIS information and the inferred information.

[0126] The new generation unit (45) can add GIS information if there is no GIS information corresponding to the inference information. This is the case where there is a block at the site but it is not entered as GIS information, and the new generation unit (45) can receive new information regarding the inference information and create information about the block within the GIS information.

[0127] Since the inference information is information extracted from an image, information regarding the specific specifications of the block, the load of the block, etc., may be omitted. Therefore, the new generation unit (45) can notify the user of the omission of GIS information based on the matching result between the inference information and the GIS information, and induce the user to input specific information about the block.

[0128] Matching will be explained again below with reference to FIGS. 10 and FIG. 11. First, referring to FIG. 10, inference information is generated as block inference is performed based on zone-specific data divided by the image processing unit (20), and the inference information and GIS information are processed as a Batch Job as shown in FIG. 5 and FIG. 10. At this time, regarding the GIS information, GIS information including data on orthophotos generated in one or more versions and GIS blocks can be input from CAD_org.

[0129] The matching unit (40) can automatically perform block matching using GIS information and inference information. However, since some errors may remain despite automatic matching and subsequent operations (layout modification, new creation), additional manual work may be performed. Manual work may involve manual adjustment of some information and may be omitted depending on the situation.

[0130] Once automatic block matching and manual block matching are completed, the block matching result (offset result) can be reflected in the GIS information, and the GIS information can be saved and managed again.

[0131] The details regarding the status / occupancy aggregation, etc. based on AI inference in Fig. 10 will be described in detail when explaining the stacking management unit (50) below.

[0132] Referring to FIG. 11, the results of automatic block matching can be transmitted to Legacy, and layout modification or new creation can be performed as needed. Layout modification may include, but is not limited to, layout deletion, position correction, or shape replacement, and various types of modification operations may be performed. When layout modification is performed, the modification details can be reflected in the GIS information.

[0133] On the other hand, in the case of new creation, input of missing GIS information is required, and GIS information entered by users, etc., can be reflected in Legacy. Both the reflection of GIS information through layout modification and the reflection of Legacy through new creation can be used for the management of GIS information.

[0134] Below, the matching of blocks considering prior mounting is described with reference to FIGS. 12 and FIGS. 13.

[0135] After undergoing small assembly, medium assembly, and large assembly, the block can be pre-erected (PE) into an extra-large block before being loaded into the dock. Pre-erecting may involve two individual blocks ((A) in FIG. 12) that were stacked adjacently in a specific space within the zone being interconnected to form a large block ((B) in FIG. 12), so that there may not be a significant difference in the image before and after pre-erecting.

[0136] The matching unit (40) can check whether a block has been pre-loaded during the matching process of inference information and GIS information. That is, the matching unit (40) can calculate the history of at least two blocks being pre-loaded as large blocks in a specific space within the area through the matching of inference information and GIS information.

[0137] The matching unit (40) performs the automatic matching described above, but may consider whether the blocks for which matching failed have been pre-loaded. For example, the matching may fail in cases where the performance record of pre-loading has been recorded in advance but the pre-loading has not yet taken place, or conversely, in cases where the pre-loading has actually taken place or the entry of the performance record is omitted. The matching unit (40) implements automatic aggregation of the pre-loading status by taking this into account. For reference, it should be noted that the descriptions regarding pre-loading may be performed by the matching implementation unit (43) described above, or by a separate pre-loading verification unit (not shown).

[0138] The matching unit (40) can query targets that failed to be matched during the matching process between GIS information and inference information, and query polygons of inference information that failed to be matched. Afterwards, the matching unit (40) checks the coordinates of the inference information or surrounding coordinates, and can identify blocks of GIS information that correspond to at least some of the coordinates of blocks of inference information that failed to be matched.

[0139] Subsequently, the matching unit (40) can check whether there is a prior installation by comparing the IoU value or the polygon area. Specifically, referring to (A) of FIG. 13, if the area of ​​the inference information is smaller than the GIS information, it can be presumed that the prior installation record has been entered in advance. Conversely, as shown in (B) of FIG. 13, if the area of ​​the inference information is larger than the GIS information, it can be presumed that the prior installation record has been omitted.

[0140] When it is confirmed that the prior loading is pre-entered or omitted in this manner, the layout correction unit (44) corrects the GIS information. For example, if the area of ​​the block in the inference information is smaller than the area of ​​the block in the GIS information, the layout correction unit (44) determines that the matching fails because the GIS information is confirmed to be pre-loaded but the inference information is confirmed to be pre-loaded, and corrects the GIS information. Conversely, if the area of ​​the block in the inference information is larger than the area of ​​the block in the GIS information, the layout correction unit (44) determines that the matching fails because the GIS information is confirmed to be pre-loaded but the inference information is confirmed to be pre-loaded, and corrects the GIS information.

[0141] However, in the latter case, since the prior loading information must be entered by the user, a notification may be provided to the user, and if information such as the management number of the prior loaded block is entered by the user, the correction of the GIS information may be performed.

[0142] Below, before explaining the stacking management unit (50), a series of operations such as image processing, inference, and matching will be summarized and explained again based on FIGS. 14 and FIGS. 15.

[0143] FIG. 14 is a conceptual diagram of a ship production management system according to a first embodiment of the present invention, and FIG. 15 is a conceptual diagram for explaining a layer generated by a ship production management system according to a first embodiment of the present invention.

[0144] Referring to FIGS. 14 and 15, in a ship production management system (1) according to the first embodiment of the present invention, area-specific data is generated as image alignment and tiling are performed on images collected through orthophotos by a shooting unit (10). At this time, a layer is generated when the image is aligned, and this layer becomes a drone orthophoto image as layer1 of FIG. 15 and can be stored in an image server or a GIS server, etc.

[0145] Based on zone-specific data formed by tiling, inference information is generated by the inference unit (30) of the inference server. The inference information can be utilized for status management, logistics change management, etc. by the storage management unit (50) described later. Additionally, the inference information is used for automatic / manual blocking by the matching unit (40). Furthermore, as previously explained, retraining, monitoring, and distribution by the learning server can be performed for the inference unit (30) to ensure model performance.

[0146] Once the inference and matching of the blocks are completed, the resulting output is transmitted to the DB server, and the DB server can be operated in conjunction with the legacy system. In addition, the results of the inference and matching are reflected in layers 2 through 5 shown in Fig. 15 and can be superimposed on the drone orthophoto of layer 1.

[0147] Layer 2 is about a lot number grid and can be derived from GIS information or extracted from a DB server. Layer 3 is information about sectors and can be information about storage areas, and like layer 2, can be extracted from a GIS server or a DB server.

[0148] Layer 4 is derived based on the matching results between inference information and GIS information, and displays the location and shape of the block, management number, etc. Since the inference information for generating layer 4 is included in layer 5, layer 4 can be generated by matching the inference results of layer 5 with GIS information.

[0149] Based on the orthophoto of layer 1, the lot number of layer 2, the storage area of ​​layer 3, the block location / shape of layer 4, and the inference result of layer 5 can be composited and displayed. Depending on the user's convenience, these layers may be displayed individually or composited, and the display device may be a monitor, a mobile terminal, etc. For example, if a user inputs a signal to check only the lot number, the composite state of layer 1 and layer 2 may be output to a monitor, etc. Alternatively, at the user's selection, the inference result of the block, which is a composite state of layer 1 and layer 5, may be output to a monitor, etc.

[0150] Through this, the present invention enables efficient production management by allowing the user to easily check the status of the storage area, the inference results of the block, and whether the GIS information and the inference information match on the screen.

[0151]

[0152] The stacking management unit (50) manages the stacking status of blocks in a zone. While the preceding configurations show the stacking status by displaying blocks actually stacked within the shipyard yard in conjunction with CAD information, the stacking management unit (50) goes further and is a configuration that automatically manages the stacking status.

[0153] The stacking management unit (50) can manage the occupation of blocks for multiple land parcels belonging to the area, and this is explained with reference to FIG. 16 and FIG. 17.

[0154]

[0155] FIGS. 16 and 17 are conceptual diagrams for explaining the lot number of a ship production management system according to the first embodiment of the present invention.

[0156] Referring to FIGS. 16 and 17, multiple lot numbers may be assigned to one area, and the stacking management unit (50) can manage lot numbers belonging to the area. That is, the stacking management unit (50) can check and manage whether a lot number is an occupied lot number occupied by a block or an empty lot number not occupied by a block, and can guide the user by indicating the stacking status of the block for each lot number with colors, as shown in FIG. 16.

[0157] Additionally, the stacking management unit (50) can manage the status of use, non-use, or non-use for each lot number. The stacking management unit (50) can assign non-use status to at least some lot numbers. The non-use status can be manually set by the user, or automatically entered by considering GIS information, etc. The stacking management unit (50) can manage the assignment of non-use status for lot numbers by matching it with GIS information.

[0158] As shown in Fig. 17, multiple lot numbers within the area may be assigned in the form of a grid, and as indicated by the box on the left, unused status may be set for a specific lot number. In this case, unused status may be set for a period, and the storage management unit (50) may reflect the unused period set for the lot number in the GIS information.

[0159] The stacking management unit (50) can manage the inference information extracted by the inference unit (30) and the land parcels assigned to unused status by integrating them. The stacking management unit (50) can check from the inference information whether a block is stacked for a land parcel set as unused, and can output the result to the user.

[0160] In addition, if there are parcel numbers assigned as unused, the matching unit (40) can match inference information and GIS information by considering the parcel numbers assigned as unused. That is, as previously explained, parcel number data can be considered when selecting block candidates, and for unused parcel numbers, it can be assumed that there is no stacking of blocks and matching can be performed.

[0161] In this way, the storage management unit (50) can establish a system to manage changes such as non-use / disuse of land parcels as time-series data. When a usage period is selected for an unused land parcel, the storage management unit (50) can check for updates to GIS information and verification of inference information in consideration of the period. In addition, when displaying the status of the occupancy distribution, the storage management unit (50) can make it easy to distinguish and manage the land parcels by outputting the unused land parcels in colors that are easily distinguishable from other land parcels.

[0162] The stacking management unit (50) can manage the occupancy status by zone / lot number by calculating the occupied area of ​​the block. This will be explained with reference to FIGS. 18 to 20.

[0163]

[0164] FIG. 18 is a conceptual diagram for explaining the occupancy management of a ship production management system according to the first embodiment of the present invention, and FIG. 19 and FIG. 20 are conceptual diagrams for explaining the occupancy status of a ship production management system according to the first embodiment of the present invention.

[0165] Referring first to FIG. 18, in the ship production management system (1) according to the first embodiment of the present invention, the stacking management unit (50) can calculate the occupied area of ​​a block based on inference information extracted by the inference unit (30) for a number of lot numbers belonging to a zone.

[0166] Specifically, the storage management unit (50) can calculate the occupied area through the number of lot numbers as shown in (A) of FIG. 18, or calculate the occupied area based on the bounding box as shown in (B) of FIG. 18.

[0167] To further explain (A) of FIG. 18, the stacking management unit (50) checks the lot numbers occupied by the block corresponding to the inference information. At this time, the lot number marked with a diagonal line on the right can be excluded from the occupied lot numbers because the area occupied by the block is less than or equal to a preset value (e.g., 10% or less). Therefore, the stacking management unit (50) can determine that the block is occupied across a total of 4 lot numbers, and the stacking management unit (50) can determine the area occupied by the block by subtracting the sum of the areas of the occupied lot numbers from the area of ​​the zone.

[0168] That is, the stacking management unit (50) can calculate the area of ​​the land parcel based on the number of land parcels over which the shape of the block detected by the inference unit (30) overlaps, and then perform a calculation of the stacking standard occupancy area by excluding it from the total sector area.

[0169] Alternatively, the stacking management unit (50) can determine the occupied area based on a bounding box as shown in (B) of FIG. 18. Specifically, the stacking management unit (50) can define a block-shaped bounding box detected by the inference unit (30), calculate the area of ​​the bounding box, and then perform a bounding box-based occupied area calculation to obtain the block's occupied area by subtracting it from the total sector area.

[0170] The stacking management department (50) can calculate the occupied area of ​​the block in this manner, such as based on the number of lot numbers or based on bounding boxes, but can also use various other methods to calculate the occupied area.

[0171] The stacking management unit (50) can determine the occupancy of blocks in a time-series manner and calculate the stacking status based on this. The stacking management unit (50) can calculate the stacking status of blocks for each zone based on the number of land parcels belonging to the zone and the number of blocks released for the zone. That is, the stacking management unit (50) can calculate the block occupancy status for the zone as time-series data, identify the occurrence of long-term stacking where the release of blocks is delayed, and induce improvements.

[0172] Referring to FIG. 19, in an image in which layer 1 and layer 5 are combined, blocks that have been stored for a long period longer than a preset period may be displayed in a separate color (e.g., orange). For example, the storage management unit (50) can monitor blocks stored for 10 days or more and classify them as long-term stored blocks and display them separately.

[0173] The stacking management unit (50) can apply different standards for long-term stacking days for each zone. That is, the stacking management unit (50) can set different stacking day standards for zone-specific data and calculate the stacking status for each zone.

[0174] In addition, the stacking management unit (50) can calculate a foundry-based turnover rate for the analysis of fluid changes over time, as shown in FIG. 20. The stacking management unit (50) can calculate the value obtained by dividing the number of blocks shipped out over a certain period by the number of lot numbers in the zone as the trend of the turnover rate. At this time, the stacking management unit (50) calculates the trend of the turnover rate for each zone and can provide the change in the trend of the turnover rate for each zone in various ways, such as the graph shown in FIG. 20.

[0175] Referring to FIG. 20, the horizontal axis represents the period (from 1 to 3 days starting from 10.01, 2024), and the vertical axis represents the turnover rate (increasing from 0.0 to 0.2 per period), and the turnover rate is defined as the number of blocks shipped out / number of lot numbers. Since the turnover rate for multiple zones can be displayed for each period on the horizontal axis, the stacking management unit (50) allows the user to easily check the turnover rate for multiple zones.

[0176] The stacking management unit (50) can manage and adjust the lot numbers belonging to the zone, and taking this into account, the stacking management unit (50) can calculate the stacking status by zone based on the number of lot numbers and the number of matching blocks released by the matching unit (40). That is, the stacking management unit (50) can perform a method of excluding the number of lot numbers set as unused from the calculation of the turnover rate, and can also calculate the stacking status based on inference information corresponding to block information at the actual site rather than based on GIS information, thereby providing a more accurate stacking status.

[0177] Below, with reference to FIGS. 21 and FIGS. 22, the details of how the stacking management unit (50) calculates time series data will be explained.

[0178]

[0179] FIGS. 21 and 22 are conceptual diagrams for explaining changes in stacking of a ship production management system according to a first embodiment of the present invention.

[0180] Referring to FIGS. 21 and 22, the stacking management unit (50) of the ship production management system (1) according to the first embodiment of the present invention can calculate changes in the stacking space for a zone as time-series data. Since the block inference and matching described above are based on images obtained at a specific time, they cannot show temporal changes in the stacking status of the blocks.

[0181] To overcome these limitations, the storage management unit (50) can provide changes in the logistics of the block by calculating changes in the storage space as time series data based on inference information extracted by the inference unit (30) for the area.

[0182] Specifically, the stacking management unit (50) can calculate the entry and exit of blocks as time-series data, and can also calculate the start of occupancy and release of occupancy for specific spaces / lots within the area as time-series data. As the stacking management unit (50) calculates such time-series data based on inference information, it is possible to manage the actual stacking status of blocks more accurately.

[0183] The storage management unit (50) automatically tracks changes in storage space (incoming, outgoing, empty space, etc.) and can calculate the turnover rate through tracking logistics changes. In addition, through the processing of time-series data of logistics changes, the storage management unit (50) can implement a simulation of the optimal space arrangement.

[0184] Referring to FIG. 21, the time series data is described as follows: it is confirmed that blocks 1, 3, and 8 in FIG. 21 (A) change into empty spaces when changing to FIG. 21 (B). Additionally, it is confirmed that blocks are stacked in section 7 of FIG. 21 (A) as time passes to FIG. 21 (B). In this way, the stacking management unit (50) can check for changes from empty spaces to occupied spaces or vice versa, and can also check for changes from the occupied space of the first block to the occupied space of the second block.

[0185] The stacking management unit (50) can output information on blocks that have been received and released for a specific space that has been released from occupancy. Specifically, the stacking management unit (50) can output information on blocks released for a specific space and information on blocks to be received in a time series.

[0186] For example, the stacking management unit (50) can provide information on what kind of block was placed in parts such as number 1 in (A) of FIG. 21 in the situation of (B) of FIG. 21. Through this, the user can check in advance whether stacking is possible when stacking new blocks in parts such as numbers 1 and 3 in (B) of FIG. 21, taking into account the area of ​​a specific space. Therefore, the stacking management unit (50) can prevent unnecessary transfer of blocks and implement an efficient transfer flow of blocks.

[0187] In addition, the stacking management unit (50) can also implement output as time-series data for the pre-loading described in the matching unit (40) above. The stacking management unit (50) can output as time-series data the history of at least two blocks being pre-loaded and converted into large blocks within a specific space.

[0188] Therefore, the stacking management unit (50) can improve the management efficiency of the stacking process by allowing the user to accurately verify and manage the pre-stacking of blocks at the actual site without relying solely on GIS information.

[0189] As previously explained, the matching unit (40) can check whether there is a block of GIS information corresponding to at least some of the coordinates of the block of inference information that failed to match, and then check whether there is a pre-loading by comparing the IoU value or area. After checking whether there is a pre-loading, the matching unit (40) can transmit this to the stacking management unit (50). Through this, the stacking management unit (50) can calculate the pre-loading status of the block as time-series data.

[0190] Below, storage management considering a special zone, specifically a passageway zone, is explained with reference to Fig. 23.

[0191]

[0192] FIG. 23 is a conceptual diagram illustrating a passageway area in a ship production management system according to the first embodiment of the present invention.

[0193] Referring to FIG. 23, the shipyard to which the ship production management system (1) according to the first embodiment of the present invention is applied has zones used for various purposes. For example, the zones may include indoor spaces not visible from the outside, docks, buildings, etc., a stacking zone where the main purpose is stacking blocks, and a passageway zone (dotted box in FIG. 23) for transporting the stacked blocks.

[0194] The stacking management unit (50) can check the stacking of blocks for the passageway area allocated within the area. The passageway area is an area where the stacking of blocks is not proposed, and it is desirable for the passageway area to be empty for the smooth logistics transport of blocks. Accordingly, the stacking management unit (50) recognizes the passageway area and can notify the stacking of blocks for the passageway area based on the inference information extracted by the inference unit (30).

[0195] However, considering the situation where a block moves along a passageway area, the block is placed in the passageway area included in the image captured in that situation, and according to the inference information obtained based on this image, the block may be recognized as being stacked in the passageway area. Therefore, to resolve this misunderstanding, the stacking management unit (50) can verify the stacking of blocks in the passageway area using time-series data.

[0196] Based on the images captured by the shooting unit (10) over time, the stacking management unit (50) confirms that blocks are not stacked in the passageway area if it is confirmed that the position of the blocks in the passageway area has moved. On the other hand, if there is no change in the position of the blocks in the passageway area, the stacking management unit (50) can conclude that transfer is necessary as blocks are stacked in the passageway area.

[0197] The stacking management unit (50) can suggest a method to improve the situation when it is confirmed that blocks are stacked in the passageway area. For example, the stacking management unit (50) can calculate the occupable space within the area where blocks are not stacked. The occupable space can be calculated based on the number or area of ​​land parcels where blocks are not stacked, excluding unused or unusable land parcels among the land parcels included in the area.

[0198] The stacking management unit (50) can calculate the possibility of improving stacking in a passageway area by comparing blocks stacked in the passageway area with available spaces for occupancy. For example, when blocks are stacked in a passageway area, the stacking management unit (50) can select a candidate space for occupancy adjacent to the passageway area where the blocks are stacked, and then identify and guide the space among the candidate spaces for occupancy where the blocks in the passageway area can move. At this time, the stacking management unit (50) can induce the movement of the blocks so that the passageway area can ensure passageway function by utilizing both GIS information and inference information.

[0199] Additionally, the stacking management unit (50) can calculate the movement path of a block based on inference information extracted by the inference unit (30) for multiple zones. The stacking management unit (50) can estimate the movement path of a block based on time series data, and can verify the consistency between the movement path and the passage zone. Depending on the consistency result, if an update or modification of the passage zone is required, the stacking management unit (50) can transmit this to the GIS server.

[0200] Since the classification of passageway zones may vary depending on the stacking status of blocks, the stacking management unit (50) can confirm from inference information that the passageway zones change due to on-site conditions and reflect this in the GIS information. In addition, the passageway zones can be modified at any time due to user input, etc.

[0201] The stacking management department (50) can adjust the number of lot numbers included in the zone by taking into account the passageway zone and calculate the stacking status for each zone. This can be done in the same way as excluding unused lot numbers from the calculation of the turnover rate, so a detailed explanation is omitted.

[0202] The stacking management unit (50) can calculate the movement history of blocks passing through the passageway area as time-series data based on the inference information extracted by the inference unit (30). Additionally, the stacking management unit (50) can determine the possibility of stacking blocks in the passageway area in preparation for situations such as when there is a shortage of stacking space for blocks or when additional stacking space is needed temporarily.

[0203] The stacking management unit (50) can determine whether the passageway area can be occupied based on the movement history of blocks for the passageway area, such as when the movement history of blocks for the passageway area is not required or when an alternative passageway area is identified. If the stacking management unit (50) can stack blocks for the passageway area, it can output a result of converting the passageway area into a stacking area.

[0204] That is, the stacking management unit (50) manages a series of logistics changes, such as stacking blocks, transporting them in a passageway area, and loading them, as time-series data, notifies the stacking of blocks in the passageway area, and can make a judgment on whether blocks are being transported along the set passageway area, or whether it is desirable to convert the passageway area into a stacking area. Through this, the stacking management unit (50) can maximize production efficiency by inducing improvements regarding the stacking of blocks, etc.

[0205]

[0206] FIGS. 24 and 25 are conceptual diagrams for explaining the management of equipment installation in a ship production management system according to a second embodiment of the present invention.

[0207] The following description will focus on the differences between this embodiment and the preceding embodiment. Any parts omitted from the description will be replaced by the preceding content. It should be noted that this applies equally to other embodiments described below.

[0208] Referring to FIGS. 24 and 25, a ship production management system (1) according to a second embodiment of the present invention includes the previously described shooting unit (10), image processing unit (20), inference unit (30), and matching unit (40).

[0209] However, the imaging unit (10) of the present embodiment can acquire images of multiple areas where large equipment is installed on the ship and areas where large equipment is stacked, in addition to photographing the area where blocks are stacked. The present embodiment can automatically manage the external work process of large equipment being loaded onto the ship by going further than block inference and matching and performing inference and matching of large equipment.

[0210] The images captured by the camera unit (10) include areas where large equipment is stored and areas where large equipment is loaded onto a ship, and in the case of the former, it may be an outdoor area for storing large equipment. However, large equipment may be stored indoors, in which case the images may be transmitted to the camera unit (10) through a separate camera provided indoors as described above.

[0211] The area where large equipment is loaded onto the ship may be a dock. The ship is formed by loading a number of blocks, and large equipment is generally loaded after the loading of blocks is completed to some extent. Therefore, the camera unit (10) can obtain an image of the dock.

[0212] The large equipment mentioned in this embodiment is equipment that generates energy by consuming fuel, and may include an engine, a generator, or a boiler. Of course, the large equipment may include all devices having a size equivalent thereto. Furthermore, as long as the image resolution is guaranteed, the large equipment may include all devices of a size that can be inferred by the inference unit (30), such as electrical devices such as transformers, mechanical devices, equipment, etc.

[0213] The image processing unit (20) of the present embodiment matches multiple zones to an image including large equipment. The image processing unit (20) can perform operations such as format conversion and tiling on the image as described above, and further details are provided in the preceding description.

[0214] The inference unit (30) extracts information about large equipment included in the image. The inference unit (30) can extract information about large equipment included in the area-specific data using a learned inference model. Additionally, the inference unit (30) can classify the large equipment being inferred into a class. In the block inference mentioned earlier, the class was a flat block, a curved block, etc., but in this embodiment, the class used for inferring large equipment may refer to the type of equipment (engine, boiler, generator, etc.). However, since the class in this embodiment may have fewer types compared to the block, relatively high-performance inference is possible. The first and second inferences of the inference unit (30) are replaced by the content above.

[0215] In addition, considering that the shape of large equipment can be identified relatively easily, the inference unit (30) can sufficiently add classes of large equipment to be inferred through retraining. For example, the inference unit (30) can expand the detection targets by training the model with various objects such as switchboards, main units, etc.

[0216] In particular, this embodiment can efficiently improve the processing process for field work performance based on images by additionally detecting targets or areas that serve as nodes of the main process.

[0217] The matching unit (40) can match the inference information extracted by the inference unit (30) with the installation information of the large equipment. The inference information may include the location, area, class, etc. of the large equipment displayed in the image as a layer, and the matching unit (40) can verify the specific information of the large equipment displayed in the image by matching the installation information, including specific specifications and installation schedule for the large equipment, with the inference information.

[0218] The matching unit (40) can check the specifications and installation schedule of large equipment scheduled to be installed on each vessel. Additionally, the matching unit (40) can match inference information and installation information based on the image acquisition date to calculate the installation history of large equipment as time-series data.

[0219] That is, the matching unit (40) can automatically check whether the large equipment is properly following the production process at the site by matching the inference information of the large equipment extracted from images of multiple viewpoints with the installation information of the large equipment. Through this, the matching unit (40) can calculate the installation details of the large equipment and the progress of the process for the ship.

[0220] Additionally, the matching unit (40) can visualize the loading schedule of large equipment for each vessel based on the vessel that can be estimated from the image. Specifically, the matching unit (40) converts the coordinate system of the inference information into pixel coordinates of the image and then converts it into a geographical coordinate system by considering the GPS coordinates of the image. Afterward, the matching unit (40) can compare the geographical coordinates with the coordinate information where a specific vessel is located within the image, and if large equipment is located on a specific vessel, it can assign the information of the specific vessel (such as the vessel's ID) to the large equipment.

[0221] After performing these operations, the matching unit (40) can verify the specifications of the large equipment itself for the large equipment identified from the inference information, and can also assign ship information. Since large equipment can have (almost) the same shape for various ships, multiple large equipment with the same shape may exist within the image. Therefore, despite inference and matching, large equipment may not be completely distinguishable individually.

[0222] However, for a vessel equipped with large equipment, the vessel information can be easily verified based on its location, etc., so the matching unit (40) can match a vessel ID by considering the location of the large equipment assigned to the class.

[0223] When the class and vessel ID, etc. of the large equipment are confirmed through the inference and matching process, the installation history of the large equipment for a specific vessel can be output so as to be superimposed within the image described as layer1, etc. At this time, the superposition of layers can be performed by the image processing unit (20).

[0224] The production process of large equipment will be explained again through the drawings below. First, referring to Fig. 24, after painting of the block is completed, outfitting work is performed, and in the pre-assembly of the block, large equipment such as a boiler may be installed first.

[0225] In addition, generators, engines, etc., may be installed after the preliminary installation, and the installation may be completed thereafter. Of course, depending on their size, the propulsion engine may be installed after the installation of the stern section of the block is completed within the dock.

[0226] Referring to FIG. 25, FIG. 25 shows an image of a dock. Based on inference and matching, the main engine (M / E), generator, etc., can be identified from the image. At this time, layer 5, which corresponds to the inference information, can be composited onto layer 1, which is an orthophoto, and installation information (class, specifications, production process, vessel ID, etc.) confirmed through matching can also be composited into the image and output as in FIG. 25.

[0227] Through this, the present embodiment allows for easy verification of mounting details for key components and progress nodes of external work processes, and enables timely response to logistics (securing yield), thereby optimizing decision-making in the production process. Furthermore, since the present embodiment enables users to accurately verify and manage material delivery dates and process progress linked to material procurement, costs incurred in production and management can be significantly reduced.

[0228]

[0229] FIG. 26 is a block diagram of a ship production management system according to a third embodiment of the present invention.

[0230] Referring to FIG. 26, the ship production management system (1) according to the third embodiment of the present invention may further include a work management unit (60) compared to the previous embodiment.

[0231] In this embodiment, the imaging unit (10) can acquire an image of the stacking area of ​​the block. Additionally, the imaging unit (10) can acquire an image of a plurality of areas where painting is performed on the ship or block.

[0232] The shooting unit (10) can acquire images in conjunction with the painting process, and the work management unit (60) can manage the painting status. This will be explained in detail below.

[0233] The image processing unit (20), the inference unit (30), and the matching unit (40) can perform tiling, block inference, and GIS information matching as described in the first embodiment above, so a detailed description is omitted. However, in this embodiment, the above components can be used to manage the painting process. Therefore, block inference and matching can be utilized in conjunction with the painting status of the blocks included in the image.

[0234] The work management unit (60) manages the painting status of a ship or block for a zone. The work management unit (60) can calculate the painting status of a block included in an image (layer1) based on the matching of inference information and GIS information.

[0235] In order to efficiently perform block matching and block coating status management, the matching of inference information and GIS information can be performed based on rapid alignment, and the calculation of the block coating status can be performed based on precise alignment.

[0236] The work management unit (60) can manage the painting process for the block by checking the painting status of the block based on time series data. That is, if it is confirmed by time series data that the block has changed from a first color to a second color (or is in the process of changing), the work management unit (60) can estimate the progress of the painting process of the block based on this.

[0237] For example, the work management unit (60) can extract changes in the coating state of the block based on multiple images according to a time series and calculate the coating interval. The work management unit (60) can verify the actual coating interval for the block and check whether the interval is appropriately performed based on information regarding the process. Alternatively, the work management unit (60) can verify the timing of the coating of the block and then provide the interval to the user.

[0238] In addition, the work management unit (60) can calculate the painting status for a specific vessel or a block related to a specific vessel within the image. That is, the work management unit (60) can output the painting status of a specific vessel so that the painting status can be managed by vessel.

[0239] The work management unit (60) can calculate the painting status of multiple blocks associated with a specific vessel within the image. As previously explained, since the corresponding vessel can be identified through inference and matching, the work management unit (60) can manage the painting status of each block while also calculating the painting status (painting volume) for each vessel based on the vessel information corresponding to the block. In other words, the work management unit (60) can manage the painting status of each block microscopically and manage the painting status for each vessel macroscopically.

[0240] At this time, the management of the painting status of each vessel, performed by the work management department (60), can be accomplished by checking the amount of painting for each block corresponding to a specific vessel and calculating the amount of painting required for that vessel.

[0241] In addition, the work management unit (60) can output the amount of paint to be painted by zone, in addition to outputting the amount of paint to be painted by ship. That is, the work management unit (60) can calculate the painting status of multiple blocks stacked in a specific zone within the image and output the amount of paint to be painted in a specific zone.

[0242] The amount of painting may vary significantly depending on the specific area. For example, in the case of the pre-loading area, most of the painting may be completed, and in the case of the large assembly area, the amount of painting may be large. Taking this into consideration, the work management department (60) can determine the appropriateness of the painting process by utilizing the expected amount of painting for each specific area and comparing it with the amount of painting for the specific area confirmed through inference and matching.

[0243] As described in the first embodiment above, the matching unit (40) can calculate the history of the block being pre-loaded through the matching of inference information and GIS information, and the work management unit (60) can manage the painting status by considering the information of the pre-loading. For example, the work management unit (60) can calculate whether the joint of the large block is painted from the painting status of the block included in the image.

[0244] When blocks are connected to each other by means of pre-loading or loading within a dock, painting of the joints must be performed. The progress of this painting process can be automatically checked by the work management unit (60) based on images collected by the shooting unit (10).

[0245] As previously explained, based on the inference information, changes in the storage space can be calculated as time-series data by the storage management unit (50). At this time, the storage management unit (50) can manage the painting status confirmed by the work management unit (60) in an integrated manner.

[0246] For example, the storage management unit (50) can manage the receipt and shipment of blocks to the painting factory included in the area, and the work management unit (60) can calculate the painting changes of the blocks received and shipped to the painting factory as time series data.

[0247] The paint factory is set up indoors due to environmental regulations, and the storage management unit (50) can manage the logistics transfer of blocks to the paint factory, and the work management unit (60) can check the work before and after at the paint factory. Based on this, the logistics load and painting work load for the paint factory can be calculated, so the productivity indicators of the paint factory can be automatically output and managed.

[0248]

[0249] Hereinafter, with reference again to FIG. 26, a ship production management system (1) according to the fourth embodiment of the present invention will be described.

[0250] Referring again to FIG. 26, the ship production management system (1) according to the fourth embodiment of the present invention includes a shooting unit (10), an image processing unit (20), an inference unit (30), a work management unit (60), etc., and can manage the sea surface within or around the shipyard.

[0251] The imaging unit (10) of the present embodiment can acquire images of the sea surface for multiple areas where the ship is loaded, launched, and operated. While the imaging unit (10) of the previous embodiment mainly acquired images of block stacking areas or large equipment loading areas, the imaging unit (10) of the present embodiment can mainly acquire images of areas including the sea surface in order to estimate and address marine pollution.

[0252] Specifically, the camera unit (10) can acquire images of a dock where the sea surface flows in, a quay where work such as loading equipment is performed after a ship is launched, and a sea area within a certain distance that may be affected by ship production (especially launching) located around a shipyard.

[0253] The image processing unit (20) performs the task of matching and tiling an image with multiple zones. In particular, the image processing unit (20) of this embodiment can mainly distinguish and tile sea surface zones with respect to the image. However, since specific image conversion or tiling operations have been described in the previous embodiment, detailed information is omitted.

[0254] The inference unit (30) extracts information about the sea surface included in the image. The inference unit (30) can extract information about the sea surface through an inference model based on multiple zone-specific data generated by tiling the image processing unit (20).

[0255] The inference unit (30) can infer the pollution status of the sea surface using a learned inference model. The learning unit (31) included in the inference unit (30) can implement the learning of the inference model based on data of changes in the sea surface during launching or quay work in which the production of the ship affects the sea surface, and the sufficiently learned inference model can be monitored and distributed.

[0256] The inference unit (30) infers the pollution state of the sea surface using the first method, but when launching or quay work is performed where the production of the ship affects the sea surface, it can infer the pollution state of the sea surface using the second method, which has improved detection capabilities compared to the first method.

[0257] For example, the inference unit (30) can infer the pollution state of the sea surface based on an image generated by rapid alignment, and when the possibility of sea surface pollution increases, it can infer the pollution state of the sea surface using an image generated by precise alignment. Alternatively, the inference unit (30) can infer the pollution of the sea surface using a first model, and when a situation such as launching occurs, it can infer the pollution of the sea surface using a second model different from the first model. In this case, the second model may have higher inference accuracy compared to the first model, but the inference speed may be lower or the inference cost may be higher.

[0258] That is, the inference unit (30) can efficiently handle situations where marine pollution is a concern by using a second method in which the resolution of the image is different as the inference model used is different from the first method or as the tiling is different from the first method.

[0259] The work management unit (60) can manage the condition of the sea surface in the area. The term work management unit (60) may refer to a unit that checks for sea surface pollution and manages disaster prevention work based on this. Of course, alternatively, the configuration that manages the condition of the sea surface in this embodiment may be referred to as a sea surface management unit, etc.

[0260] The work management unit (60) can estimate the pollution status of the sea surface based on the production schedule of the ship and the inference information of the inference unit (30). At this time, the work management unit (60) can calculate the pollution status of the sea surface as time-series data in conjunction with the production schedule of the ship.

[0261] For example, when sea surface pollution is inferred by considering launching or quay work, the work management department (60) can estimate the degree of sea surface pollution, work history, and the correlation between the pollution location and the work space, and manage them in an integrated manner. Therefore, the work management department (60) can trace back the cause of the pollution occurring on the sea surface.

[0262] Additionally, the work management unit (60) can induce work to treat estimated pollution on the sea surface. For example, the work management unit (60) can communicate with a disaster prevention vessel capable of moving on the sea surface and treating pollution, and the work management unit (60) can transmit a treatment signal to the disaster prevention vessel capable of treating the pollution when pollution is estimated. Therefore, this embodiment enables the disaster prevention vessel to quickly recover the pollutants, thereby enabling rapid suppression of environmental pollution.

[0263] As such, this embodiment photographs the sea surface using a filming unit (10), such as a drone, and checks for sea surface pollution in conjunction with internal shipyard processes. At this time, this embodiment may consider the status of the launching process, where pollutants flow into the sea surface, as a control variable, and also transmits a signal to a pollution control vessel to quickly process the recovery of pollutants through the pollution control vessel. Therefore, this embodiment has the effect of inspecting and preventing marine pollution around the shipyard.

[0264]

[0265] FIG. 27 is a conceptual diagram illustrating the indoor status in a ship production management system according to the fifth embodiment of the present invention.

[0266] Referring to FIG. 27, the ship production management system (1) according to the fifth embodiment of the present invention can implement production management for indoors, going further than the first embodiment which performs block inference and matching based on an outdoor image confirmed through a shooting unit (10).

[0267] In this embodiment, the indoor area among the areas where blocks are stacked can be separately indicated based on an image from the shooting unit (10) or a previously stored image. That is, as shown on the left side of FIG. 27, this embodiment can highlight the indoor area, but an image of the actual site is not necessarily required to provide information on the stacking status of the indoor area. Therefore, instead of using an image from the shooting unit (10), a pre-stored image of the shipyard yard may be used.

[0268] Even so, the present embodiment may include a shooting unit (10). In this case, the shooting unit (10) is provided indoors and can collect images of the current status of blocks stacked indoors. In addition, the present embodiment can implement appropriate image processing, inference, and matching with GIS information for the corresponding images, and can output layers such as indoor images and inference information by merging them. This is similar to the content described in the preceding first embodiment.

[0269] Alternatively, regarding the indoor area, even if an image is acquired using a camera unit (10) installed at the highest point inside the building, it may be difficult to obtain an image that looks down at the indoor area at a glance. This is because the indoor area is equipped with multiple gantry cranes, etc., which can significantly increase the difficulty of inferring the blocks. In addition, since small-scale or medium-scale assembly is performed in the indoor area, the shape of the blocks is not clear, making inference or matching difficult.

[0270] Considering this situation, the present embodiment can verify only GIS information for indoor areas and output it to the user. That is, as shown on the right side of FIG. 27, for indoor areas, the shape of a block drawn arbitrarily and the corresponding information may be displayed, and a display based on an orthophoto may not be made.

[0271] Of course, this embodiment does not completely exclude shooting and inference. Even in an indoor area, tasks such as securing an image through the shooting unit (10), converting the image through the image processing unit (20), generating inference information by the inference unit (30), and block matching by the matching unit (40) can be applied.

[0272] However, as explained above, even if the blocks of the indoor area are displayed only as GIS information, the shooting unit (10) can be used. In this case, the shooting unit (10) can obtain an image to implement matching between the information of the blocks placed at the site and the GIS information without needing to obtain an image to show a flat photograph. That is, this embodiment can verify cases where correction, deletion, or addition is required regarding the GIS information by going through all the series of processes described in the first embodiment from the shooting unit (10) to the matching, and can provide additional guidance only on such errors in the GIS information.

[0273] Of course, it should be noted that even if inference information about the block is not secured, this embodiment can implement management that guides the flow of logistics changes or occupancy status by utilizing only GIS information.

[0274]

[0275] In addition to the embodiments described above, the present invention encompasses all embodiments resulting from a combination of the above embodiments and known technology.

[0276] Although the present invention has been described in detail through specific embodiments, this is for the purpose of specifically explaining the invention, and the invention is not limited thereto. It will be apparent that modifications or improvements can be made by those skilled in the art within the technical scope of the invention.

[0277] All simple variations or modifications of the present invention fall within the scope of the present invention, and the specific scope of protection of the present invention will be clarified by the appended claims.

Claims

1. A shooting unit that acquires images of multiple zones for loading blocks forming a ship; An image processing unit that matches the above image with the above plurality of zones; and It includes an inference unit that extracts information of the block included in the image above, and The above-mentioned imaging unit is, It includes an aircraft that collects shooting data while moving along the aforementioned multiple zones, The above image processing unit is, The above-mentioned shooting data is aligned to generate aligned data, and the above-mentioned aligned data is tiled according to preset zones to obtain multiple zone-specific data. The above inference unit, A ship production management system that extracts information on blocks included in the above-mentioned zone-specific data using a learned inference model.

2. In claim 1, the inference unit, A ship production management system that extracts information of the block through the inference model based on the above matching data and the above multiple zone-specific data.

3. In Paragraph 1, It further includes a matching unit that matches inference information extracted by the above inference unit with GIS information, and The above matching unit is, A GIS input unit that retrieves GIS information of a block for the above-mentioned area; A batch modification unit that matches the above inference information and GIS information and updates the GIS information depending on whether the matching is successful; and A ship production management system comprising a new generation unit that receives new information regarding the inference information when there is no GIS information corresponding to the inference information.

4. In Clause 3, the above-mentioned placement modification unit, A ship production management system that deletes the GIS information, corrects the location of the GIS information, or replaces the GIS information with the inference information, depending on the matching situation between the inference information and the GIS information.

5. In Paragraph 1, A ship production management system further comprising a learning unit that retrains and shares the inference model when the performance of the inference model of the inference unit deteriorates.

6. In claim 1, the inference unit, A primary inference unit that infers information of the block using a plurality of library models for the above-mentioned zone-specific data; A secondary inference unit that infers information of the block for at least some regions identified based on polygon coordinates; and A ship production management system comprising an inference result output unit that reflects the inference result in the above image.

7. In Paragraph 1, It further includes a stacking management unit that manages the stacking status of blocks for the above-mentioned area, The above storage management department is, A ship production management system that calculates the occupied area of ​​the block based on inference information extracted by the inference unit for a number of lot numbers belonging to the above area.

8. In Clause 7, the storage management department above, A ship production management system that calculates the occupied area based on the number of land parcels over which the shape of the block corresponding to the above inference information overlaps, or calculates the occupied area based on the bounding box of the block corresponding to the above inference information.

9. A shooting unit that acquires images of multiple zones for loading blocks forming a ship; An inference unit for extracting information of the block included in the above image; and It further includes a matching unit that matches inference information extracted by the above inference unit with GIS information, and The above matching unit is, A block selection unit that selects a block candidate group based on the above GIS information; and A ship production management system comprising a matching implementation unit that aligns the center point of the GIS information and the inference information, calculates an IoU value while changing the direction of the GIS information, and produces a matching block from the inference information according to the IoU value.

10. In Paragraph 9, The above block selection unit is, Based on the above GIS information, blocks located within the up, down, left, right, or diagonal ranges of the above inference information are selected as the block candidate group, and The above matching implementation unit is, A ship production management system that calculates a block among the above block candidate group whose IoU value is greater than or equal to a preset value as a matching block of the above GIS information, and calculates a matching block among the above block candidate group using Hausdorff calculation when the IoU value in the above inference information is less than the preset value and is included within a preset range.

11. In Paragraph 9, The image processing unit further includes the image obtained by the above-mentioned capturing unit, which tiles the image according to preset zones to obtain multiple zone-specific data. The above inference unit, Based on the above multiple zone-specific data, information on the above block is extracted, and The above block selection unit is, A ship production management system that selects the block candidate group by considering the lot number data included in the above area.

12. A shooting unit that acquires images of multiple zones for loading blocks forming a ship; An image processing unit that matches the above image with the above plurality of zones; An inference unit for extracting information of the block included in the above image; and It includes a stacking management unit that manages the stacking status of blocks for the above-mentioned area, The above storage management department is, A ship production management system that calculates changes in storage space as time-series data based on inference information extracted by the inference unit for the above area.

13. In Paragraph 12, The above storage management department is, A ship production management system that calculates the receiving and shipping of the above-mentioned blocks and the start and release of occupancy of specific spaces within the above-mentioned zone as time-series data, and calculates the trend of the turnover rate for each above-mentioned zone by dividing the number of the above-mentioned blocks shipped out by the number of lot numbers belonging to the above-mentioned zone.

14. In Paragraph 13, The above storage management department is, A ship production management system that outputs information on blocks received and released for the specific space that has been released from occupancy, outputs information on blocks released for the specific space and information on blocks to be received according to a time series, and outputs a history of at least two blocks being pre-loaded as large blocks within the specific space as time series data.

15. In Paragraph 14, It further includes a matching unit that matches inference information extracted by the above inference unit with GIS information, and The above matching unit is, A ship production management system that identifies a block of GIS information corresponding to at least some of the coordinates of a block of inference information where matching failed, compares an IoU value or area, and transmits whether to pre-load to the stacking management unit.

Images