Building cost estimation system and building cost estimation method

Abstract

To provide a building cost estimation system and method that can accurately recognize building elements and significantly reduce the effort required for calculating the necessary building materials. [Solution] A building cost estimation system 100 for estimating the amount of building materials needed from building drawing data D, comprising: dimension acquisition means 1 for obtaining dimension data D1 included in the drawing data D; element recognition means 2 for recognizing building elements E to be estimated from the drawing data D; noise removal means 3 for removing noise N within the surrounding region A1 surrounded by the building elements E recognized by the element recognition means 2; and cost estimation means 4 for estimating the amount of building materials needed from the noise removal region A2 obtained by the noise removal means 3 and the dimension data D1.

Description

Technical Field

[0001] The present invention relates to a building estimation system and a building estimation method for calculating the quantity and cost of building materials required for a building from architectural drawings.

Background Art

[0002] Buildings such as houses are designed using a building CAD (Computer Aided Design) system, and various drawings are created at the design stage. These drawings include a foundation plan showing the shape and arrangement of the foundation, a floor plan showing the shape and size of rooms, etc. These drawings are used for the occupant to confirm the floor plan and for the construction contractor to use in construction. In addition, they are also used to calculate the quantity, size, cost, etc. required for construction of building materials such as foundation concrete and walls. Thus, the work of calculating the quantity and cost of required building materials is generally called "estimation".

[0003] When performing estimation using drawings such as a floor plan and a foundation plan, the procedure is to check the drawings, pick out the quantity of each building element, list them by specification, and then multiply by the unit price corresponding to each building element and compile them into documents. Currently, these tasks are mainly carried out manually.

[0004] The main reason why manual work is required is that a common system has not been introduced among various business operators involved in design to construction, such as design engineers and construction contractors. Therefore, drawing data is often exchanged in paper media or electronic data in PDF (Portable Document Format) format. Although this data contains information such as shape and dimensions, it remains visual information and is not data in a format that can be directly used for estimation processing. As a result, the estimation work may require a great deal of effort and there may be estimation errors.

[0005] To address these problems, systems have been developed that can automatically estimate the necessary building materials from blueprints. For example, the technology described in Patent Document 1 involves transmitting electronic data of a floor plan to a server device, which then detects the size and type of space, such as rooms and corridors, and automatically identifies and places the necessary equipment and building materials using the information about that space. The technology described in Patent Document 1 states that spatial information is detected using a program that includes a trained vertex output model that has been machine-learned to output the positions of vertices in a top view of space. It also states that if there is an error in the detection result, the accuracy of the detection can be improved by inputting training data for correction.

[0006] On the other hand, technologies for accurately detecting spaces (such as rooms) have also been developed. For example, Patent Document 2 discloses technology for an image analysis device that appropriately recognizes walls from images of floor plans. The technology described in Patent Document 2 involves having the user specify a corner on the outer edge of a building, forming a grid based on the corner information, and recognizing walls from the relationship between the grid and line segments on a floor plan. In wall recognition, it is determined that a wall is present when the intersection of the grid lies between two line segments that constitute a wall, and the distance between those line segments matches the thickness of the wall. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Publication No. 2023-119559 [Patent Document 2] Japanese Patent Publication No. 2025-135780 [Overview of the project] [Problems that the invention aims to solve]

[0008] The technology described in Patent Document 1 attempts to automatically recognize walls in spatial recognition by using a machine learning model that has been trained to identify the vertices of line segments that make up rooms, etc. However, architectural drawings also contain various information other than walls, such as line segments and text. Therefore, the possibility of misidentifying vertices cannot be ruled out. In this regard, Patent Document 1 states that if an error occurs, training data for correction will be input, but the possibility that the error may be used in cost estimation without being noticed cannot be ruled out, and there is a problem that training the system with correction data every time an error occurs is time-consuming.

[0009] In contrast, Patent Document 2 allows the user to specify the position of the corner, and the detection is determined based on whether the intersection of the grid lies between two line segments that constitute the wall. Therefore, it is believed that the accuracy of wall detection will be higher. However, the technology described in Patent Document 2 had the problem that, even if the wall detection accuracy was high, it required a great deal of effort due to the many operations performed by the user.

[0010] This invention has been made in view of the above-mentioned problems, and its purpose is to provide a building cost estimation system and method that can accurately recognize building elements and significantly reduce the effort required for estimating the necessary building materials. [Means for solving the problem]

[0011] The means employed by the inventors to solve the above problems are described below. The present invention relates to a building cost estimation system for estimating the amount of building materials needed from building drawing data. Its basic configuration includes a dimension acquisition means for obtaining dimension data included in the drawing data, an element recognition means for recognizing building elements to be estimated from the drawing data, a noise reduction means for removing noise within the surrounding region surrounded by the building elements recognized by the element recognition means, and an estimation means for estimating the amount of building materials needed from the noise reduction region obtained by the noise reduction means and the dimension data. In this invention, an architectural element refers to a portion of a region having a predetermined area that corresponds to a line segment constituting that region, and examples include walls that make up a room and concrete that makes up the foundation.

[0012] Drawings contain various types of noise, such as text, symbols, and dimension lines, in addition to the architectural elements necessary for cost estimation. They also include elements that complicate the shape of an area, such as columns and partitions. Furthermore, architectural elements themselves have thickness, and this thickness must be taken into account when obtaining coordinate values. Here, if we directly obtain coordinates from the architectural elements recognized by the element recognition means and calculate the necessary building materials, there is a possibility of misrecognizing the shape of the area by misidentifying noise as an architectural element, obtaining incorrect coordinates, or by ignoring columns, partitions, etc., when obtaining coordinates. Furthermore, there is a possibility of misrecognizing the shape of the area by obtaining coordinates from both the front and back line segments in the thickness direction of the architectural element.

[0013] Therefore, in this invention, the element recognition means determines the enclosed region surrounded by the recognized building elements, and removes noise such as characters, symbols, and irrelevant line segments contained within that enclosed region. Then, the necessary building materials are calculated from the noise-removed region obtained as a result of noise removal and the dimensional data acquired from the drawing data. The surrounding region is determined by finding the inner area enclosed by multiple image-recognized building elements, and removing the noise components from the pixels within that region. The resulting noise-free pixel region is used as the denoising region, and the length of the building elements can be determined from the length of its outer perimeter, allowing for the calculation of building material quantities.

[0014] By recognizing architectural elements using image recognition and then defining the inner region enclosed by these multiple architectural elements as the surrounding region, it is possible to identify the inner region of all architectural elements, even if the region has a complex shape due to the intertwining of architectural elements. Furthermore, by removing noise within the surrounding region through image processing, the resulting denoised region can be used as the region to be integrated. This allows for accurate determination of the region's shape and size without being affected by noise.

[0015] Another possible means to address the problem is to adopt a configuration in the above-described setup in which the noise reduction means replaces pixels within the surrounding region in the drawing data with a single color. If any pixels are included within the surrounding region, all of those pixels can be considered noise.

[0016] Therefore, in addition to recognizing and removing each individual noise, the noise reduction method can also effectively remove noise by performing a process that replaces all pixels within the surrounding region with a single color, as shown in the above configuration. By replacing all pixels within the surrounding region with a single color, the resulting denoising region becomes a solid colored area. This makes it easier to understand the shape of the denoising region and accurately calculate the coordinates of inflection points (including vertices and corners).

[0017] Another possible means to address the problem is to provide the noise reduction means with a closure determination means for determining whether or not to close the opening in the surrounding region, and a closure means for closing the opening when the closure determination means determines that the opening should be closed. When determining the surrounding area based on multiple architectural elements obtained by element recognition means, for example, if entrances to living spaces are detected as openings, the surrounding area may end up being a connected area consisting of multiple regions.

[0018] In the noise removal means with the above configuration, when there is such an opening, if the occlusion determination means determines that the opening should originally be occluded as a region (for example, since it is the entrance of a living room, it should originally be closed by a door to form one region), the occlusion means is configured to occlude the opening. This can prevent the regions adjacent to each other due to the opening from being determined as one region, and prevent the number and size of necessary building elements from being erroneously calculated.

[0019] As still another means that can be adopted for solving the problem, in the above configuration, the building element can be set as a wall, and the integration means can be configured to integrate the length of the wall required for the building from the dimensional data and the noise removal region. The building elements include all things that constitute the "regions" in a building. In the above configuration, the building elements are set as walls, and the lengths of those walls are integrated. Thereby, from the floor plan, the size, number, and cost of the walls required for the interior construction can be integrated.

[0020] As still another means that can be adopted for solving the problem, it can also be solved as the method of the following procedure without using the above-described building integration system. Its basic configuration can be a configuration including a dimension acquisition step for obtaining the dimensional data included in the drawing data, an element recognition step for recognizing the building elements to be integrated from the drawing data, a noise removal step for removing the noise in the surrounding region surrounded by the building elements recognized by the element recognition step, and an integration step for integrating the necessary building materials from the noise removal region obtained by the noise removal step and the dimensional data.

[0021] [[ID=十八]]With the above method, after recognizing the building elements by image recognition, by setting the inner region surrounded by those multiple building elements as the surrounding region, even when the regions have a complex shape due to the building elements being intertwined, etc., the inner regions of all those building elements can be specified. Furthermore, by removing noise within the surrounding region through image processing, the resulting denoised region can be used as the region to be integrated. This allows for accurate determination of the region's shape and size without being affected by noise.

[0022] In addition to the above configuration, the noise reduction step can replace pixels within the surrounding region in the drawing data with a single color. By replacing all pixels within the surrounding region with a single color, the resulting denoising region becomes a solid colored area. This makes it easier to understand the shape of the denoising region and accurately calculate the coordinates of inflection points (including vertices and corners).

[0023] Furthermore, the noise reduction step may also include a blocking determination step that determines whether or not to close an opening in the surrounding region when the surrounding region has a partial opening, and a blocking step that closes the opening when the blocking determination step determines that the opening should be closed. With the method described above, if there is an opening in the surrounding area, the closure determination step will determine that the opening should be closed off as part of the area (for example, since it is the entrance to a living room, the door should be closed to form a single area), and the closure means will close the opening. This prevents adjacent areas from being judged as a single area due to an opening, which could lead to incorrect calculations of the number and size of necessary building elements.

[0024] Furthermore, it is also possible to configure the building element to be a wall, and the calculation step to calculate the length of the wall required for the building from the dimensional data and the noise reduction area. Using the method described above, it is possible to calculate the size, number, and cost of the walls required for the interior construction from the floor plan. [Effects of the Invention]

[0025] The menu creation support system and menu creation support method of the present invention allows for accurate recognition of building elements by recognizing them using image recognition and removing noise within the surrounding area enclosed by the recognized building elements. Furthermore, by calculating the building materials using the resulting noise-removed area and dimensional data, the effort required to calculate the necessary building materials can be significantly reduced. [Brief explanation of the drawing]

[0026] [Figure 1] This is an explanatory diagram showing the overall configuration of the building cost estimation system of the present invention. [Figure 2] This is a flowchart illustrating the processing flow of the building cost estimation system of the present invention. [Figure 3] This is an explanatory diagram illustrating the contents of the dimension acquisition means in the building cost estimation system of the present invention. [Figure 4] This is an explanatory diagram illustrating the contents of the element recognition means in the building cost estimation system of the present invention. [Figure 5] This is an explanatory diagram illustrating the contents of the noise reduction means in the building cost estimation system of the present invention. [Figure 6] This is an explanatory diagram showing the contents of the blockage determination means and blockage means in the building cost estimation system of the present invention. [Figure 7] This is an explanatory diagram illustrating the contents of the cost estimation means in the building cost estimation system of the present invention. [Figure 8] This is an explanatory diagram showing a modified example of the menu creation support system of the present invention. [Modes for carrying out the invention]

[0027] Embodiments for carrying out the present invention will be described below with reference to Figures 1 to 7. Note that in the following description, each figure is schematically represented for the sake of simplification. As shown in Figure 1, the building cost estimation system (hereinafter also simply referred to as the system) 100 of the present invention can be operated by a user U using a terminal P as an application stored on a server S in the cloud. Information processed by the server S is displayed on the user U's terminal P in a timely manner, allowing the user U to visually confirm the information as the process progresses. In this case, since sending and receiving the entire data to be processed each time would increase the communication time, the processing speed can be improved by sending and receiving only the coordinate values ​​and numerical values ​​of the cost estimation results, which will be described later, from the server S to the terminal P. In the configuration shown in Figure 1, the system is a SaaS (Software as a Service) type that can be used on the web without being installed on terminal P. However, it may also be an on-premise type system that is installed on terminal P and operates standalone; the configuration is not limited.

[0028] System Configuration First, the overall configuration of this system 100 will be explained based on Figure 1, with reference to Figures 2 to 7 as appropriate. User U is a user of this system 100, and the system 100 reads architectural drawing data D from terminal P based on user U's operations. The drawing data D may be scanned paper drawings converted into image data, or it may be PDF data output directly from architectural CAD software. This system 100 treats the read drawing data D as raster data (image data), and subsequent calculations are performed as image processing. For ease of image processing, it is preferable that the drawing data D is converted to grayscale at the time of reading.

[0029] The drawing data D to be imported can be based on various drawings, but the target is a drawing that shows an area A having a predetermined area formed by building elements E, E, etc. For example, the drawing data D can be floor plan data in which an area A having a predetermined area, such as a living room or corridor, is shown by a wall, which is a building element E. In the following explanation, as an example, as shown in Figure 1, the drawing data D will be floor plan data, the building element E will be a wall, and the area A will be a living room or corridor.

[0030] The dimension acquisition means 1 acquires reference dimension data D1 from the drawing data D that has been read as described above. The acquisition of dimension data D1 can be performed by recognizing any dimension line D11 contained in the drawing data D and its corresponding dimension value D12 through image recognition processing, and defining the length per pixel. Alternatively, the user U may manually input the value of the recognized dimension value D12, or the user U may specify the coordinate values ​​of the start and end points of the dimension line D11 by clicking with the mouse.

[0031] The element recognition means 2 recognizes the building elements E·E… to be calculated. While the recognition of building elements E may be performed uniformly based on their image features, it is preferable to use a training dataset L to extract the pixel range of the wall, which is the building element E. Since walls have a predetermined thickness, they are generally represented as closed shapes such as rectangles in floor plans. The element recognition means 2 recognizes the pixels in the area enclosed by this closed shape as walls. In the element recognition means 2 when using the training dataset L, a large amount of floor plan data containing diverse floor plans is prepared, and the training dataset L obtained by machine learning a large amount of wall features described in these plans is used. Various machine learning methods can be used, but supervised learning that pre-identifies the wall parts in the floor plans is preferred, and deep learning using a neural network is more preferred.

[0032] The noise reduction means 3 comprises a region definition means 31, a pixel replacement means 32, a blockage determination means 33, and a blockage means 34. While the blockage determination means 33 and the blockage means 34 are not essential components, their inclusion is preferable. In the region definition means 31, the parts of the building elements E·E… extracted by the element recognition means 2 that are not pixels are defined as surrounding regions A1·A1…. Generally, since a floor plan has multiple rooms and corridors, multiple surrounding regions A1·A1… are defined within a single drawing data D, but it is also acceptable for only one surrounding region A1 to be defined.

[0033] Once the surrounding region A1·A1… is defined by the region definition means 31, the pixel replacement means 32 then replaces the noise pixels N·N… within the surrounding region A1·A1… with predetermined pixels. Noise N·N… includes, for example, characters representing the type or area of ​​a room, symbols representing housing equipment, dimension extension lines, etc., and any pixels other than the building elements E·E… that are subject to calculation. The pixel replacement means 32 replaces these noise N·N… pixels. For example, noise N·N… pixels can be replaced by filling them with white "RGB(255,255,255)". In the replacement, only the noise N·N… pixels may be replaced with a predetermined color, but by uniformly replacing the entire surrounding region A1·A1… with a single color, the noise N·N… pixels can be replaced as a result. In this way, the noise reduction means 3 defines the noise reduction region A2·A2, which is replaced with pixels of a predetermined color. In this case, the noise reduction region A2 may include an opening O such as a door or window. In this case, the opening O can be closed by the blockage determination means 33 and the blockage means 34. This process will be described later.

[0034] The calculation means 4 includes a coordinate identification means 41, a coordinate adjustment means 42, a length calculation means 43, a material information input means 44, and a material calculation means 45. These means calculate the length, number, and cost of the building elements E to be calculated from the noise reduction region A2. The coordinate identification means 41 identifies the coordinate values ​​of pixels at the inflection points (vertices and corners) of line segments that constitute the noise reduction region A2. As described above, the noise reduction region A2 is a closed region that is filled (replaced) with a predetermined color (e.g., white). Therefore, by connecting the pixels at the boundary between the replaced color and other colors, the outermost pixels of the noise reduction region A2 can be easily determined. At this time, the position where the directional continuity of adjacent pixels is lost is identified as the inflection point of the line segment. For example, if pixels that were continuous toward the X-axis change direction toward the Y-axis at a certain point, the position of the pixel immediately before that change in direction is identified as the coordinate value of the inflection point.

[0035] The coordinate adjustment means 42, when there are pairs of coordinate values ​​in a single noise reduction region A2 that have slight differences in the X-axis or Y-axis direction, unifies and aligns them to one of the coordinate values ​​or rounds them. In this case, a coordinate value normalized with one of the coordinate values ​​as the origin of the relative coordinates may be used, and rounding or alignment may be performed in predetermined units. This coordinate adjustment means 42 is not an essential component, but it is preferable to have it.

[0036] The length calculation means 43 calculates the actual length of each building element E·E… from the specified coordinate values. After determining the distance between adjacent bending points from the difference in coordinate values, it uses the previously acquired dimension data D1 (using the length per pixel) to calculate the actual length of the wall, which is the building element E, of the line segment between those bending points.

[0037] The material information input means 44 is a means for inputting various information about the wall materials to be used in the target building. This means may be provided as one of the cost estimation means 4, but it may also be pre-entered as master information. The material information input means 44 is used to input information such as the thickness and unit price per unit length of the wall material. In addition to this, information such as the manufacturer, material, and supplier may also be included.

[0038] The material calculation means 45 calculates information such as the required number of wall materials and the total cost to be used as building materials, based on the length of the building element E calculated by the length calculation means 43 and the unit price per length entered by the material information input means 44. The information to be calculated may include not only numbers and amounts, but also, in the case of walls, information such as wall height and wall area.

[0039] "Processing Procedures for This System" Next, an example of a specific calculation method for system 100 will be explained based on Figures 2 to 7. First, in the loading step S1 of the flowchart in Figure 2, user U operates terminal P to send drawing data D to system 100. If the drawing data D read by system 100 is raster data obtained by scanning a paper drawing, it may be slightly tilted due to the effects of scanning. In this case, processing to correct the tilt may be performed, for example, by using the straight lines of the drawing frame.

[0040] Next, in step S2 of the dimension acquisition step in the flowchart of Figure 2, a process is performed to extract an arbitrary dimension from the drawing data D and acquire dimension data D1. Dimension data D1 acquires the length (number of pixels) of the dimension line D11, and the dimension value D12 corresponding to the dimension line D11 is read by OCR (Optical Character Recognition). By dividing the read dimension value D12 by the number of pixels of the dimension line D11, the length per pixel can be calculated as a reference value. Using this reference value, it is possible to calculate the actual length from the number of pixels in the drawing data D.

[0041] At this time, the dimensions to be extracted are determined after recognizing multiple dimensions in the drawing data D, and the user U can select any dimension. In this case, it is preferable to define the length per pixel using the largest dimension value D12 and the dimension line D11, as this reduces errors. Figure 3 shows the process of actually acquiring dimensions. In the example in Figure 3, a dimension value D12 and dimension line D11 of 7,000, and a dimension value D12' and dimension line D11' of 6,675 are recognized. When user U selects the dimension value D12 of 7,000, the recognized dimension value D12 is displayed. At this point, user U can also manually modify the dimension value D12.

[0042] Next, in the element recognition step S3 of the flowchart in Figure 2, a process is performed to recognize the walls, which are the building elements E·E… to be calculated. In the example in Figure 4, using a training dataset L that has been trained by machine learning to recognize the characteristics of walls in a large number of floor plans, artificial intelligence technology is used to extract wall pixels from the drawing data D. Figure 4(a) shows the clothing storage area from drawing data D. While the clothing storage area appears to have a nearly square layout, there is a pipe space (PS) in the upper right corner and a large column protruding inward in the upper left corner. Closets are located on both the left and right sides, with doors at the bottom.

[0043] When element recognition step S3 recognizes walls as architectural elements E·E…, it is possible to recognize the walls surrounding the clothing storage area as a single unit, but it is also possible to recognize each face individually. Figure 4(b) shows the state where each wall is recognized individually. Note that the recognition target may also include columns and other elements other than architectural elements E.

[0044] Next, in the noise reduction step S4 in the flowchart of Figure 2, the parts of the multiple building elements E·E… recognized in the element recognition step S3 that are not pixels are defined as surrounding regions A1·A1…. The noise reduction means S4 consists of a region definition step S41, a pixel replacement step S42, a blockage determination step S43, and a blockage step S44. First, in the region demarcation step S41, the enclosed region A1, surrounded by the recognized architectural elements E·E…, is defined. In Figure 5(a), the recognized architectural elements E·E… are shown in light gray for convenience. It can be seen that not only the four walls, but also the wall constituting the PS in the upper right is recognized, as well as the wall surrounding the column in the upper left. In this way, the region surrounded by architectural elements E·E… is defined as enclosed region A1. This surrounding area A1 includes a "clothing storage room" and "25.51m 2 In addition to the characters ", the code includes various noises such as lines for closets, auxiliary lines for dimension indications, and lines indicating the opening range of doors.

[0045] Next, in the pixel replacement step S42, all pixels in the surrounding region A1 are replaced with a single color. In the example in Figure 5(b), the pixels are filled with white "RGB(255,255,255)", but other colors may be used, or only the pixels in the noise N·N… region may be replaced. The pixel replacement step S42 results in the appearance of a predetermined area unified by a single color. This area is designated as the noise reduction area A2.

[0046] Incidentally, as shown in Figure 6, the surrounding region A1 sometimes includes an opening O, such as an entrance or exit. If the pixel replacement step S42 is performed in a state that includes such an opening O, the noise reduction region A2 will recognize multiple adjacent regions connected by the opening O as a single region. Therefore, if there is a portion with a predetermined interval within the noise reduction region A2, the occlusion determination step S43 is executed to determine whether or not that portion is an opening O that should be closed. At this time, the determination may be made based on the size and position of the interval, but it may also be made using artificial intelligence technology with the training dataset L.

[0047] If the closure determination step S43 determines that the opening O should be closed, the closure step S44 is executed to close the opening O. Specifically, a rectangle inscribed in the portion excluding the opening O is calculated, and a noise reduction region A2 closed by that rectangle is obtained.

[0048] Once the noise reduction region A2 is obtained, the integration step S5 in the flowchart of Figure 2 is executed. The integration step S5 consists of a coordinate identification step S51, a coordinate adjustment step S52, a length calculation step S53, a material information input step S54, and a material integration step S55. In the coordinate identification step S51, the coordinates of all inflection points in the obtained noise reduction region A2 are identified. Specifically, the entire drawing data D is treated as a Cartesian coordinate system with X and Y axes, and an arbitrary inflection point in the noise reduction region A2, such as the lower left inflection point shown in Figure 7(a), is set as the reference coordinate (0,0) of the relative coordinate system. Then, the coordinates of each inflection point are identified as coordinates using the relative coordinate system. For example, the coordinate value at this time can be determined by the number of pixels from the reference coordinate as a distance.

[0049] Once the coordinates of each inflection point are identified, the coordinate adjustment step S52 is performed. For example, there may be two inflection points at very close coordinates, or there may be fractional values ​​in the coordinate values. Also, for multiple inflection points aligned in the X-axis direction, there may be slight differences in the Y-axis coordinate values ​​of those inflection points. In the coordinate adjustment step S52, for coordinate values ​​with such slight deviations, it is possible to remove all but one of several adjacent inflection points, round off any fractional coordinate values ​​as shown in Figure 7(b), or align them to a predetermined pitch.

[0050] Based on the coordinate values ​​of the inflection points organized in this way, the length of each building element E·E… is calculated in the length calculation step S53. The length is calculated by determining the number of pixels between adjacent inflection points based on the coordinate values ​​and using the length per pixel obtained in advance from the dimension data D1. In the example shown in Figure 7(c), it can be seen that the width of the clothing storage room (4.84m) and the depth to the PS (pipe storage) (4.33m) have been accurately calculated. Note that the calculated length values ​​may be rounded or aligned at this stage. Once the length between each bending point has been calculated, the material information input step S54 is executed. In the material information input step S54, the user U is prompted to input various information about the building element E to be used for the noise reduction region A2. For example, as shown in Figure 7(d), the wall thickness, wall height, unit price, etc., are input for each calculated wall. In addition, the user may also be prompted to input ceiling height, material specifications (type), etc.

[0051] Once the material information has been entered in the material information input step S54, the cost estimation step S55 is executed. In the cost estimation step S55, the required number and cost of materials for each building element E·E…, as well as the total cost for the entire area A, are calculated. In the example shown in Figure 7(d), the calculation result is displayed on the same screen used in the material information input step S54.

[0052] As described above, this system 100 can accurately recognize walls by using artificial intelligence technology for image recognition and removing noise N within the surrounding region A1 enclosed by the recognized walls. Then, the required wall size and cost can be easily calculated using the resulting noise-removed region A2 and dimensional data D1.

[0053] "Differentiation Example 1" Next, a modified building cost estimation system 101 according to the present invention will be described with reference to Figure 8. In the following description, the same parts will be referred to by the same reference numerals, and redundant explanations will be omitted.

[0054] In this modified example, the differences from the forms shown in Figures 1 to 7 are that the drawing data D is based on the foundation plan, the building element E is the rising portion of the foundation, and area A is the space enclosed by the rising portion. In this modified version, system 101 calculates the amount and cost of concrete used for the entire raft foundation by calculating the amount of concrete used for the foundation's rising section, which is a building element E, from drawing data D based on the foundation plan, as well as calculating the amount of concrete used for the floor slab.

[0055] The modified system 101 is executed with the same system configuration as in Figure 1 and the same steps as in Figure 2. When the drawing data D is read in the reading step S1, the dimension acquisition means 1 executes the dimension acquisition step S2 and acquires the dimension data D1. Next, the element recognition means 2 executes the element recognition step S3 and recognizes the foundation rising section, which is the building element E. This element recognition means 2 prepares a large amount of foundation plan drawing data containing various foundation shapes and uses a training dataset L obtained by machine learning a large amount of the features of the foundation rising sections described therein. Next, the noise reduction means 3 performs the noise reduction step S4. The drawing data D obtained based on the foundation plan includes figures indicating anchor bolts within the rising section and symbols indicating foundation stones within area A. These symbols are removed as noise N by the noise reduction means 3.

[0056] Once the noise reduction means 3 obtains the noise reduction region A2, the integration means 4 performs the integration step S5. In the material cost estimation step S55 of cost estimation step S5, not only the amount and cost of concrete used for the rising section are calculated, but also the amount of floor concrete used for area A is calculated from the area of ​​noise reduction area A2 surrounded by the rising section. At this time, the area used for boundary area A3 between multiple areas A·A is also included in the cost estimation.

[0057] Thus, in this modified example, the foundation's rising section is recognized using image recognition technology based on artificial intelligence, and the noise N within the surrounding region A1 enclosed by the recognized rising section is removed, thereby enabling accurate recognition of the rising section. The resulting noise-removed region A2 and dimensional data D1 can then be used to easily calculate the required amount and cost of concrete.

[0058] The present invention is not limited to the embodiments described above, and can be modified as appropriate within the scope of the claims of the present invention. For example, the drawing data may include architectural drawings such as floor plans, structural drawings such as foundation plans, or data based on equipment drawings that include electrical wiring and piping. Furthermore, the calculation may include not only the number and cost of necessary building materials, but also the labor hours required to construct that section. [Explanation of symbols]

[0059] 100,101 Architectural Cost Estimation System 1. Means for obtaining dimensions 2 element recognition means 3. Noise reduction means 31. Means for defining the area 32 Pixel replacement means 33 Obstruction determination means 34 Obstruction means 4. Calculation method 41 Coordinate identification means 42 Coordinate adjustment means 43 Length Calculation Means 44. Material information input means 45. Material Calculation Method Area A A1 Surrounding area A2 Noise Reduction Area A3 boundary area C Customer Terminal D Drawing data D1 Dimension Data D11 Dimension Line D12 Dimension Value E Architectural elements L Training dataset N Noise O opening P terminal U users

Claims

1. A building cost estimation system for calculating the necessary building materials from architectural drawing data, A dimension acquisition means for obtaining dimension data included in the aforementioned drawing data, An element recognition means for recognizing building elements to be calculated from the aforementioned drawing data, A noise reduction means that replaces all pixels within the surrounding region surrounded by the architectural elements recognized by the element recognition means with a single color, A calculation means for calculating the necessary building materials from the noise reduction region obtained by the noise reduction means and the dimensional data, A building cost estimation system characterized by comprising the following features.

2. The noise reduction means includes, in the case where the surrounding region has a partial opening, A means for determining whether or not the opening should be closed, When the blocking determination means determines that the opening should be closed, a blocking means for closing the opening, The building cost estimation system according to claim 1, characterized by comprising:

3. The aforementioned building element is a wall, The building cost estimation system according to claim 1, characterized in that the estimation means estimates the length of the wall required for construction from the dimension data and the noise reduction region.

4. A construction cost estimation method for calculating the necessary building materials from architectural drawing data, A dimension acquisition step for obtaining dimension data included in the aforementioned drawing data, The element recognition step involves recognizing the building elements to be calculated from the aforementioned drawing data, The noise reduction step involves replacing all pixels within the surrounding region enclosed by the architectural elements recognized in the element recognition step with a single color. A calculation step is performed to calculate the necessary building materials from the noise reduction region obtained by the noise reduction step and the dimensional data, A method for calculating building costs, characterized by comprising the following features.

5. The noise reduction step includes, in the case where the surrounding region has a partial opening, A blocking determination step to determine whether or not the opening should be closed, When the aforementioned closure determination step determines that the opening should be closed, a closure step is performed to close the opening. The building cost estimation method according to claim 4, characterized by comprising:

6. The aforementioned building element is a wall, The building cost estimation method according to claim 5, characterized in that the estimation step involves estimating the length of the walls required for construction from the dimensional data and the noise reduction region.

Images