Method, apparatus, and system for providing an AI-based image processing solution for automated estimation of service cost and moving volume in small-scale relocation
Patent Information
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- NATIONWIDE TRANSPORTATION CO LTD
- Filing Date
- 2025-12-01
- Publication Date
- 2026-08-03
Smart Images
Figure 112025134916690-PAT00001_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to a technology field that automatically identifies the condition of an indoor space based on video analysis technology using artificial intelligence, estimates the type, quantity, and volume of moving items, and automatically calculates estimate information for small-scale moving.
[0002] More specifically, the invention relates to a technology that analyzes photos or videos of an indoor space provided by a user terminal to generate captured geometric information, recognizes furniture, home appliances, box-shaped items, and irregular items as individual objects through an image analysis procedure including object segmentation and perspective correction, and then calculates the size and scale of said objects.
[0003] Furthermore, the present invention relates to a technology field that automatically determines narrow sections of a movement path and sections with insufficient turning radius by estimating the corridor width, effective opening width of a door, floor area, and internal dimensions of an elevator based on images, and derives the difficulty of moving household goods using these analysis results. Furthermore, it relates to a service automation technology field that automatically generates an estimated cost required for a small-scale move by combining the calculated volume and difficulty of the household goods, the moving distance, and building structural conditions, and supports driver assignment by predicting the type of suitable vehicle and the need for transport personnel.
[0004] Furthermore, the present invention relates to a technical field for minimizing dispatch errors caused by insufficient or inconsistent information directly entered by customers by comparing user input information with image analysis results to detect omissions or inaccuracies in the input information and, if necessary, providing the user with a request for correction. It also relates to a technology that automatically verifies whether a task has been completed by aligning image data before and after the move to determine changes in item locations, omissions, and the state of organization. Background Technology
[0006] In general, the market size for small-scale moving services is continuously expanding due to the increase in single-person households and the miniaturization of housing types, and there is a steady increase in demand for short-distance moving services centered on moving within relatively limited spaces such as studio apartments, officetels, and urban residential units.
[0007] In response to these changes, individual users desire a simpler moving process; however, in actual service environments, the insufficient accuracy of the information provided by users frequently places an unexpected burden on moving companies and drivers. While a system has been established in the general freight transportation sector where shippers clearly provide cargo information and work requirements according to standardized procedures, individual users utilizing small-scale moves often have limited moving experience and are unfamiliar with the item information entry process, leading to frequent omissions, under-information, or over-information.
[0008] In particular, there are many cases where only a few photos or brief descriptions are provided without accurately determining the actual dimensions or quantities of furniture and home appliances, leading to a growing discrepancy between the actual scale of the work and the reservation information.
[0009] Furthermore, photos or videos taken arbitrarily by users are prone to distortion depending on the shooting angle, distance, tilt, and lighting conditions. Since items are often obscured by others or perspective is exaggerated, the simple image upload method has limitations in accurately identifying the actual size of items and the spatial structure. Due to these limitations, existing services have remained limited to using uploaded images only as reference material rather than directly for quantitative analysis, which has further increased the likelihood that the actual scale of the moving belongings differs from the information entered by the customer.
[0010] Small-scale moves are significantly affected by physical constraints during the moving process, such as the effective opening width of doors, the height of entrance thresholds, narrow corridor sections, the turning radius of stairwells, and the internal dimensions of elevators, but existing brokerage services are not sufficiently equipped with the ability to automatically determine this spatial structure information or quantify the difficulty of movement.
[0011] As a result, there were many instances where on-site work delays or dispatch errors occurred because critical moving conditions, such as situations where vehicle access is impossible, the presence of furniture requiring disassembly, or the need to dispatch a small vehicle instead of a 1-ton truck, could not be predicted in advance. This leads to a chain reaction of delays in driver schedules, increased customer waiting times, and a decline in service quality, which poses a problem that significantly reduces the efficiency of overall service operations.
[0012] Furthermore, due to the nature of person-to-person services, small-scale moves often involve bookings based on simple phone consultations or text messages. Consequently, the information input process is unsystematic, and the data provided by customers is often poorly structured, making it difficult to secure objective data for accurate work preparation. Customers frequently feel burdened by the task of classifying their belongings or recording quantities, leading them to request a vehicle with only minimal information. This often results in the actual workload growing larger than anticipated or requires additional personnel, causing inconvenience for both the driver and the customer.
[0013] In many cases, sufficient features are not provided to objectively verify work quality even after the move is completed, creating a potential for disputes regarding completion status, missing items, or the state of organization. Existing simple photo verification methods are limited by the difficulty of comparison due to differences in shooting conditions and the reliance on manual human judgment, resulting in a lack of systematic service quality management.
[0014] Meanwhile, video analysis technology utilizing AI models has recently been advancing across various fields, and while basic functions such as object recognition and spatial classification are being implemented at a high level, integrated analysis technology capable of estimating the actual size of items, analyzing spatial geometry, assessing the difficulty of movement routes, and linking vehicle selection and personnel allocation—to provide features specialized for moving services—has not yet been sufficiently secured. AI technology based on simple object detection lacks perspective and tilt correction, making it difficult to calculate physical dimensions; it also lacks the technical foundation to automatically interpret structural constraints of movement routes; furthermore, technology directly linked to the moving service estimate calculation process remains insufficient.
[0015] Therefore, advanced technology is required to calculate the actual volume of items based on user-captured images, automatically analyze spatial structures, quantify the difficulty of movement routes, determine the suitability of vehicles and personnel, and even compare and analyze the condition before and after the move. Such technology lays the foundation for service providers to accurately and efficiently perform dispatching and estimate calculations while reducing the burden of information input on customers, and is an essential element for maintaining stable service quality.
[0016] As such, in order to enhance the accuracy and efficiency of small-scale moving services, there is an urgent need for a new type of technological means that combines image-based automatic analysis technology with moving load size estimation technology; therefore, the development of technology that provides these functions is currently required. Prior art literature
[0018] Korean Patent Publication No. 10-2021-0073087 (Method for Providing Moving Estimate Service) The problem to be solved
[0019] The purpose of this invention is to provide technical means to resolve the limitations of existing small-scale moving services, where it is difficult to sufficiently determine the actual scale of belongings and movement route conditions based solely on the limited information provided by the user. Items entered directly by the user are prone to omitting item types or quantities, and item sizes are often not accurately specified, frequently resulting in discrepancies between the assigned vehicles or personnel and the actual on-site conditions.
[0020] Therefore, there is a need for technology that can quantitatively determine the spatial structure and item sizes based on photos or videos taken by the user, and automatically calculate the difficulty of the task and the scale of the moving based on this.
[0021] Furthermore, perspective distortion occurs in captured images depending on the shooting position, angle, and tilt, and there is a problem where the overall shape cannot be accurately identified when part of an object is obscured. Since it is difficult to ensure the reliability of image-based analysis without correcting these geometric shooting problems, the present invention focuses on a technical configuration that calculates information on vanishing points, horizons, and camera tilt through geometric analysis of the shooting and incorporates this into perspective correction and size calculation.
[0022] In small-scale moves, not only the volume of goods but also on-site structural conditions, such as narrow sections of the movement path, elevator dimensions, and the effective opening width of doors, are important. It is difficult to accurately determine this information using existing user input methods, and since it is often confirmed only on-site, it leads to work delays or dispatch errors. The present invention aims to fundamentally resolve these problems by automatically determining structural constraints of the indoor space through image analysis and reflecting this information in vehicle selection and personnel allocation decisions.
[0023] Moving services require a clear comparison of the condition before and after the move to minimize disputes between customers and technicians; however, current methods lack objectivity and are difficult to repeat because they require a person to manually compare photos. This invention aims to solve these problems by providing a structure that aligns image data before and after the move into the same coordinate system and automatically analyzes changes in item locations and floor areas to determine whether the work is complete and what items are missing. means of solving the problem
[0025] A method for providing a solution for automatically calculating service estimates and the scale of moving based on image processing using an artificial intelligence model according to an embodiment of the present invention comprises: an image reception step in which a small-scale moving platform server determines a photograph or video of an indoor space transmitted from a user terminal as input data satisfying a preset minimum resolution value and shooting distance estimation criteria, and receives and stores it; and a shooting geometry analysis step in which the small-scale moving platform server extracts candidates for floor lines, wall corner lines, and ceiling lines using a straight line detection algorithm on the photograph or video stored in the image reception step, calculates a vanishing point and a horizontal line using the intersection of the extracted straight lines, and calculates shooting angle, shooting position, and camera tilt information of the photograph or video using the calculated vanishing point and the horizontal line. An object volume calculation step in which the small-scale moving platform server executes an object segmentation model using the photographic geometric information calculated in the photographic geometric analysis step to separate furniture, home appliances, box-shaped items, and irregularly shaped items from the input photo or video into independent objects, extracts the representative edge length, side ratio, and texture-based features of each object, and calculates the height, width, and depth of each object by performing calculations according to the photographic tilt correction formula and the perspective distance correction formula; and a spatial structure analysis step in which the small-scale moving platform server determines the actual scale of the floor surface by comparing the camera position information calculated in the photographic geometric analysis step with reference length candidates such as floor tile specifications or standard specifications for the effective opening width of a door, and calculates the floor surface area of the indoor space, the corridor width, the effective opening width of the door, and the internal dimensions of the elevator using the determined scale.Moving load size calculation step in which the small-scale moving platform server calculates the total moving load volume by summing the volumes and quantities of each object calculated in the object volume calculation step, and determines a difficulty grade according to a difficulty grade standard table stored in the small-scale moving platform server based on the passage dimensions and the degree of narrowing of the movement path width calculated in the spatial structure analysis step, and assigns a difficulty value; user input verification step in which the small-scale moving platform server corresponds the item list entered from the user terminal with the object list calculated in the object volume calculation step to determine whether there are any omissions, and verifies the accuracy of the user input by calculating whether the absolute error and ratio error between the estimated volume for each input item and the volume calculated based on actual measurement exceed a preset standard value range; and user information correction request step in which the small-scale moving platform server provides an information correction request message including detailed information on the error item to the user terminal if a deviation exceeding the standard value exists in the user input verification step. A quotation amount calculation step in which the small-scale moving platform server combines the total volume of moving goods and the movement difficulty value calculated in the moving goods size calculation step with the movement distance and building structure conditions, calculates the basic cost, movement distance cost, difficulty weighting cost, packing personnel requirement cost, and vehicle accessibility cost according to the cost calculation formula stored in the small-scale moving platform server, and sums the respective calculation results to generate a small-scale moving quotation amount; a vehicle and personnel judgment step in which the small-scale moving platform server determines a suitable vehicle among a Damas vehicle, a Labo vehicle, or a 1-ton vehicle based on a vehicle suitability judgment criterion that compares the total volume and maximum dimensions of goods calculated in the moving goods size calculation step with the movement path dimensions, compares the internal dimensions of the vehicle cargo compartment with the maximum dimensions of the goods, and determines whether transport personnel are required based on the movement difficulty and the quantity of goods;
[0026] The small-scale moving platform server transmits the result of the vehicle and manpower determination step to an external driver assignment platform server and receives driver information including the driver's location, working time, and vehicle ownership status from the external driver assignment platform server, and prioritizes selecting a driver who satisfies the conditions of work start time, travel distance, and vehicle suitability; the small-scale moving platform server transmits the driver selected in the driver list lookup step to the user terminal to provide the assignment result; the small-scale moving platform server stores the data set generated in the image reception step, shooting geometry analysis step, object volume calculation step, spatial structure analysis step, moving load size calculation step, user input verification step, estimate amount calculation step, and driver list lookup step as training data, and performs a parameter update procedure configured to periodically update an AI model capable of training using the training data, thereby improving image analysis accuracy and estimate calculation accuracy, wherein the straight line detection algorithm used in the shooting geometry analysis step applies the Canny edge detection algorithm to a photo or video to extract edge points, and applies the Hough transform to the extracted edge points to detect the position and direction of the straight line.
[0027] The above-described shooting geometric analysis step may include a shooting tilt correction step that extracts wall boundary lines, floor boundary lines, and ceiling boundary lines using the above-described straight line detection algorithm, determines a vanishing point from the distribution of intersection points of the extracted multiple straight lines, and calculates a shooting tilt correction value using the relative positional relationship between the location of the vanishing point and the image center point.
[0028] The object volume calculation step may include an object shape correction step in which, when a part of the object's outline in a photograph or video is not detected by another object or background, a standard shape template to which the object belongs is searched using the detected outline fragments, color information, and texture information, and the size of the object is corrected by estimating the shape of the undetected area based on the template matching result.
[0029] The above spatial structure analysis step may include a movement path analysis step that sequentially searches indoor corridors, entrances, doors, and elevators on an image, calculates the width, height, and effective opening dimensions of each area, and reflects them in the movement difficulty parameter.
[0030] The above user input verification step may include a missing item detection step that compares an image analysis-based object list with a user input list and provides the missing item to the user terminal if major furniture or home appliances are missing.
[0031] The above-mentioned estimate amount calculation step may include a detailed cost calculation step that calculates the basic travel distance cost, the high-rise work cost, the stair transport cost, and the narrow corridor work cost, respectively, and calculates the final estimate amount by summing the calculated cost values.
[0032] The above vehicle and personnel determination step may include a vehicle selection step for selecting a vehicle by determining whether vehicle accessibility is restricted based on the width of the narrowest area in the travel path, and if vehicle accessibility is restricted, comparing whether the width of the narrowest area is greater than or equal to the width through which a vehicle can pass and whether it is greater than or equal to the minimum turning radius through which a vehicle can turn.
[0033] The above article assignment notification step may include a schedule verification step that determines whether there is a conflict between the available work time and the user's desired time, and if a conflict exists, compares the article's available time range with the user's requested time range to re-select an article with an overlapping time range.
[0034] A method for estimating a small-scale move and automatically calculating the size of moving goods using an artificial intelligence model-based image analysis may further include a step of classifying irregular items and recommending packing boxes, wherein the small-scale moving platform server classifies the item type for irregularly shaped items other than standardized furniture or home appliances among the objects separated in the object volume calculation step using contour complexity, surface texture features, shape asymmetry index, and texture-based features, and automatically calculates the size and quantity of packing boxes and recommended packing methods by referring to a pre-stored standard packing specification database for each classified type.
[0035] A method for estimating a small-scale move and automatically calculating the size of moving belongings using an artificial intelligence model-based image analysis may further include a step of comparing before and after moving images and verifying the work, wherein the small-scale moving platform server receives a photo or video of the move transmitted from the user terminal or the driver terminal after the moving work is completed, aligns the photo or video of the move before the move with the photo or video of the move after the move into the same coordinate system, and then determines whether the change in the location of an item is within an allowable range, whether the floor area has increased or decreased compared to the initial arrangement, and whether the spacing between item objects has changed compared to the initial state, by comparing the presence and location change of items, the change in pixel area classified as a floor area in the before and after moving images, and the change in the relative spacing between item objects, thereby determining whether the work is completed and identifying missing items.
[0036] A method for estimating and automatically calculating the size of moving goods for small-scale moving using an artificial intelligence model-based image analysis may further include a movement path obstacle analysis and movement path correction step in which the small-scale moving platform server generates a movable path as a virtual movement trajectory using floor area information, corridor width information, effective door opening width information, and elevator dimension information calculated in the space structure analysis step, detects the presence of obstacles on the generated trajectory, detects narrow sections consisting of sections with insufficient curb height and turning radius, and sections where the width is smaller than a preset threshold width, and calculates a movement difficulty value by comparing height, width, and turning radius values, which indicate the degree to which each obstacle restricts the movement path, with the maximum dimensions of the goods and the clearance width within the movement path. Effects of the invention
[0038] According to the present invention, users can systematically identify the total scale of moving belongings and spatial structural constraints simply by taking a photo or video of the indoor space, thereby significantly reducing the burden of complex information input that was previously required in the process of booking small-scale moves. While the process of users manually listing items or finding and inputting specifications is cumbersome and prone to errors, the present invention can generate objective data by automatically analyzing the type, arrangement, shape, and size of items from captured images, thereby greatly reducing the intensity of customer participation and naturally improving the quality of input information.
[0039] The image processing procedure of the present invention, which combines photographic geometric analysis, perspective correction, and texture-based feature analysis, can calculate the actual physical dimensions of an object after correcting distortion factors appearing in the image; thus, it enables the calculation of accurate specifications that were difficult to achieve with simple object recognition technology. This function reflects not only the actual size of furniture or home appliances constituting the moving shipment but also the morphological characteristics of box-shaped or irregular items, allowing for the calculation of values close to the actual volume and significantly improving the precision of moving estimates.
[0040] Furthermore, the present invention not only calculates the volume of the item but also automatically analyzes structural constraints of the indoor space, thereby enabling a more precise assessment of the difficulty of the entire movement path. By analyzing elements such as narrow corridor widths, effective door opening widths, and elevator interior dimensions from captured images, it is possible to determine in advance whether the item can actually pass through the path. This function plays a crucial role in the technician assignment stage. In existing services, it was common to discover belatedly on-site that the item could not pass or required disassembly; however, the present invention prevents such prediction failures in advance, thereby reducing factors that cause delays in technician schedules or conflicts with customers.
[0041] The vehicle and personnel assignment judgment function of the present invention determines the appropriate vehicle type and personnel size by synthesizing information on item size calculated based on images, the total volume of moving goods, and obstacles along the movement path. This enables much more accurate and rational dispatching compared to existing experience-dependent judgment methods. By automatically determining whether a vehicle can pass through a route and selecting a vehicle while considering items requiring disassembly or sections with insufficient turning radii, it provides predictable schedules and a stable service experience for both drivers and customers. This significantly contributes to the efficiency of driver schedules and helps them safely complete a greater number of assignments within the same daily workload.
[0042] The present invention includes a function that can verify the accuracy of input data by automatically comparing user input information with image analysis results, thereby effectively resolving the problem of missing information that commonly occurs in existing services. Since it is difficult for users to identify every item in their home individually, they often fail to input important items or underestimate their volume. The present invention automatically recognizes the type of item during the image analysis stage and cross-verifies user input information; by immediately notifying the user if there are any missing items or incorrect information, it reduces situations where both the customer and the technician experience difficulties due to unexpected variables.
[0043] Furthermore, the present invention possesses the technical advantage of objectively determining whether a task has been completed, as it can automatically verify changes in item positions, floor surface exposure, and spacing between items by aligning pre- and post-moving images into the same coordinate system. Unlike conventional subjective photo comparison methods, this judgment is based on automatically calculated quantitative results, significantly reducing the possibility of disputes and providing platform service providers with a foundation to technically establish a quality assurance system.
[0044] The artificial intelligence model learning function of the present invention has a structure that accumulates data sets generated during the image analysis process as training data to iteratively improve the model's performance. Therefore, as the application period of the present invention lengthens, product classification accuracy, volume calculation precision, movement path analysis accuracy, and vehicle selection prediction performance naturally become more advanced. This provides the effect of continuously and stably growing the technical foundation of the platform service and serves as a basis for securing a long-term technological advantage over competitors.
[0045] As such, the present invention is highly useful in that it can comprehensively achieve, within a single system, a reduction in the burden of entering customer information, precise analysis based on captured images, automatic determination of narrow sections in the movement path, improved accuracy in assigning vehicles and personnel, automation of comparing pre- and post-moving conditions, reduction of disputes, maximization of operational efficiency, assurance of service quality, and long-term improvement of AI performance. Consequently, the present invention significantly enhances the quality and reliability of the entire process of small-scale moving services, provides a stable and predictable moving experience for both customers and drivers, and offers platform service providers the effect of simultaneously achieving operational efficiency and strengthening technological competitiveness.
[0046] The effects of the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by a person skilled in the art from the description below. Brief explanation of the drawing
[0048] FIG. 1 is a diagram illustrating a method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention. FIG. 2 is a schematic diagram illustrating a method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention. FIG. 3 is a block diagram showing the configuration of a processor, memory, and network interface included in a small-scale moving platform server of a method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention. FIG. 4 is a configuration diagram showing a system environment to which a method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention is applied. Specific details for implementing the invention
[0049] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.
[0050] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.
[0051] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another.
[0052] For example, the first component may be named the second component, and similarly, the second component may also be named the first component.
[0053] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or joined to that other component, or that there may be other components in between.
[0054] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.
[0055] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.
[0056] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.
[0057] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but may be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims.
[0058] In the embodiments of the present invention, unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in the embodiments of the present invention.
[0059] The shapes, sizes, ratios, angles, numbers, etc. disclosed in the drawings for explaining embodiments of the present invention are exemplary, and therefore the present invention is not limited to the depicted details. Furthermore, in describing the present invention, if it is determined that a detailed description of related known technology may unnecessarily obscure the essence of the present invention, such detailed description is omitted. Where terms such as "includes," "has," or "is made up" are used in this specification, other parts may be added unless "only" is used. Where a component is expressed in the singular, it includes cases where it includes the plural unless specifically stated otherwise.
[0060] In interpreting the components, they are interpreted to include a margin of error even in the absence of a separate explicit statement.
[0061] In the case of describing positional relationships, for example, when the positional relationship between two parts is described using expressions such as 'on,' 'upper,' 'lower,' or 'next to,' one or more other parts may be located between the two parts unless 'immediately' or 'directly' is used.
[0062] When elements or layers are referred to as "on" another element or layer, this includes cases where another layer or element is placed directly on top of or in between. Throughout the specification, the same reference numerals refer to the same components.
[0063] The size and thickness of each component shown in the drawings are illustrated for convenience of explanation, and the present invention is not necessarily limited to the size and thickness of the illustrated components.
[0064] The features of each of the various embodiments of the present invention may be combined or combined with one another, either partially or wholly, and as will be fully understood by those skilled in the art, various technical interlocking and operation are possible, and each embodiment may be implemented independently of one another or together in an interlocking relationship.
[0066] The present invention relates to a technology field that automatically identifies the condition of an indoor space based on video analysis technology using artificial intelligence, estimates the type, quantity, and volume of moving items, and automatically calculates estimate information for small-scale moving.
[0067] More specifically, the invention relates to a technology that analyzes photos or videos of an indoor space provided by a user terminal to generate captured geometric information, recognizes furniture, home appliances, box-shaped items, and irregular items as individual objects through an image analysis procedure including object segmentation and perspective correction, and then calculates the size and scale of said objects.
[0068] Furthermore, the present invention relates to a technology field that automatically determines narrow sections of a movement path and sections with insufficient turning radius by estimating the corridor width, effective opening width of a door, floor area, and internal dimensions of an elevator based on images, and derives the difficulty of moving household goods using these analysis results. Furthermore, it relates to a service automation technology field that automatically generates an estimated cost required for a small-scale move by combining the calculated volume and difficulty of the household goods, the moving distance, and building structural conditions, and supports driver assignment by predicting the type of suitable vehicle and the need for transport personnel.
[0069] Furthermore, the present invention relates to a technical field for minimizing dispatch errors caused by insufficient or inconsistent information directly entered by customers by comparing user input information with image analysis results to detect omissions or inaccuracies in the input information and, if necessary, providing the user with a request for correction. It also relates to a technology that automatically verifies whether a task has been completed by aligning image data before and after the move to determine changes in item locations, omissions, and the state of organization.
[0071] However, generally speaking, the market size for small-scale moving services is continuously expanding due to the increase in single-person households and the miniaturization of housing types, and there is a steady increase in demand for transporting household goods focused on short-distance moves within relatively confined spaces such as studio apartments, officetels, and urban residential units. In response to these changes, individual users desire to proceed with moving through simpler procedures; however, in actual service environments, the accuracy of the information provided by users is often insufficient, frequently resulting in unexpected burdens for moving companies and drivers.
[0072] In the general cargo transportation sector, a system has been established where shippers clearly provide cargo information and work requirements according to standardized procedures. However, individual users of small-scale moves often have limited moving experience and are unfamiliar with the item information entry process, leading to frequent omissions, under-information, or over-information. In particular, as users often provide only a few photos or brief descriptions without accurately determining the actual dimensions or quantities of furniture and appliances, a significant discrepancy arises between the actual scale of the work and the reservation information.
[0073] Furthermore, photos or videos taken arbitrarily by users are prone to distortion depending on the shooting angle, distance, tilt, and lighting conditions. Since items are often obscured by others or perspective is exaggerated, the simple image upload method has limitations in accurately identifying the actual size of items and the spatial structure. Due to these limitations, existing services have remained limited to using uploaded images only as reference material rather than directly for quantitative analysis, which has further increased the likelihood that the actual scale of the moving belongings differs from the information entered by the customer.
[0074] Small-scale moves are significantly affected by physical constraints during the relocation process, such as the effective opening width of doorways, the height of entrance thresholds, narrow corridors, turning radii of stairwells, and internal elevator dimensions. However, existing brokerage services are not sufficiently equipped with the capabilities to automatically determine this spatial structure information or quantify the difficulty of movement. Consequently, critical moving conditions—such as situations where vehicle access is impossible, the presence of furniture requiring disassembly, or the need to dispatch a small vehicle instead of a 1-ton truck—cannot be predicted in advance, often leading to on-site work delays or dispatch errors. This results in a chain reaction of driver schedule delays, increased customer waiting times, and a decline in service quality, thereby significantly reducing the overall efficiency of service operations.
[0075] Furthermore, due to the nature of person-to-person services, small-scale moves often involve bookings based on simple phone consultations or text messages. Consequently, the information input process is unsystematic, and the data provided by customers is often poorly structured, making it difficult to secure objective data for accurate work preparation. Customers frequently feel burdened by the task of classifying their belongings or recording quantities, leading them to request a vehicle with only minimal information. This often results in the actual workload growing larger than anticipated or requires additional personnel, causing inconvenience for both the driver and the customer.
[0076] In many cases, sufficient features are not provided to objectively verify work quality even after the move is completed, creating a potential for disputes regarding completion status, missing items, or the state of organization. Existing simple photo verification methods are limited by the difficulty of comparison due to differences in shooting conditions and the reliance on manual human judgment, resulting in a lack of systematic service quality management.
[0077] Meanwhile, video analysis technology utilizing AI models has recently been advancing across various fields, and while basic functions such as object recognition and spatial classification are being implemented at a high level, integrated analysis technology capable of estimating the actual size of items, analyzing spatial geometry, assessing the difficulty of movement routes, and linking vehicle selection and personnel allocation—to provide features specialized for moving services—has not yet been sufficiently secured. AI technology based on simple object detection lacks perspective and tilt correction, making it difficult to calculate physical dimensions; it also lacks the technical foundation to automatically interpret structural constraints of movement routes; furthermore, technology directly linked to the moving service estimate calculation process remains insufficient.
[0078] Therefore, advanced technology is required to calculate the actual volume of items based on user-captured images, automatically analyze spatial structures, quantify the difficulty of movement routes, determine the suitability of vehicles and personnel, and even compare and analyze the condition before and after the move. Such technology lays the foundation for service providers to accurately and efficiently perform dispatching and estimate calculations while reducing the burden of information input on customers, and is an essential element for maintaining stable service quality.
[0079] As such, in order to enhance the accuracy and efficiency of small-scale moving services, there is an urgent need for a new type of technological means that combines image-based automatic analysis technology with moving load size estimation technology; therefore, the development of technology that provides these functions is currently required.
[0081] Furthermore, the present invention aims to provide technical means to resolve the limitations of existing small-scale moving services, where it is difficult to sufficiently determine the actual scale of belongings and movement route conditions based solely on the limited information provided by the user. Items entered directly by the user are prone to omitting item types or quantities, and item sizes are often not accurately specified, frequently resulting in discrepancies between the assigned vehicles or personnel and the actual on-site conditions.
[0082] Therefore, there is a need for technology that can quantitatively determine the spatial structure and item sizes based on photos or videos taken by the user, and automatically calculate the difficulty of the task and the scale of the moving based on this.
[0083] Furthermore, perspective distortion occurs in captured images depending on the shooting position, angle, and tilt, and there is a problem where the overall shape cannot be accurately identified when part of an object is obscured. Since it is difficult to ensure the reliability of image-based analysis without correcting these geometric shooting problems, the present invention focuses on a technical configuration that calculates information on vanishing points, horizons, and camera tilt through geometric analysis of the shooting and incorporates this into perspective correction and size calculation.
[0084] In small-scale moves, not only the volume of goods but also on-site structural conditions, such as narrow sections of the movement path, elevator dimensions, and the effective opening width of doors, are important. It is difficult to accurately determine this information using existing user input methods, and since it is often confirmed only on-site, it leads to work delays or dispatch errors. The present invention aims to fundamentally resolve these problems by automatically determining structural constraints of the indoor space through image analysis and reflecting this information in vehicle selection and personnel allocation decisions.
[0085] Moving services require a clear comparison of the condition before and after the move to minimize disputes between customers and technicians; however, current methods lack objectivity and are difficult to repeat because they require a person to manually compare photos. This invention aims to solve these problems by providing a structure that aligns image data before and after the move into the same coordinate system and automatically analyzes changes in item locations and floor areas to determine whether the work is complete and what items are missing.
[0087] Hereinafter, a method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing utilizing the artificial intelligence model of the present invention will be described with reference to FIGS. 1 to 4. In the following description, some components have been omitted from the drawings to facilitate clear understanding, but these are configurations that can be obviously implemented by those skilled in the art within the technical scope of the present invention in accordance with the claims and the detailed description. That is, the drawings of this specification merely illustrate specific embodiments by way of example, and it is obvious that configurations not described in the drawings are also included within the technical scope of the present invention. Each module is sequentially linked through electrical signals or data packets, and the arrows indicate the data flow.
[0089] FIG. 1 is a diagram showing a method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention; FIG. 2 is a diagram schematically showing a method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention; FIG. 3 is a block diagram showing the configuration of a processor, memory, and network interface included in a small-scale moving platform server of a method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention; and FIG. 4 is a configuration diagram showing a system environment to which a method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention is applied.
[0091] Referring to FIGS. 1 to 4, a method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention is a method performed in a system environment including a small-scale moving platform server connected to each other via a network, a user terminal, and an external driver assignment platform server. The method includes an entire procedure in which the small-scale moving platform server automatically calculates the scale of moving goods using photos or videos of an indoor space, verifies the reliability of a list of items entered by the user, and finally determines the estimated amount, vehicle type, and whether transport personnel are required, thereby automatically processing up to driver assignment.
[0092] In addition, the method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention includes an artificial intelligence model learning procedure that continuously improves the accuracy of an object classification model, a shooting error correction model, and an item scale calculation model by repeatedly utilizing a data set generated in the entire procedure as training data.
[0093] The small-scale moving platform server (10) may be composed of one or more computing devices installed, for example, in a cloud server or data center, and transmits and receives data between a user terminal and an external driver assignment platform server (10) through a network interface. The user terminal may be a smartphone, tablet, laptop, or desktop computer that has a camera function, transmits photos or videos to the small-scale moving platform server (10) via a network, and receives quote results and driver assignment results. The external driver assignment platform server is a server of a driver matching service linked with a small-scale moving service, manages information such as driver location, working hours, vehicle ownership status, and vehicle type for multiple moving driver accounts, and provides driver information in response to a driver lookup request from the small-scale moving platform server (10).
[0094] A small-scale moving platform server (10) includes a processor (11), memory (12), and a network interface (13). The processor (11) is a computing device that loads and executes program modules to perform an image reception step (S10), a captured geometry analysis step (S20), an object volume calculation step (S30), a spatial structure analysis step (S40), a moving goods size calculation step (S50), a user input verification step (S60), a user information correction request step (S70), an estimate amount calculation step (S80), a vehicle and manpower judgment step (S90), a driver list lookup step (S100), a driver assignment notification step (S110), an artificial intelligence model training step (S120), an unstructured item classification and packaging box recommendation step (S130), a pre- and post-moving image comparison and work verification step (S140), a movement path obstacle analysis and movement path correction step (S150), and a site risk factor detection and safety index calculation step (S160).
[0095] Memory (12) is a storage device that stores program modules, step execution parameters, image data, object classification results, spatial structure analysis results, cost calculation formulas, difficulty rating criteria tables, learning data and article information, etc.
[0096] The network interface (13) is responsible for transmitting and receiving data with the user terminal (20) and the technician terminal (30), and performs network-based communication including image data transmission, provision of quotation results, reception of technician information, and notification of technician assignment.
[0097] A small-scale moving platform server (10) stores program modules and parameters corresponding to an image reception step, a shooting geometry analysis step, an object volume calculation step, a spatial structure analysis step, a moving load size calculation step, a user input verification step, a user information correction request step, an estimate amount calculation step, a vehicle and manpower judgment step, a driver list lookup step, a driver assignment notification step, and an artificial intelligence model learning step in a database, and sequentially executes the corresponding program modules using a processor (11) to implement a method for providing an automatic calculation solution for service estimates and sizes for small-scale moving based on image processing using the artificial intelligence model of the present invention.
[0098] The image reception step (S10) is a step in which the small-scale moving platform server (10) determines that a photo or video of an indoor space transmitted from a user terminal owned by a user who intends to use the small-scale moving service is input data that satisfies a preset minimum resolution value and a shooting distance estimation standard, and receives and stores it.
[0099] The small-scale moving platform server (10) verifies whether the horizontal resolution and vertical resolution of a photo or video frame uploaded by a user terminal are greater than or equal to the minimum resolution standard, and calculates whether the shooting distance estimation standard is satisfied by comparing the size of a representative object in the indoor space with the size of a pixel in the image.
[0100] The small-scale moving platform server (10) sends a re-shooting request message to the user terminal for photos or videos that do not satisfy the minimum resolution value and shooting distance estimation criteria, and stores only photos or videos that satisfy the conditions as valid input data for the image reception stage.
[0101] The shooting geometry analysis step (S20) is a step in which the small-scale moving platform server (10) extracts floor line, wall corner line, and ceiling line candidates using a straight line detection algorithm on a photo or video stored in the image reception step, calculates a vanishing point and a horizontal line using the intersection of the extracted straight lines, and calculates the shooting angle, shooting position, and camera tilt information of the photo or video using the calculated vanishing point and horizontal line. Here, the straight line detection algorithm used in the shooting geometry analysis step is an algorithm that extracts edge points by applying a Canny edge detection algorithm to the photo or video, and detects the position and direction of a straight line by applying a Hough transform to the extracted edge points.
[0102] A small-scale moving platform server (10) first applies a Canny edge detection algorithm to extract edge points near the boundaries of the floor, walls, and ceiling, and then performs a Hough transform on the edge points as input to detect a number of straight line candidates for the floor line corresponding to the boundary line where the floor and walls meet, the wall edge line corresponding to the corner part where the walls meet, and the ceiling line corresponding to the boundary line where the walls and ceiling meet.
[0103] In the shooting geometry analysis step (S20), the small-scale moving platform server (10) calculates the distribution of intersection points of multiple straight lines by combining the floor line, wall corner line, and ceiling line candidates detected as described above, determines a vanishing point corresponding to the camera line of sight direction among the intersection point distributions, and calculates a horizontal line from the boundary lines located on the same plane. This operation corresponds to the shooting tilt correction step, and in the shooting tilt correction step, the small-scale moving platform server (10) calculates a shooting tilt correction value using the relative positional relationship between the location of the vanishing point and the image center point, calculates the shooting angle and camera tilt information of the photo or video using the shooting tilt correction value, and estimates the approximate shooting position of the camera. The shooting tilt correction value, shooting angle, and camera tilt information calculated in the shooting tilt correction step are subsequently used as shooting geometry information to correct shooting geometry distortion in the object volume calculation step and the spatial structure analysis step.
[0104] The object volume calculation step (S30) is a step in which the small-scale moving platform server (10) executes an object segmentation model using the photographic geometric information calculated in the photographic geometric analysis step to separate furniture, home appliances, box-shaped items, and irregularly shaped items from the input photo or video into independent objects, extracts the representative edge length, side ratio, and texture-based features of each object, and calculates the height, width, and depth of each object by performing calculations according to the photographic tilt correction formula and the perspective distance correction formula. The object segmentation model may be composed of an artificial intelligence-based semantic segmentation or instance segmentation model, and outputs labels such as floor, wall, ceiling, furniture, home appliance, box-shaped item, and irregularly shaped item through the classification result for each pixel.
[0105] The small-scale moving platform server (10) generates an independent object mask for each object based on these label results and extracts the ratio of the representative corner length and the side from the outer boundary of each object mask. In addition, to extract texture-based features, it analyzes the color change, texture pattern, and repeating structure of the object surface pixels to calculate texture features for each object.
[0106] In the object volume calculation step (S30), if a portion of the outline of an object within a photo or video is not detected by another object or background, the small-scale moving platform server (10) performs an object shape correction step. In the object shape correction step, the small-scale moving platform server (10) searches for a standard shape template to which the object belongs using the detected outline fragments, color information, and texture information, and corrects the size of the object by estimating the shape of the undetected area based on the matching result with the previously stored standard shape template. For example, if only a portion of a bookshelf is photographed and the top or bottom is partially obscured, the small-scale moving platform server (10) analyzes the vertical straight line pattern, shelf spacing, surface texture, etc. of the bookshelf outline and compares them with a standard bookshelf template, estimates the total height and width based on the template matching result, and then calculates the volume by calculating the actual height, width, and depth according to the shooting tilt correction formula and the perspective distance correction formula.
[0107] The spatial structure analysis step (S40) is a step in which the small-scale moving platform server (10) determines the actual scale of the floor surface by comparing the camera position information calculated in the shooting geometry analysis step with reference length candidates such as floor tile specifications or standard specifications for the effective opening width of the door, and calculates the floor surface area of the indoor space, the corridor width, the effective opening width of the door, and the internal dimensions of the elevator using the determined scale.
[0108] A small-scale moving platform server (10) detects the repeating pattern of floor tiles, the door frame, and the standard dimensions of the door frame from multiple photo or video frames to derive a standard length candidate, and calculates a scale value that converts the pixel unit length into an actual distance unit by using the camera position information calculated in the shooting geometry analysis step. Using the determined scale value, the number of pixels in the floor area is converted into an actual floor surface area value, and the pixel width corresponding to the corridor area is calculated as the corridor width, the pixel width corresponding to the door opening area is calculated as the effective door opening width, and the pixel width and height corresponding to the elevator interior area are calculated as the elevator interior dimensions.
[0109] The spatial structure analysis step (S40) may include a movement path analysis step. In the movement path analysis step, the small-scale moving platform server (10) sequentially searches for indoor corridors, entrances, doors, and elevators in the image, calculates the width, height, and effective opening dimensions of each area, and reflects them in the movement difficulty parameter.
[0110] A small-scale moving platform server (10) estimates a path starting from a living room or room location that serves as a starting point in an indoor space, passing through a door and a hallway to an elevator or building entrance, and evaluates the movable path conditions by calculating the hallway width, the effective opening width of the door, and the internal length, width, and height of the elevator for each section of the path.
[0111] The moving goods volume calculation step (S50) is a step in which the small-scale moving platform server (10) calculates the total moving goods volume by summing the volume and quantity of each object calculated in the object volume calculation step, and determines the difficulty grade according to a predefined difficulty grade standard table based on the passage dimensions and the degree of narrowing of the movement path width calculated in the space structure analysis step, and assigns a difficulty value.
[0112] The small-scale moving platform server (10) evaluates whether the corridor width is narrower than the standard, whether the effective opening width of the doorway is close to the maximum width of the item, the margin of the maximum size of the item relative to the internal dimensions of the elevator, the presence or absence of stairs, the height of the threshold, and the number of corners requiring turning, by referring to the difficulty grade standard table stored in the database, and calculates the difficulty grade according to each condition. The difficulty grade standard table may be subdivided into multiple grades, and the small-scale moving platform server (10) stores the difficulty value corresponding to the final difficulty grade as the result of the moving goods size calculation step.
[0113] The user input verification step (S60) is a step in which the small-scale moving platform server (10) determines whether there are any omissions by matching the list of items entered from the user terminal with the list of objects calculated in the object volume calculation step, and verifies the accuracy of the user input by calculating whether the absolute error and ratio error between the estimated volume for each input item and the volume calculated based on actual measurement exceed a preset standard value range.
[0114] The small-scale moving platform server (10) compares the image analysis-based object list and the user input list item by item to determine whether the same type of furniture or home appliance is included in the user input list, and if not, detects the missing items through the missing item detection step. In the missing item detection step, if major furniture or home appliances that significantly affect the size of the moving load, such as a sofa, bed, refrigerator, or washing machine, exist in the image analysis results but not in the user input list, the small-scale moving platform server (10) marks the item as a missing item and generates missing item information to guide to the user terminal.
[0115] In addition, the small-scale moving platform server (10) compares the user input's estimated volume with the volume calculated based on actual measurements to calculate the absolute error and the ratio error, and evaluates the reliability of the user input by determining whether the value is within the standard value range.
[0116] Each threshold value of the minimum resolution value, shooting distance estimation standard, standard value, threshold width, and difficulty grade standard table can be stored and managed as an adjustable parameter in the settings screen or administrator console of the small-scale moving platform server (10), and can be changed from time to time according to service policy and statistical data.
[0117] The user information correction request step (S70) is a step in which the small-scale moving platform server (10) provides an information correction request message containing detailed information of an error item to the user terminal when there is a deviation exceeding a standard value in the user input verification step.
[0118] The small-scale moving platform server (10) generates a user information correction request message containing information such as whether the estimated volume of an item is entered as being excessively smaller or larger than the volume calculated based on actual measurements, whether major furniture or home appliances are missing, and how many missing items are there, and transmits it to a user terminal so that the user can modify and correct the item list. In this way, the user input verification step and the user information correction request step are linked, thereby improving the accuracy and reliability of the user input.
[0119] The estimate amount calculation step (S80) is a step in which the small-scale moving platform server (10) calculates the basic cost, travel distance cost, difficulty weighting cost, packing manpower requirement cost, and vehicle accessibility cost by combining the total moving volume and movement difficulty value calculated in the moving volume calculation step with the travel distance and building structure conditions according to a pre-set cost calculation formula, and generates a small-scale moving estimate amount by summing each calculation result.
[0120] The small-scale moving platform server (10) calculates the basic cost, travel distance cost, and difficulty weighted cost by assigning weights to the total volume, travel distance, and difficulty values using a cost calculation formula stored in the database, and additionally calculates the high-rise work cost, stair transport cost, and vehicle accessibility cost depending on whether the building structure is high-rise, whether only stairs exist, whether entry to an underground parking lot is possible, and whether there is an elevator.
[0121] In addition, the necessity of deploying packing personnel is determined based on the difficulty of packing and the types of belongings, and the required costs are calculated and added together. The process of summing these detailed cost elements can be structured as a detailed cost calculation step, and the sum of each cost value calculated in this step becomes the final small-scale moving estimate amount in the estimate calculation step.
[0122] The cost calculation formula can be implemented as a linear combination formula that multiplies the total moving volume, moving distance, and difficulty values by their respective weights and then sums them up, or as an adjustment formula that assigns non-linear weights to some sections, and can be stored in the form of a table or function in the database of the small-scale moving platform server (10).
[0123] The vehicle and manpower determination step (S90) is a step in which the small-scale moving platform server (10) determines a suitable vehicle among a Damas vehicle, a Labo vehicle, or a 1-ton vehicle based on a vehicle suitability determination criterion that compares the internal dimensions of the vehicle cargo box and the maximum dimensions of the goods calculated in the moving goods size calculation step with the dimensions of the moving path, and determines whether transport manpower is required based on the difficulty of movement and the quantity of goods.
[0124] The small-scale moving platform server (10) determines whether vehicle accessibility is restricted based on the width of the narrowest area of the movement path, and if vehicle accessibility is restricted, performs a vehicle selection step. In the vehicle selection step, the small-scale moving platform server (10) compares whether the width of the narrowest area of the movement path is greater than the passing width for each type of vehicle and whether it is greater than the minimum turning radius required in the section where the vehicle must turn, selects vehicle candidates that can actually enter and exit, and among the candidates, compares the total volume of moving goods and the maximum dimensions of the items with the internal dimensions of the cargo box to finally select one of the Damas vehicle, Labo vehicle, or 1-ton vehicle.
[0125] In addition, the small-scale moving platform server (10) calculates the number of transport personnel required based on the total volume of moving goods, difficulty level, and quantity of items, and determines whether, for example, one basic person or two or more personnel need to be deployed as a result of the vehicle and personnel judgment step.
[0126] The article list lookup step (S100) is a step in which the small-scale moving platform server (10) transmits the result of the vehicle and manpower judgment step to an external article assignment platform server, receives article information including the article location, work availability time, and vehicle ownership status from the external article assignment platform server, and prioritizes selecting an article that satisfies the work start time, travel distance, and vehicle suitability conditions.
[0127] The small-scale moving platform server (10) transmits a driver search request to an external driver assignment platform server, which includes information such as the scheduled moving date and time, the work location, the type of vehicle required, and the number of personnel required, and receives driver information including driver location information, available working hours, and whether the driver possesses a vehicle. Subsequently, the small-scale moving platform server (10) calculates the travel distance between the driver's location and the work location, checks whether the desired work time entered by the user overlaps with the range of available working hours of the driver, selects candidate drivers, and determines the optimal driver by prioritizing the selection of a driver who actually possesses the type of vehicle determined in the vehicle and personnel judgment step among the candidate drivers.
[0128] The driver assignment notification step (S110) is a step in which the small-scale moving platform server (10) transmits the driver selected in the driver list lookup step to the user terminal to provide an assignment result. The small-scale moving platform server (10) generates assignment result information including the name of the selected driver, vehicle type, estimated arrival time, work start time, quoted amount, and number of transport personnel, and transmits it to the user terminal so that the user can check the assignment result.
[0129] Additionally, the driver assignment notification step (S110) may include a schedule verification step. In the schedule verification step, the small-scale moving platform server (10) determines again whether there is a conflict between the available time of the selected driver and the user's desired time. If a conflict exists, it compares the driver's available time range with the user's requested time range to select another driver whose time range overlaps, and notifies the user terminal of the modified assignment result by reflecting the information of the re-selected driver.
[0130] The artificial intelligence model training step (S120) is a step in which the small-scale moving platform server (10) stores the data set generated in the image reception step, the shooting geometry analysis step, the object volume calculation step, the spatial structure analysis step, the moving goods size calculation step, the user input verification step, the estimate amount calculation step, and the article list lookup step as training data, and performs a parameter update procedure configured to periodically update the artificial intelligence model that can be trained using the training data, thereby improving the image analysis accuracy and the estimate calculation accuracy.
[0131] The small-scale moving platform server (10) stores image data, object segmentation results, shooting geometric information, actual movement results per object, item volume and movement difficulty confirmed after actual moving completion, actual billed costs and driver assignment results in a learning database, and periodically updates the parameters of the object classification model, the shooting error correction model, and the item size calculation model using this learning database.
[0132] In addition, the small-scale moving platform server (10) classifies shooting error cases by type, such as low image quality, backlight, shaking, and poor shooting angle, and learns a correction pattern corresponding to each shooting error type, so that when the same error pattern is detected, it generates corrected shooting geometric information in the shooting geometric analysis stage, thereby enabling more accurate image analysis.
[0133] The step of classifying irregular items and recommending packaging boxes (S130) is a step in which the small-scale moving platform server (10) classifies the item type for irregularly shaped items other than standardized furniture or home appliances among the objects separated in the object volume calculation step using contour complexity, surface texture feature quantity, shape asymmetry index, and texture-based feature quantity, and automatically calculates the size and quantity of packaging boxes and recommended packaging methods by referring to a standard packaging specification database stored for each classified type.
[0134] The small-scale moving platform server (10) analyzes the contour complexity regarding how irregular the outline of an irregularly shaped item is, surface texture features such as surface roughness and pattern, shape asymmetry index indicating the degree of left-right or up-down symmetry, and texture-based features together to classify the irregularly shaped item into multiple types, such as fragile items, long items, and items with many curved surfaces. Then, for each classified type, it selects a packaging box size by referring to a standard packaging specification database and automatically calculates specific recommended packaging methods, such as the number of boxes to package one item, whether cushioning material is used, whether double packaging is necessary, and the direction of item placement.
[0135] The step of comparing images before and after moving and verifying the work (S140) is a step in which a small-scale moving platform server (10) receives a photo or video of the moving after the moving is completed from a user terminal or a technician terminal, aligns the photo or video of the moving before moving and the photo or video of the moving after moving into the same coordinate system, and then determines whether the change in the location of an item is within an allowable range, whether the floor area has increased or decreased compared to the initial arrangement, and whether the spacing between item objects has changed compared to the initial state by using the presence of items, the change in location, the change in pixel area classified as a floor area in the image before moving and the image after moving, and the change in the relative spacing between item objects, and determines whether the work is completed and whether there are missing items.
[0136] A small-scale moving platform server (10) extracts unchanging reference points, such as wall corner lines, floor lines, and window positions, from images before and after moving to align the coordinate systems of the two images, and applies an object segmentation model on the same coordinate system to compare the location and area of each item object. In this process, if a specific item object existed before moving but does not exist after moving, it is determined to be a missing item. It can also determine whether the item object is organized by comprehensively evaluating whether the location of the item object has been moved appropriately within the allowed range, whether the arrangement has been excessively changed and is in a messy state, and whether the organized state has improved by increasing the pixel area classified as the floor area.
[0137] The step of analyzing movement path obstacles and correcting the movement path (S150) is a step in which a small-scale moving platform server (10) generates a movable path as a virtual movement trajectory using floor area information, corridor width information, effective door opening width information, and elevator dimension information calculated in the space structure analysis step, detects the presence of obstacles on the generated trajectory, detects narrow sections consisting of sections with insufficient curb height and turning radius, and sections where the width is smaller than a preset threshold width, and calculates a movement difficulty value by comparing the height, width, and turning radius values, which indicate the degree to which each obstacle restricts the movement path, with the maximum dimensions of the item and the clearance width within the movement path.
[0138] A small-scale moving platform server (10) sets one or more virtual movement paths starting from a living room or room, passing through a door and a hallway, and proceeds along each virtual movement path, analyzing factors such as the presence or absence of obstacles, floor threshold height, turning radius required at corners, and hallway width. At this time, if the hallway width, the effective opening width of the door, or the width of the elevator entrance is smaller than the maximum dimensions of the goods, or if there is a narrow section smaller than the critical width, the small-scale moving platform server (10) marks the section as a section with high difficulty of movement, and assigns a high difficulty value if there are consecutive stairs or high thresholds. The movement difficulty value calculated in this movement path obstacle analysis and movement path correction step can be combined with the difficulty value assigned in the moving goods size calculation step to be used for a more precise evaluation of the difficulty of movement.
[0139] As such, the method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model according to an embodiment of the present invention can significantly improve the efficiency and accuracy of small-scale moving services by having a small-scale moving platform server (10) automatically identify the structure of the indoor space and the scale of the moving goods based on photos or videos, verify the accuracy of user input, calculate an estimate amount reflecting the difficulty of movement, the distance traveled, and the building structure, determine a suitable vehicle and transport personnel through vehicle suitability judgment criteria and a vehicle selection step, and then automatically assign an optimal driver by considering the driver's location, working hours, and vehicle ownership status in the driver list lookup step and driver assignment notification step.
[0141] A method for providing an automatic calculation solution for service estimates and scale for small-scale moving based on image processing using an artificial intelligence model according to another embodiment of the present invention may further include the following on-site risk factor detection and safety index calculation step (S160) in addition to the previously described image reception step (S10), shooting geometry analysis step (S20), object volume calculation step (S30), spatial structure analysis step (S40), moving goods scale calculation step (S50), user input verification step (S60), user information correction request step (S70), estimate amount calculation step (S80), vehicle and manpower judgment step (S90), driver list lookup step (S100), driver assignment notification step (S110), artificial intelligence model training step (S120), non-standard item classification and packaging box recommendation step (S130), before and after moving image comparison and work verification step (S140), and movement path obstacle factor analysis and movement path correction step (S50).
[0142] The on-site risk factor detection and safety index calculation step (S160) is a step in which the small-scale moving platform server (10) applies an artificial intelligence-based risk factor detection model to the indoor environment image, corridor area image, door area image, and elevator interior image obtained in the image reception step and spatial structure analysis step to detect risk factors with a high probability of safety accidents occurring during work, and calculates a safety index by synthesizing the type, location of occurrence, and risk value of the detected risk factors.
[0143] A small-scale moving platform server (10) runs a risk detection model to detect risk factors included in an indoor image, such as floor slip risk factors (water, shine, lubrication traces), obstacles (small objects on the floor, cables, small boxes), damaged or cracked flooring, slopes, damaged door frames, protruding metal parts, low light areas, and unsecured furniture. The risk detection model may be composed of an object detection model or a semantic segmentation model that includes risk labels, and identifies risk factors within the image on a pixel-by-pixel or object-by-object basis.
[0144] The small-scale moving platform server (10) calculates a risk score by considering the risk weights for each type of risk factor (e.g., slippery 0.6, obstacle 0.4, poor lighting 0.5, slope 0.7, etc.) among the risk factor detection results, and calculates an overall safety index by applying the importance weights of the space where the risk factor is located (corridor, door threshold, elevator entrance, indoor corner, etc.). The safety index can be expressed as a value from 0 to 1, for example, and the higher the value, the higher the risk.
[0145] The small-scale moving platform server (10) can generate a final difficulty correction value by combining the calculated safety index with the difficulty value of the moving load size calculation stage, and reflect it in the difficulty-weighted cost during the estimate amount calculation stage so that costs for securing a safer working environment are included. Additionally, if the safety index is above a certain standard, it can provide a risk factor guidance message to the user terminal or control the system to prioritize recommending experienced drivers who can consider the risk factor during the driver list lookup stage.
[0146] Thus, the on-site risk factor detection and safety index calculation step according to another embodiment of the present invention provides a technical effect that enables more accurate cost estimation and enhanced safety by analyzing potential safety accident factors that may occur at the actual site in advance, going beyond a simple assessment of moving load size and difficulty of movement.
[0148] As described above, according to the present invention, the user can systematically identify the total scale of moving belongings and spatial structural constraints simply by taking a photo or video of the indoor space, thereby significantly reducing the burden of complex information input that was previously required in the process of booking a small-scale move. While the process of the user manually listing items one by one or finding and inputting specifications is cumbersome and prone to errors, the present invention can generate objective data by automatically analyzing the type, arrangement, shape, and size of items from the captured image, thereby greatly reducing the intensity of customer participation and naturally improving the quality of input information.
[0149] The image processing procedure of the present invention, which combines photographic geometric analysis, perspective correction, and texture-based feature analysis, can calculate the actual physical dimensions of an object after correcting distortion factors appearing in the image; thus, it enables the calculation of accurate specifications that were difficult to achieve with simple object recognition technology. This function reflects not only the actual size of furniture or home appliances constituting the moving shipment but also the morphological characteristics of box-shaped or irregular items, allowing for the calculation of values close to the actual volume and significantly improving the precision of moving estimates.
[0150] Furthermore, the present invention not only calculates the volume of the item but also automatically analyzes structural constraints of the indoor space, thereby enabling a more precise assessment of the difficulty of the entire movement path. By analyzing elements such as narrow corridor widths, effective door opening widths, and elevator interior dimensions from captured images, it is possible to determine in advance whether the item can actually pass through the path. This function plays a crucial role in the technician assignment stage. In existing services, it was common to discover belatedly on-site that the item could not pass or required disassembly; however, the present invention prevents such prediction failures in advance, thereby reducing factors that cause delays in technician schedules or conflicts with customers.
[0151] The vehicle and personnel assignment judgment function of the present invention determines the appropriate vehicle type and personnel size by synthesizing information on item size calculated based on images, the total volume of moving goods, and obstacles along the movement path. This enables much more accurate and rational dispatching compared to existing experience-dependent judgment methods. By automatically determining whether a vehicle can pass through a route and selecting a vehicle while considering items requiring disassembly or sections with insufficient turning radii, it provides predictable schedules and a stable service experience for both drivers and customers. This significantly contributes to the efficiency of driver schedules and helps them safely complete a greater number of assignments within the same daily workload.
[0152] The present invention includes a function that can verify the accuracy of input data by automatically comparing user input information with image analysis results, thereby effectively resolving the problem of missing information that commonly occurs in existing services. Since it is difficult for users to identify every item in their home individually, they often fail to input important items or underestimate their volume. The present invention automatically recognizes the type of item during the image analysis stage and cross-verifies user input information; by immediately notifying the user if there are any missing items or incorrect information, it reduces situations where both the customer and the technician experience difficulties due to unexpected variables.
[0153] Furthermore, the present invention possesses the technical advantage of objectively determining whether a task has been completed, as it can automatically verify changes in item positions, floor surface exposure, and spacing between items by aligning pre- and post-moving images into the same coordinate system. Unlike conventional subjective photo comparison methods, this judgment is based on automatically calculated quantitative results, significantly reducing the possibility of disputes and providing platform service providers with a foundation to technically establish a quality assurance system.
[0154] The artificial intelligence model learning function of the present invention has a structure that accumulates data sets generated during the image analysis process as training data to iteratively improve the model's performance. Therefore, as the application period of the present invention lengthens, product classification accuracy, volume calculation precision, movement path analysis accuracy, and vehicle selection prediction performance naturally become more advanced. This provides the effect of continuously and stably growing the technical foundation of the platform service and serves as a basis for securing a long-term technological advantage over competitors.
[0155] As such, the present invention is highly useful in that it can comprehensively achieve, within a single system, a reduction in the burden of entering customer information, precise analysis based on captured images, automatic determination of narrow sections in the movement path, improved accuracy in assigning vehicles and personnel, automation of comparing pre- and post-moving conditions, reduction of disputes, maximization of operational efficiency, assurance of service quality, and long-term improvement of AI performance. Consequently, the present invention significantly enhances the quality and reliability of the entire process of small-scale moving services, provides a stable and predictable moving experience for both customers and drivers, and offers platform service providers the effect of simultaneously achieving operational efficiency and strengthening technological competitiveness.
[0157] Although embodiments of the present invention have been described in more detail with reference to the attached drawings, the present invention is not necessarily limited to these embodiments and may be modified in various ways within the scope of the technical spirit of the present invention. Accordingly, the embodiments disclosed in the present invention are intended to explain, not limit, the technical spirit of the present invention, and the scope of the technical spirit of the present invention is not limited by these embodiments. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. The scope of protection of the present invention shall be interpreted by the claims below, and all technical spirits within an equivalent scope shall be interpreted as being included within the scope of rights of the present invention.
[0158] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.
Claims
Claim 1 A method for estimating a small-scale move and automatically calculating the size of moving goods using an artificial intelligence model-based image analysis comprises: an image reception step in which a small-scale moving platform server determines a photo or video of an indoor space transmitted from a user terminal as input data satisfying a preset minimum resolution value and shooting distance estimation criteria, and receives and stores it; a shooting geometry analysis step in which the small-scale moving platform server extracts candidates for floor lines, wall corner lines, and ceiling lines using a straight line detection algorithm on the photo or video stored in the image reception step, calculates vanishing points and horizontal lines using the intersection points of the extracted straight lines, and calculates shooting angles, shooting positions, and camera tilt information of the photo or video using the calculated vanishing points and horizontal lines; and an object volume calculation step in which the small-scale moving platform server executes an object segmentation model using the shooting geometry information calculated in the shooting geometry analysis step to separate furniture, home appliances, box-shaped items, and irregularly shaped items from the input photo or video into independent objects, extracts representative corner lengths, side ratios, and texture-based features of each object, and calculates the height, width, and depth of each object by performing operations according to a shooting tilt correction formula and a perspective distance correction formula. A spatial structure analysis step in which the above-mentioned small-scale moving platform server determines the actual scale of the floor surface by comparing the camera position information calculated in the above-mentioned shooting geometric analysis step with reference length candidates, such as floor tile specifications or standard specifications for the effective opening width of a door, and calculates the floor surface area of the indoor space, the corridor width, the effective opening width of the door, and the internal dimensions of the elevator using the determined scale;Moving load size calculation step in which the small-scale moving platform server calculates the total moving load volume by summing the volumes and quantities of each object calculated in the object volume calculation step, and determines a difficulty grade according to a difficulty grade standard table stored in the small-scale moving platform server based on the passage dimensions and the degree of narrowing of the movement path width calculated in the spatial structure analysis step, and assigns a difficulty value; user input verification step in which the small-scale moving platform server corresponds the item list entered from the user terminal with the object list calculated in the object volume calculation step to determine whether there are any omissions, and verifies the accuracy of the user input by calculating whether the absolute error and ratio error between the estimated volume for each input item and the volume calculated based on actual measurement exceed a preset standard value range; and user information correction request step in which the small-scale moving platform server provides an information correction request message including detailed information on the error item to the user terminal if a deviation exceeding the standard value exists in the user input verification step. A quotation amount calculation step in which the small-scale moving platform server combines the total volume of moving goods and the movement difficulty value calculated in the moving goods size calculation step with the movement distance and building structure conditions, calculates the basic cost, movement distance cost, difficulty weighting cost, packing personnel requirement cost, and vehicle accessibility cost according to the cost calculation formula stored in the small-scale moving platform server, and sums the respective calculation results to generate a small-scale moving quotation amount; a vehicle and personnel judgment step in which the small-scale moving platform server determines one of a Damas vehicle, a Labo vehicle, or a 1-ton vehicle based on a vehicle suitability judgment criterion that compares the total volume and maximum dimensions of goods calculated in the moving goods size calculation step with the movement path dimensions, and compares the internal dimensions of the vehicle cargo compartment with the maximum dimensions of the goods, and determines whether transport personnel are required based on the movement difficulty and the quantity of goods;A small-scale moving platform server transmits the result of the vehicle and manpower determination step to an external driver assignment platform server, receives driver information including the driver's location, available working time, and vehicle ownership status from the external driver assignment platform server, and prioritizes selecting a driver who satisfies the conditions for available work time, travel distance, and vehicle suitability; and a driver assignment notification step in which the small-scale moving platform server transmits the driver selected in the driver list lookup step to the user terminal to provide an assignment result. A method for providing an automatic calculation solution for service estimates and scales based on image processing utilizing an artificial intelligence model, comprising: an AI model training step in which the small-scale moving platform server stores data sets generated in the image reception step, the shooting geometric analysis step, the object volume calculation step, the spatial structure analysis step, the moving goods scale calculation step, the user input verification step, the estimate amount calculation step, and the driver list lookup step as training data, and performs a parameter update procedure configured to periodically update an AI model trainable using the training data, thereby improving image analysis accuracy and estimate calculation accuracy; wherein the line detection algorithm used in the shooting geometric analysis step applies a Canny edge detection algorithm to a photograph or video to extract edge points, and applies a Hough transform to the extracted edge points to detect the position and direction of a straight line. Claim 2 In claim 1, the shooting geometric analysis step includes a shooting tilt correction step that extracts wall boundary lines, floor boundary lines, and ceiling boundary lines using the line detection algorithm, determines a vanishing point from the distribution of intersection points of the extracted multiple lines, and calculates a shooting tilt correction value using the relative positional relationship between the location of the vanishing point and the image center point; the object volume calculation step includes an object shape correction step that, when a part of the outline of an object in a photograph or video is not detected by another object or background, searches for a standard shape template to which the object belongs using the detected outline fragments, color information, and texture information, and estimates the shape of the undetected area based on the template matching result to correct the size of the object; the spatial structure analysis step includes a movement path analysis step that sequentially searches for indoor corridors, entrances, doors, and elevators on the image, calculates the width, height, and effective opening dimensions of each area, and reflects them in the movement difficulty parameter; the user input verification step includes a missing item detection step that compares an image analysis-based object list with a user input list and provides the missing item to the user terminal if major furniture or home appliances are missing; and the estimated amount calculation step includes a basic travel distance It includes a detailed cost calculation step for calculating costs, high-rise work costs, stair transport costs, and narrow corridor work costs respectively, and summing the calculated cost values to calculate the final estimated amount; the vehicle and personnel judgment step includes a vehicle selection step for determining whether vehicle accessibility is restricted based on the width of the narrowest area in the movement path, and if vehicle accessibility is restricted, selecting a vehicle by comparing whether the width of the narrowest area is greater than or equal to the width of a vehicle's passable width and greater than or equal to the minimum turning radius for a vehicle to turn; and the driver assignment notification step includes determining whether there is a conflict between available work hours and the user's desired time, and if a conflict exists,A method for providing an automatic calculation solution for service estimates and scales for small-scale moving based on image processing using an artificial intelligence model, comprising a schedule verification step that compares the time range of an article's available move with the time range requested by a user and re-selects an article whose time ranges overlap. Claim 3 A method for providing an automatic calculation solution for service estimates and scales based on image processing using an artificial intelligence model, wherein, in claim 1, the small-scale moving platform server classifies the type of an irregularly shaped item other than standardized furniture or home appliances among the objects separated in the object volume calculation step using contour complexity, surface texture feature quantity, shape asymmetry index, and texture-based feature quantity, and automatically calculates the size and quantity of packaging boxes and recommended packaging methods by referring to a previously stored standard packaging specification database for each classified type.