A method and system for monitoring an interior decoration process

Abstract

This invention discloses a monitoring method and system for interior decoration processes, relating to the fields of computer vision and construction monitoring technology. The method includes: identifying construction objects through video analysis and generating a sequence of semantic descriptors; constructing a temporal dependency graph representing the temporal dependencies of the construction objects; calculating the node mapping coverage and edge connectivity preservation of the temporal dependency graph and a preset process flow diagram to identify the current construction stage; propagating construction specification constraints along the edges of the temporal dependency graph and adjusting the spatial constraint range according to the temporal dependencies to generate an extended constraint set; inferring the complete spatial location for construction objects with incomplete spatial locations; matching the updated semantic descriptor sequence with the extended constraint set to calculate the deviation, generating anomaly judgment results, and outputting monitoring instructions.

Description

Technical Field

[0001] This invention relates to the fields of computer vision and construction monitoring technology, specifically to a monitoring method and system for the interior decoration process. Background Technology

[0002] Interior decoration construction involves multiple processes, each with complex temporal dependencies and spatial constraints. Traditional construction monitoring methods rely primarily on manual on-site inspections, which suffer from low monitoring efficiency, limited coverage, and difficulty in real-time anomaly detection. Furthermore, these methods typically identify and judge construction objects as independent entities, failing to fully consider the temporal dependencies between them. In decoration construction, different processes have strict sequential orders; for example, electrical and plumbing wiring must be completed before wall sealing, and waterproofing must be completed before tiling. The construction quality and spatial layout of the current process directly affect the constraints of subsequent processes, but current technologies struggle to establish such dynamic relationships.

[0003] Existing technologies typically rely on static construction specifications for judgment, failing to dynamically adjust the constraint range based on actual construction progress and procedural dependencies. For example, the installation location constraint of a wall socket depends not only on static construction specifications but also on factors such as the location of completed conduit wiring, wall structure, and the space available for adjacent construction processes. Existing technologies struggle to achieve dynamic propagation and adjustment of constraints. Furthermore, when processing construction objects that are obscured or partially visible in video footage, incomplete information is often discarded or marked as unidentified, failing to fully utilize temporal dependencies and the spatial location information of identified objects for inference and completion, resulting in blind spots in monitoring.

[0004] Therefore, there is a need for an indoor decoration construction monitoring method that can establish temporal dependencies between construction objects, realize dynamic propagation and adjustment of constraints, and infer and complete incomplete spatial location information. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for monitoring the interior decoration process, aiming to at least solve one of the technical problems existing in the prior art.

[0006] The technical solution of this invention is: a method for monitoring the interior decoration process, comprising the following steps: The video data stream of the indoor decoration work area is acquired, and the construction objects are identified through temporal segmentation and semantic segmentation. The category identifier, spatial location and time identifier are extracted to generate a semantic descriptor sequence. A temporal dependency graph is constructed based on a sequence of semantic descriptors, where nodes represent construction objects and edges represent the temporal dependencies between construction objects. The current construction stage is identified by calculating the node mapping coverage and edge connectivity preservation of the temporal dependency graph and the preset process flow diagram, and the stage identifier is output. Extract construction specification constraints based on stage identifiers, propagate these constraints along the edges of the temporal dependency graph, adjust the spatial constraint range based on temporal dependencies, and generate an extended constraint set. Analyze the category identifiers and spatial locations of construction objects in the semantic descriptor sequence, infer the complete spatial locations for construction objects marked with incomplete spatial locations, and update the semantic descriptor sequence; The updated semantic descriptor sequence is matched with the extended constraint set, the deviation of the spatial location from the spatial constraint range is calculated, the spatial compliance is generated, the anomaly judgment result is generated based on the spatial compliance, and the monitoring instruction is output.

[0007] Construction objects are identified through temporal and semantic segmentation, and a sequence of semantic descriptors is generated by extracting category identifiers, spatial locations, and temporal identifiers, including: The video data stream is parsed into a sequence of frames arranged in chronological order. The frame sequence is segmented temporally to identify changes in the motion state of construction objects and scene content between consecutive frames, determine the start and end frames of the operation, and segment the frame sequence into multiple operation segments based on the start and end frames. Semantic segmentation is performed on frames in the work segment to identify construction objects, and pixel regions of construction objects are extracted and labeled with category identifiers of construction objects. Calculate the boundary range of the pixel region and convert the boundary range into the spatial location of the construction object; Detect the occlusion situation in the pixel area of ​​the construction object and count the number of occluded pixels. Calculate the occlusion ratio of the number of occluded pixels to the total number of pixels in the pixel area. Mark the spatial position of the construction object as a complete spatial position or an incomplete spatial position according to the occlusion ratio. Generate a sequence of semantic descriptors by identifying the category, spatial location, and time of the construction object.

[0008] Constructing a temporal dependency graph based on a sequence of semantic descriptors includes: Based on the time identifier of the construction object in the semantic descriptor sequence, the construction object is mapped to the node of the temporal dependency graph, and the spatial location of the construction object corresponding to the temporally adjacent node is extracted. Calculate the horizontal projection area of ​​the spatial positions of the construction object corresponding to the previous node and the construction object corresponding to the subsequent node, and identify the temporal dependency relationship type between the construction objects by the overlap state of the horizontal projection area; When the horizontal projection areas overlap, the vertical height difference is extracted and the temporal dependency type is marked as a bearing dependency. When the horizontal projection areas do not overlap, the horizontal spacing value is extracted and the temporal dependency type is marked as an adjacency dependency. A vertical constraint propagation path is constructed based on the vertical height difference of the bearing dependency, and a horizontal constraint isolation region is constructed based on the horizontal spacing of the adjacency dependency. In the temporal dependency graph, edges are created from the previous node to the next node. The temporal dependency type, vertical constraint propagation path, or horizontal constraint isolation region are stored as edge attributes in the edges to complete the construction of the temporal dependency graph.

[0009] The current construction stage is identified by calculating the node mapping coverage and edge connectivity preservation between the temporal dependency graph and the preset process flow diagram. The output stage identifier includes: Extract the category identifiers and spatial location markers of nodes in the temporal dependency graph, and extract the standard node category set and standard edge connection relationship set corresponding to each construction stage in the preset process flow diagram; For nodes whose spatial location is marked as incomplete spatial location, the corresponding construction stage is searched in the standard edge connection relationship set as a candidate construction stage based on the category identifier of the starting node of the edge pointing to the node with incomplete spatial location in the temporal dependency graph. Calculations are performed for each construction stage in the preset process flow diagram, and the number of matching nodes is obtained by matching the category identifier of the complete spatial location node with the standard node category set. Incomplete spatial location nodes that are the same as the candidate construction stage and the current construction stage are included in the number of matching nodes. The proportion of the number of matching nodes to the total number of nodes in the standard node category set is calculated as the node mapping coverage. Constraint propagation is performed along the edges based on the edge attributes of the temporal dependency graph. The number of edge connections covered by the propagation that match the standard edge connection set is counted. The proportion of the number of matches to the total number of edge connections in the standard edge connection set is calculated as the edge connection preservation degree. The comprehensive matching degree of each construction stage is calculated based on the node mapping coverage and edge connectivity preservation degree. The construction stage with the highest comprehensive matching degree is selected as the current construction stage and the stage identifier is output.

[0010] Construction specification constraints are extracted based on stage identifiers, propagated along the edges of the temporal dependency graph, and the spatial constraint range is adjusted according to the temporal dependencies to generate an extended constraint set including: Based on the stage identifier, extract the construction specification constraints corresponding to the current construction stage from the preset construction specification library. The construction specification constraints include the spatial constraint range of each type of construction object. Extract the edge attributes of each edge in the temporal dependency graph and identify the types of temporal dependencies in the edge attributes; For edges with a bearing dependency relationship, extract the vertical constraint propagation path and vertical height difference from the edge attributes, propagate the spatial constraint range corresponding to the starting node along the vertical constraint propagation path to the ending node, and superimpose the vertical height difference onto the boundary coordinates of the spatial constraint range corresponding to the ending node in the vertical direction to generate the vertically adjusted spatial constraint range. For edges with adjacency dependency relationships, extract the horizontal constraint isolation region and horizontal spacing value from the edge attributes. Based on the horizontal spacing value, adjust the horizontal spacing between the spatial constraint ranges corresponding to the start node and the end node to generate the horizontally adjusted spatial constraint range. For cases where multiple edges point to the same termination node, the adjusted spatial constraint ranges propagated from each edge to the termination node are extracted, and the intersection of the adjusted spatial constraint ranges is calculated as the spatial constraint range corresponding to the termination node. The adjusted spatial constraint ranges for each node are aggregated to generate an extended constraint set.

[0011] Analyzing the category identifiers and spatial locations of construction objects in the semantic descriptor sequence, inferring the complete spatial locations for construction objects marked with incomplete spatial locations, and updating the semantic descriptor sequence includes: Extract construction objects whose spatial location is marked as incomplete spatial location from the semantic descriptor sequence, and extract each incoming edge of the node corresponding to the construction object with incomplete spatial location in the temporal dependency graph; Extract the category identifier and spatial position of the construction object corresponding to the starting node of each edge, obtain the corresponding standard size information based on the category identifier, and superimpose the standard size information onto the spatial position of the construction object corresponding to the starting node to generate the spatial boundary range of the construction object corresponding to the starting node. Extract the temporal dependency type from the edge attributes of each edge. For the carrying dependency type, extract the upper boundary of the spatial boundary in the vertical direction as the vertical inference boundary. For the adjacency dependency type, extract the side boundary of the spatial boundary in the horizontal direction as the horizontal inference boundary. Obtain the standard dimension information corresponding to the construction object with incomplete spatial location, and perform spatial matching between the standard dimension information and the vertical and horizontal reasoning boundaries to calculate the complete spatial location of the construction object with incomplete spatial location. Update the complete spatial location of the construction object whose spatial location is incomplete to the semantic descriptor sequence.

[0012] The updated semantic descriptor sequence is matched with the extended constraint set to calculate the deviation of spatial location from the spatial constraint range, generate spatial compliance, and generate anomaly judgment results based on the spatial compliance, and output monitoring instructions including: Extract the complete spatial location of each construction object from the updated semantic descriptor sequence, and extract the corresponding spatial constraint range from the extended constraint set; In the time-dependent graph, identify the set of nodes that form a closed-loop path. For the construction objects in the closed-loop path, calculate the deviation distance between the complete spatial location and the spatial constraint range. Accumulate the propagation deviation distance along the closed-loop path to obtain the closed-loop cumulative deviation value. The closed-loop deviation amplification factor is calculated based on the cumulative deviation value and the closed-loop path length. For nodes in the closed-loop path, the deviation amplification factor is multiplied by the deviation distance to obtain the deviation degree. For nodes in the non-closed-loop path, the deviation distance is used as the deviation degree. Based on the deviation, spatial compliance is generated; based on the spatial compliance, anomaly judgment results are generated; based on the anomaly judgment results, construction objects in the closed-loop path where the closed-loop deviation amplification coefficient exceeds the preset amplification threshold are extracted; monitoring instructions are generated and output.

[0013] This invention provides a monitoring system for the interior decoration process, the system comprising: The semantic recognition module is used to acquire video data streams of the indoor decoration work area, identify construction objects through temporal segmentation and semantic segmentation, and extract category identifiers, spatial locations and time identifiers to generate a sequence of semantic descriptors; The dependency graph construction module is used to construct a temporal dependency graph based on a sequence of semantic descriptors, where nodes represent construction objects and edges represent the temporal dependencies between construction objects. The stage identification module is used to identify the current construction stage by calculating the node mapping coverage and edge connectivity preservation degree between the time-series dependency graph and the preset process flow diagram, and output the stage identifier. The constraint generation module is used to extract construction specification constraints based on stage identifiers, propagate the construction specification constraints along the edges of the temporal dependency graph, adjust the spatial constraint range according to the temporal dependency relationship, and generate an extended constraint set. The location reasoning module is used to analyze the category identifier and spatial location of construction objects in the semantic descriptor sequence, and to reason about the complete spatial location of construction objects marked with incomplete spatial locations and update the semantic descriptor sequence. The anomaly monitoring module is used to match the updated semantic descriptor sequence with the extended constraint set, calculate the deviation degree of spatial location from the spatial constraint range, generate spatial compliance degree, generate anomaly judgment results based on spatial compliance degree, and output monitoring instructions.

[0014] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0015] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps in any of the aforementioned methods.

[0016] This invention establishes temporal relationships between construction objects by constructing a temporal dependency graph, enabling dynamic propagation of construction specification constraints and adaptive adjustment of spatial constraint ranges. This solves the problem of missing process dependencies caused by independent judgment of construction objects in existing technologies. By using temporal dependencies to infer incomplete spatial locations, it reduces monitoring blind spots caused by video occlusion, improving the comprehensiveness and accuracy of construction monitoring. Through the constraint propagation mechanism along the edges of the temporal dependency graph, it achieves dynamic influence modeling of the spatial constraints of preceding processes on subsequent processes, making constraint judgments more consistent with actual construction scenarios. By calculating deviation and spatial conformity, it quantifies the degree of deviation between construction objects and specification constraints, providing an interpretable basis for anomaly detection. This facilitates construction managers in quickly locating problems and taking targeted measures, improving the intelligence level of construction quality control. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a method for monitoring an interior decoration process, provided as an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of an indoor decoration process monitoring system according to an embodiment of the present invention. Detailed Implementation

[0018] like Figure 1 As shown, Figure 1 A flowchart of a method for monitoring an interior decoration process provided by an embodiment of the present invention, the method comprising the following steps: Step 101: Obtain video data stream of the indoor decoration work area, identify construction objects through temporal segmentation and semantic segmentation, and extract category identifiers, spatial location and time identifiers to generate a semantic descriptor sequence.

[0019] In some embodiments of the present invention, step 101 may specifically include the following sub-steps: Sub-step 1011: Parse the video data stream into a sequence of frames arranged in chronological order; Sub-step 1012: Perform temporal segmentation on the frame sequence, identify changes in the motion state of construction objects and changes in scene content between consecutive frames, determine the start and end frames of the operation, and segment the frame sequence into multiple operation segments based on the start and end frames. Sub-step 1013: Semantic segmentation is performed on the frames in the work segment to identify construction objects, extract the pixel regions of the construction objects, and label the category identifiers of the construction objects; Sub-step 1014: Calculate the boundary range of the pixel region and convert the boundary range into the spatial location of the construction object; Sub-step 1015: Detect the occlusion situation in the pixel area of ​​the construction object and count the number of occluded pixels. Calculate the occlusion ratio of the number of occluded pixels to the total number of pixels in the pixel area. Mark the spatial position of the construction object as a complete spatial position or an incomplete spatial position according to the occlusion ratio. Sub-step 1016 generates a semantic descriptor sequence from the category identifier, marked spatial location, and time identifier of the construction object.

[0020] During indoor renovation monitoring, after acquiring the video data stream of the work area, the video data stream is parsed into discrete image frames in chronological order. Each image frame carries timestamp information as its time position identifier within the entire video sequence. The parsing process is performed according to the frame rate parameter of the video encoding format. If the video frame rate is 30fps, then the video data stream per second is parsed into 30 independent image frames, and the time identifier of each image frame is accurate to the millisecond level.

[0021] When performing temporal segmentation on the parsed frame sequence, changes in the motion state of the construction object are identified by calculating image differences between consecutive frames. Specifically, this involves extracting image features from the current frame and the previous frame, calculating the difference in grayscale values ​​or color features at corresponding pixel positions between the two frames, and determining a change in motion state when the difference exceeds a set threshold. Simultaneously, changes in scene content are monitored, and the boundaries of the work actions are determined by analyzing the appearance and disappearance of construction tools, materials, or workers in consecutive frames. When the difference value of multiple consecutive frames continuously increases and then stabilizes, the frame where the difference value begins to increase is marked as the starting frame of the work action, and the frame where the difference value stabilizes is marked as the ending frame. Based on the marked starting and ending frames, the complete frame sequence is segmented into multiple independent work segments, each corresponding to a complete construction operation process, such as painting a wall, installing pipelines, or laying flooring.

[0022] Semantic segmentation is performed on image frames within each work segment to identify construction objects within the frame. The semantic segmentation process assigns each pixel in the image to a specific object category, achieving precise localization of construction objects through pixel-level classification. During segmentation, each pixel in the image is traversed, and the object type of the pixel is determined based on its color, texture, and relationship with surrounding pixels. Pixels belonging to the same construction object are grouped into continuous pixel regions, which fully depict the shape and extent of the construction object on the image plane. Identified construction objects are labeled with category identifiers, including specific construction item types such as paint buckets, electric drills, tiles, cement bags, and ladders.

[0023] When calculating the pixel region boundary of the construction object, the coordinates of all pixels within the pixel region are traversed. The minimum and maximum values ​​of the horizontal coordinates are found as the left and right endpoints of the boundary, and the minimum and maximum values ​​of the vertical coordinates are found as the top and bottom endpoints of the boundary. The boundary range forms a rectangular bounding box, and the coordinates of the four vertices of this bounding box represent the spatial position of the construction object on the image plane. When converting the image plane coordinates to the actual spatial position, the installation height, shooting angle, and lens focal length parameters of the video acquisition equipment are considered, and perspective transformation is used to map the two-dimensional image coordinates to three-dimensional spatial coordinates. If the installation height of the acquisition equipment is 2.5m and the shooting angle is 45°, the actual position coordinates of the construction object on the ground plane are calculated based on trigonometric relationships. The position coordinates are expressed as the relative distance from a fixed point in the work area as the origin.

[0024] When detecting occlusion in the pixel region of a construction object, the system analyzes whether there are pixels belonging to other object categories within the pixel region. When a portion of the construction object is occluded by other objects, the occluded pixels are semantically segmented and labeled as belonging to the category of the occluding object rather than the construction object itself. The number of pixels labeled as not belonging to the construction object category is counted, and this number is taken as the number of occluded pixels. The occlusion ratio is calculated by dividing the number of occluded pixels by the total number of pixels in the pixel region. When the occlusion ratio is less than 0.3, the main features of the construction object are considered still visible, and the spatial location of the construction object is marked as a complete spatial location. When the occlusion ratio is greater than or equal to 0.3, it indicates that a significant part of the construction object is occluded, and the spatial location is marked as an incomplete spatial location. A complete spatial location indicates that the location information of the construction object is reliable and accurate, while an incomplete spatial location suggests that the location information may be inaccurate and needs to be supplemented by information from other frames.

[0025] The extracted category identifier, labeled spatial location, and corresponding time identifier are combined to generate a semantic descriptor. Each construction object corresponds to one semantic descriptor, which is stored in structured data format, including a category field, a location coordinate field, an integrity flag field, and a timestamp field. All semantic descriptors are arranged according to the chronological order of their appearance in the frame sequence, forming a semantic descriptor sequence. This sequence fully records the category, location, and visibility status information of each construction object within the work area at different times, providing a structured semantic data foundation for subsequent work behavior analysis and anomaly detection.

[0026] This invention, through the aforementioned processing method, transforms continuous video data streams into a structured sequence of semantic descriptors, enabling refined analysis of the interior decoration construction process. This method accurately identifies the category and location of construction objects, effectively handles object occlusion, and ensures the integrity and reliability of extracted information, laying a data foundation for real-time monitoring and intelligent analysis of the decoration process.

[0027] Step 102: Construct a temporal dependency graph based on the semantic descriptor sequence, where nodes represent construction objects and edges represent the temporal dependencies between construction objects.

[0028] In some embodiments of the present invention, step 102 may specifically include the following sub-steps: Sub-step 1021: Based on the time identifier of the construction object in the semantic descriptor sequence, map the construction object to the node of the temporal dependency graph, and extract the spatial location of the construction object corresponding to the temporally adjacent node; Sub-step 1022: Calculate the horizontal projection area of ​​the spatial position of the construction object corresponding to the previous node and the construction object corresponding to the subsequent node, and identify the temporal dependency relationship type between the construction objects by the overlap state of the horizontal projection area. Sub-step 1023: When the horizontal projection areas overlap, extract the vertical height difference and mark the temporal dependency type as a bearing dependency; when the horizontal projection areas do not overlap, extract the horizontal spacing value and mark the temporal dependency type as an adjacency dependency. Sub-step 1024: Construct a vertical constraint propagation path based on the vertical height difference of the bearing dependency relationship, and construct a horizontal constraint isolation region based on the horizontal spacing value of the adjacency dependency relationship; Sub-step 1025: In the temporal dependency graph, establish edges from the previous node to the next node, and store the temporal dependency type, vertical constraint propagation path or horizontal constraint isolation region as edge attributes in the edge to complete the construction of the temporal dependency graph.

[0029] When constructing a temporal dependency graph based on a semantic descriptor sequence, the time identifier of each construction object is extracted from the sequence as the basis for node generation. The semantic descriptor sequence is traversed, and a corresponding node is created in the temporal dependency graph for each construction object. Each node carries the category identifier, spatial location, and time identifier information of the construction object. The nodes are sorted according to the chronological order of their time identifiers to establish temporal relationships between them. Nodes corresponding to adjacent time identifiers form temporally adjacent node pairs; the node appearing from the previous time step is called the preceding node, and the node appearing from the next time step is called the following node. The spatial location coordinates of the construction object corresponding to the preceding node and the following node are extracted from the temporally adjacent node pairs. The spatial location coordinates include the horizontal, vertical, and vertical height coordinates of the construction object in three-dimensional space.

[0030] Calculate the horizontal projection areas of the preceding and following construction objects, projecting their spatial coordinates onto the ground plane, ignoring the vertical height dimension and retaining only the horizontal and vertical coordinates. A rectangular projection area is formed on the ground plane based on the boundary range of the construction object, with its four sides parallel to the horizontal and vertical axes, respectively. Determine if the horizontal projection areas of the preceding and following construction objects overlap by comparing the boundary coordinate ranges of the two rectangular projection areas. If the horizontal coordinate ranges of the preceding and following construction objects intersect, and the vertical coordinate ranges of both intersect, the horizontal projection areas are considered to overlap. If neither the horizontal nor vertical coordinate ranges intersect, the horizontal projection areas are considered not to overlap.

[0031] When the horizontal projection areas overlap, it indicates a spatial vertical superposition between the ground projection positions of the subsequent construction object and the preceding construction object. Extract the vertical height coordinates of the preceding and subsequent construction objects and calculate their vertical height difference. The vertical height difference is the difference between the vertical height coordinates of the subsequent and preceding construction objects; a positive difference indicates the subsequent construction object is above the preceding one, and a negative difference indicates the subsequent construction object is below it. This type of temporal dependency is marked as a load-bearing dependency, indicating that the subsequent construction object may be supported by or bear the load of the preceding construction object. Load-bearing dependencies reflect the physical contact and mechanical transmission characteristics of the construction objects in the vertical direction; load-bearing dependencies exist between flooring and tiles, and between walls and coatings.

[0032] When the horizontal projection areas do not overlap, it indicates that the construction object at the subsequent node is separated from the construction object at the preceding node on the ground plane, and there is no vertical overlap between them. The shortest distance between the horizontal projection area boundaries of the preceding and subsequent node construction objects is calculated as the horizontal spacing value. All points on the boundary of the preceding node's projection area and all points on the boundary of the subsequent node's projection area are traversed, and the Euclidean distance between point pairs is calculated. The minimum value among all distances is taken as the horizontal spacing value. This type of temporal dependency is marked as an adjacency dependency, indicating that the construction object at the subsequent node is adjacent to the construction object at the preceding node in the horizontal direction but not in direct contact. Adjacency dependencies reflect the layout and arrangement of construction objects in horizontal space; there are adjacency dependencies between painting on adjacent walls and stacking materials side-by-side.

[0033] A vertical constraint propagation path is constructed based on the vertical height difference of the load-bearing dependency relationship, recording the height relationship between construction objects and possible mechanical transmission paths in the vertical direction. The vertical constraint propagation path starts from the previous node and ends at the next node, with the path weight set to the absolute value of the vertical height difference. When multiple construction objects form a vertically superimposed sequence, the vertical constraint propagation path connects adjacent nodes in the sequence, forming a propagation chain from bottom to top or from top to bottom. This path is used to track the transmission of construction dependencies in the vertical direction; changes in the lower structure will affect the stability of the upper structure through the propagation path.

[0034] A horizontal constraint isolation zone is constructed based on the horizontal spacing value of the adjacency dependency relationship, defining the range of horizontal safe distances that must be maintained between construction objects. The horizontal constraint isolation zone extends outward from the horizontal projection area boundary of the preceding node construction object, forming a buffer zone. No other construction objects that would interfere with the preceding node construction object should appear within this buffer zone. The width of the isolation zone is equal to the horizontal spacing value, and its direction points towards the location of the following node construction object. When the horizontal spacing value is less than the preset minimum safe distance threshold, the adjacency dependency relationship is marked as having a spatial conflict risk. The horizontal constraint isolation zone is used to plan the reasonable layout of construction objects, avoiding overly dense stacking of construction tools and materials that could restrict working space or cause collisions.

[0035] In the temporal dependency graph, directed edges are established from the previous node to the next node. The direction of the directed edges represents the chronological order and the temporal relationship of the construction objects. The identified temporal dependency types are stored as edge type attributes, with values ​​of either carrier dependency or adjacency dependency. For edges of carrier dependency type, the corresponding vertical constraint propagation path is added as an edge attribute, containing the vertical height difference and propagation direction information. For edges of adjacency dependency type, the corresponding horizontal constraint isolation region is added as an edge attribute, containing the horizontal spacing value and the spatial coordinates of the isolation region. After establishing edges for all temporally adjacent node pairs, the temporal dependency graph is fully constructed. The nodes in the graph are connected by directed edges to form a network structure, and the edge attributes record the spatial constraint relationships between construction objects in detail.

[0036] Through the aforementioned construction process, this invention organically combines the temporal evolution and spatial relationships of construction objects into a temporal dependency graph, clearly reflecting the order of appearance and interdependence of construction objects during the renovation process. This graph structure provides a complete topological foundation for subsequent construction conflict detection, work rationality assessment, and abnormal behavior identification, effectively supporting intelligent monitoring and risk warning functions throughout the renovation process.

[0037] Step 103: Identify the current construction stage by calculating the node mapping coverage and edge connectivity retention of the time-series dependency graph and the preset process flow diagram, and output the stage identifier.

[0038] In some embodiments of the present invention, step 103 may specifically include the following sub-steps: Sub-step 1031: Extract the category identifier and spatial location marker of the nodes in the temporal dependency graph, and extract the standard node category set and standard edge connection relationship set corresponding to each construction stage in the preset process flow diagram; Sub-step 1032: For nodes whose spatial location is marked as incomplete spatial location, search for the corresponding construction stage in the standard edge connection relationship set as a candidate construction stage based on the category identifier of the starting node of the edge pointing to the node with incomplete spatial location in the temporal dependency graph. Sub-step 1033: Calculate the number of matching nodes by matching the category identifier of the complete spatial location node with the standard node category set for each construction stage in the preset process flow diagram. Sub-step 1034: Incomplete spatial location nodes that are the same as the current construction stage in the candidate construction stage are included in the number of matching nodes, and the proportion of the number of matching nodes to the total number of nodes in the standard node category set is calculated as the node mapping coverage. Sub-step 1035: Based on the edge attributes of the edges in the time-series dependency graph, perform constraint propagation along the edges, count the number of edge connections covered by the propagation that match the standard edge connection set, and calculate the ratio of the number of matches to the total number of edge connections in the standard edge connection set as the edge connection preservation degree. Sub-step 1036: Calculate the comprehensive matching degree of each construction stage based on the node mapping coverage and edge connectivity preservation degree, select the construction stage with the highest comprehensive matching degree as the current construction stage and output the stage identifier.

[0039] When identifying the current construction stage by calculating the node mapping coverage and edge connectivity preservation of the temporal dependency graph and the preset process flow diagram, the category identifier and spatial location marker of all nodes are extracted from the temporal dependency graph. The category identifier describes the specific type of construction object, and the spatial location marker distinguishes between complete and incomplete spatial locations. The standard node category set and standard edge connectivity set corresponding to each construction stage are extracted from the preset process flow diagram. The preset process flow diagram predefines the standard construction process for interior decoration, dividing the entire decoration process into multiple construction stages, such as the foundation construction stage, water and electricity installation stage, masonry construction stage, carpentry construction stage, painting stage, and equipment installation stage. Each construction stage corresponds to a standard node category set, which contains the categories of construction objects that should appear in that stage. The standard node category set for the foundation construction stage includes material categories such as cement, sand, bricks, and steel bars. The standard edge connectivity set defines the standard dependencies between construction objects within that stage, stored in the form of node category pairs, recording the dependency types that should be established between preceding and subsequent node categories.

[0040] For nodes with incomplete spatial locations, a candidate construction stage determination process is performed. This involves traversing all incoming edges in the temporal dependency graph that point to the node with the incomplete spatial location. The starting node of each incoming edge is extracted, and its category identifier is obtained. The standard edge connection relationship set of the pre-defined process flow diagram is searched for all edge connection relationship records with the starting node's category identifier as the preceding node's category. This search process traverses the standard edge connection relationship sets of each construction stage. When a record exists in the standard edge connection relationship set of a construction stage whose preceding node category matches the starting node's category identifier, that construction stage is marked as a candidate construction stage for the node with the incomplete spatial location. An incomplete spatial location node may correspond to multiple candidate construction stages; all candidate construction stages are stored as a set of possible stages to which the node belongs.

[0041] For each construction stage in the pre-defined process flow diagram, node mapping coverage is calculated, and nodes marked as complete spatial locations in the time-series dependency graph are extracted. The category identifiers of complete spatial location nodes are traversed, and a search is performed to see if a matching category identifier exists in the standard node category set for the current construction stage. When a complete spatial location node's category identifier has a corresponding category in the standard node category set, that node is counted in the matching node count. The total number of successfully matched complete spatial location nodes is then used as the initial matching node count.

[0042] When processing the contribution of incomplete spatial location nodes to the number of matching nodes, the candidate construction stage set of all incomplete spatial location nodes is traversed. It is determined whether the currently calculated construction stage belongs to the candidate construction stage set of a certain incomplete spatial location node. If the current construction stage is the same as a candidate construction stage, the incomplete spatial location node is included in the count of matching nodes for the current construction stage. This approach is based on the logic that although incomplete spatial location nodes have limited visibility, their stage affiliation can still be inferred through dependencies. After completing the statistics for all complete spatial location nodes and eligible incomplete spatial location nodes, the total number of matching nodes for the current construction stage is obtained. A standard node category set for the current construction stage is extracted, and the total number of node categories in the set is taken as the total number of standard nodes. The node mapping coverage is calculated as the total number of matching nodes divided by the total number of standard nodes. This ratio reflects the degree to which the actual construction objects appearing in the time-series dependency graph conform to the standard requirements of the construction stage.

[0043] When calculating edge connectivity preservation, constraints are propagated along the edges based on their attributes in the time-series dependency graph. Starting from the initial node of each edge, the scope and direction of constraint propagation are determined based on the vertical constraint propagation path or horizontal constraint isolation region information stored in the edge attributes. For edges carrying dependency relationships, constraints propagate from the preceding node to the following node along the vertical constraint propagation path, covering the dependency relationship between the two nodes connected by the edge. For edges with adjacency dependency relationships, constraints identify the spatial adjacency relationship between the preceding and following nodes within the horizontal constraint isolation region. All edge connections in the time-series dependency graph covered by constraint propagation are counted, and the starting node category identifier and target node category identifier of each edge are extracted to form a node category pair. The standard edge connection relationship set for the current construction stage is searched for a standard edge connection relationship that matches this node category pair. When an edge connection relationship in the time-series dependency graph matches a record in the standard edge connection relationship set in all three dimensions—previous node category, following node category, and dependency relationship type—it is considered a successful match, and the edge is counted in the number of matched edges. After completing the matching statistics for all edges, the total number of matched edges in the current construction stage is obtained. Extract the set of standard edge connections for the current construction phase, and count the total number of edge connections in the set as the total number of standard edges. Calculate the edge connection preservation degree by dividing the total number of matching edges by the total number of standard edges. This ratio reflects the degree of consistency between the actual dependencies between construction objects in the time dependency graph and the standard requirements for this construction phase.

[0044] The overall matching degree for each construction stage is calculated based on the node mapping coverage and edge connectivity preservation. The overall matching degree is obtained through a weighted summation and is expressed as: Overall Matching Degree = w n × Node mapping coverage + w e ×Edge connectivity retention, where node mapping coverage is the node matching ratio in the current construction phase, and edge connectivity retention is the edge matching ratio in the current construction phase, w n w represents the weighting coefficient for node mapping coverage. e The weights for edge connectivity preservation are 1, and the sum of the two weights is 1. The weights are set according to the actual application scenario; a larger weight is set when accurate identification of construction object categories is more important. n When more emphasis is placed on accurately identifying the dependencies between construction objects, a larger value (w) should be set. e Value. After calculating the overall matching degree of all construction stages, compare the overall matching degree values ​​of each construction stage and select the construction stage with the highest overall matching degree as the current construction stage. Output the stage identifier corresponding to this construction stage. The stage identifier represents the specific construction stage of the current decoration progress in the form of a predefined label.

[0045] This invention achieves accurate identification of the current construction stage by performing structured matching between the temporal dependency graph and the preset process flow diagram through the above processing method. This method comprehensively considers two dimensions: category matching of construction objects and dependency matching. It can effectively handle the impact of incomplete spatial location nodes on stage identification, ensuring the reliability and accuracy of construction stage judgment.

[0046] Step 104: Extract construction specification constraints based on stage identifiers, propagate the construction specification constraints along the edges of the temporal dependency graph, adjust the spatial constraint range according to the temporal dependency relationship, and generate an extended constraint set.

[0047] In some embodiments of the present invention, step 104 may specifically include the following sub-steps: Sub-step 1041: Extract the construction specification constraints corresponding to the current construction stage from the preset construction specification library according to the stage identifier. The construction specification constraints include the spatial constraint range of each type of construction object. Sub-step 1042: Extract the edge attributes of each edge in the temporal dependency graph and identify the type of temporal dependency relationship in the edge attributes; Sub-step 1043: For edges carrying dependency relationships, extract the vertical constraint propagation path and vertical height difference from the edge attributes, propagate the spatial constraint range corresponding to the starting node along the vertical constraint propagation path to the ending node, and superimpose the vertical height difference onto the boundary coordinates of the spatial constraint range corresponding to the ending node in the vertical direction to generate the vertically adjusted spatial constraint range. Sub-step 1044: For edges with adjacency dependency relationship type, extract the horizontal constraint isolation region and horizontal spacing value from the edge attributes, and adjust the horizontal spacing between the spatial constraint ranges corresponding to the start node and the end node according to the horizontal spacing value to generate the horizontally adjusted spatial constraint range. Sub-step 1045: For the case where multiple edges point to the same termination node, extract the adjusted spatial constraint range of each edge propagating to the termination node, and calculate the intersection region of the adjusted spatial constraint ranges as the spatial constraint range corresponding to the termination node. Sub-step 1046: Summarize the spatial constraint ranges corresponding to each node after adjustment to generate an extended constraint set.

[0048] When extracting construction specification constraints based on stage identifiers, the pre-set construction specification library is accessed, and the corresponding construction specification constraint entries are queried based on the stage identifier of the current construction stage. The pre-set construction specification library stores standard construction specifications for each construction stage. Each specification constraint record includes the category identifier of the construction object and the spatial constraint range that the construction object of that category must comply with. The spatial constraint range defines the allowed positional boundaries of the construction object in three-dimensional space, represented in the form of a three-dimensional spatial coordinate system, including the horizontal coordinate range, the vertical coordinate range, and the vertical height coordinate range. In the construction specification constraints of the masonry construction stage, the spatial constraint range for wall tiles requires its vertical height coordinate range to be between 0.2m and 2.8m above the ground, while the spatial constraint range for floor tiles requires its vertical height coordinate range to be between 0m and 0.05m above the ground. The spatial constraint ranges corresponding to all construction object categories involved in the current construction stage are extracted to construct an initial set of spatial constraint ranges.

[0049] Extract the edge attributes of all edges in the temporal dependency graph, and read the temporal dependency type field stored in the edge attributes. Temporal dependency types are divided into two categories: carrier dependency and adjacency dependency. Different edge types correspond to different spatial constraint propagation and adjustment methods. Traverse each edge in the temporal dependency graph, and classify the edge into either the carrier dependency edge set or the adjacency dependency edge set according to the temporal dependency type in the edge attributes.

[0050] For edges with load-bearing dependency relationships, vertical constraint propagation and adjustment are performed. The vertical constraint propagation path and vertical height difference are extracted from the edge attributes. The vertical constraint propagation path records the vertical transmission channel of the constraint from the starting node to the ending node along the edge. The vertical height difference reflects the relative positional relationship between the construction objects at the starting and ending nodes in the vertical direction. The initial spatial constraint range of the construction object corresponding to the starting node is obtained, and this range is propagated along the vertical constraint propagation path. During the propagation process, the horizontal and vertical coordinate ranges remain unchanged, only the vertical height coordinate range is adjusted. The vertical height difference is superimposed on the vertical height coordinate boundary of the starting node's spatial constraint range to generate the propagated vertical height coordinate range. When the lower boundary of the starting node's spatial constraint range is 0.2m, the upper boundary is 2.8m, and the vertical height difference is 3.0m, the lower boundary of the propagated vertical height coordinate is adjusted to 3.2m, and the upper boundary is adjusted to 5.8m. The adjusted spatial constraint range is used as the vertically adjusted spatial constraint range corresponding to the ending node.

[0051] For edges with adjacency dependencies, horizontal constraint adjustments are performed, extracting the horizontal constraint isolation region and horizontal spacing value from the edge attributes. The horizontal constraint isolation region identifies the horizontal separation range that should be maintained between the starting and ending construction objects, and the horizontal spacing value quantifies the minimum horizontal distance between them. The initial spatial constraint ranges of the starting and ending construction objects are obtained. Based on the horizontal spacing value, the horizontal spacing between the starting and ending node spatial constraint ranges is adjusted. The lateral and longitudinal coordinate boundaries of the starting node's spatial constraint range are calculated, and the horizontal spacing value is extended outside the boundaries to form the extended isolation boundary. It is ensured that the lateral and longitudinal coordinate boundaries of the ending node's spatial constraint range maintain a distance of at least the horizontal spacing value from the extended isolation boundary. When the right lateral boundary of the starting node's spatial constraint range is 5.0m and the horizontal spacing value is 0.8m, the left lateral boundary of the ending node's spatial constraint range should be no less than 5.8m. The lateral and longitudinal coordinate ranges of the ending node's spatial constraint range are adjusted to generate the horizontally adjusted spatial constraint range.

[0052] For time-dependent graphs where multiple edges point to the same terminal node, a comprehensive processing of spatial constraint ranges is performed, identifying all incoming edges with that terminal node as the target node. The adjusted spatial constraint range propagated to the terminal node for each incoming edge is extracted. Different incoming edges may originate from different starting nodes, thus the propagated spatial constraint ranges may differ. A candidate set of spatial constraint ranges is formed by collecting all spatial constraint ranges propagated to the terminal node from all incoming edges. The intersection region of all spatial constraint ranges in the candidate set is calculated, with intersection calculations performed in the horizontal, vertical, and height coordinate dimensions. The horizontal coordinate intersection range is the overlapping portion of the horizontal coordinate ranges of all candidate spatial constraint ranges, specifically from the maximum value of the left boundary of all horizontal coordinates to the minimum value of the right boundary of all horizontal coordinates. The vertical coordinate intersection range is the overlapping portion of the vertical coordinate ranges of all candidate spatial constraint ranges, specifically from the maximum value of the front boundary of all vertical coordinates to the minimum value of the back boundary of all vertical coordinates. The vertical height coordinate intersection range is the overlapping portion of the vertical height coordinate ranges of all candidate spatial constraint ranges, specifically from the maximum value of the lower boundary of all vertical height coordinates to the minimum value of the upper boundary of all vertical height coordinates. The intersection ranges of the three dimensions are combined to form the final spatial constraint range of the termination node, which simultaneously satisfies the constraint requirements for the propagation of all incoming edges. When the termination node has three incoming edges, the spatial constraint ranges for the propagation of the three incoming edges are extracted respectively. In the horizontal coordinate dimension, the maximum value of the left boundary and the minimum value of the right boundary of the three ranges are taken to form the horizontal intersection range. In the vertical coordinate dimension, the maximum value of the front boundary and the minimum value of the back boundary of the three ranges are taken to form the vertical intersection range. In the vertical height coordinate dimension, the maximum value of the lower boundary and the minimum value of the upper boundary of the three ranges are taken to form the vertical intersection range. The combination of the intersection ranges of the three dimensions is the final spatial constraint range of the termination node.

[0053] Traverse all nodes in the temporal dependency graph and collect the final spatial constraint range of each node after constraint propagation and adjustment. For source nodes without incoming edges, their spatial constraint range remains the initial spatial constraint range extracted from the construction specification constraints. For nodes with incoming edges, their spatial constraint range is the adjusted spatial constraint range calculated based on incoming edge propagation and intersection. Organize the spatial constraint ranges corresponding to all nodes by indexing them according to their node identifiers to form a complete extended constraint set. The extended constraint set is stored in key-value pairs, where the key is the node identifier and the value is the 3D coordinate boundary of the spatial constraint range corresponding to that node.

[0054] This invention combines static construction specification constraints with dynamic temporal dependencies through the aforementioned processing method, achieving intelligent propagation and adaptive adjustment of spatial constraint range. This method can dynamically correct spatial constraint boundaries based on the bearing capacity and adjacency relationships between construction objects, effectively handling the comprehensive impact of multi-source constraints on the same construction object and ensuring the accuracy and consistency of constraint propagation.

[0055] Step 105: Analyze the category identifier and spatial location of construction objects in the semantic descriptor sequence, infer the complete spatial location for construction objects marked as having incomplete spatial locations, and update the semantic descriptor sequence.

[0056] In some embodiments of the present invention, step 105 may specifically include the following sub-steps: Sub-step 1051: Extract construction objects whose spatial location is marked as incomplete spatial location from the semantic descriptor sequence, and extract each incoming edge of the node corresponding to the construction object in the temporal dependency graph that points to the incomplete spatial location. Sub-step 1052: Extract the category identifier and spatial position of the construction object corresponding to the starting node of each edge, obtain the corresponding standard size information based on the category identifier, and superimpose the standard size information onto the spatial position of the construction object corresponding to the starting node to generate the spatial boundary range of the construction object corresponding to the starting node. Sub-step 1053: Extract the temporal dependency type from the edge attributes of each edge; for the bearing dependency type, extract the upper boundary of the spatial boundary range in the vertical direction as the vertical inference boundary; for the adjacency dependency type, extract the side boundary of the spatial boundary range in the horizontal direction as the horizontal inference boundary. Sub-step 1054: Obtain the standard dimension information corresponding to the construction object with incomplete spatial location, and perform spatial matching with the standard dimension information and the vertical inference boundary and the horizontal inference boundary to calculate the complete spatial location of the construction object with incomplete spatial location; Sub-step 1055: Update the complete spatial location of the construction object with incomplete spatial location to the semantic descriptor sequence.

[0057] Construction objects with incomplete spatial locations are extracted from the semantic descriptor sequence. All construction object records in the semantic descriptor sequence are traversed to identify entries marked as incomplete in the spatial location field. An incomplete spatial location refers to a construction object whose three-dimensional spatial coordinate information is missing or incomplete; for example, it may only contain horizontal and vertical coordinates but lack vertical height coordinates, or only contain vertical height coordinates but lack horizontal and vertical coordinates. The identified construction objects with incomplete spatial locations are used to form a set of construction objects to be inferred, recording the category identifier and partial spatial location information of each object. For each construction object in the set, the corresponding node is located in the temporal dependency graph, and all incoming edges pointing to that node are extracted. Incoming edges represent the temporal dependency relationship between a preceding construction object and the current construction object to be inferred. The spatial location and spatial boundary range of the preceding construction object provide a spatial reference for inferring the complete spatial location of the current construction object to be inferred.

[0058] Traverse each incoming edge, extract the construction object record corresponding to the starting node of the incoming edge, and read the category identifier and spatial location information of the construction object from the construction object record. The category identifier is used to query the standard size information of the construction object of that category, and the spatial location information includes the coordinate position of the construction object in three-dimensional space. Access the preset construction object standard size database, and query the standard size information of the construction object of that category according to the category identifier of the construction object corresponding to the starting node. The standard size information includes the standard width, standard length, and standard height of the construction object in the horizontal direction, the vertical direction, and the vertical direction. The standard size information of wall tiles is 0.3m in the horizontal direction, 0.45m in the vertical direction, and 0.008m in the vertical direction. The standard size information of floor tiles is 0.8m in the horizontal direction, 0.8m in the vertical direction, and 0.015m in the vertical direction. Take the spatial location of the construction object corresponding to the starting node as the spatial starting point, and superimpose the standard size information on the spatial starting point to generate the spatial boundary range of the construction object corresponding to the starting node. The spatial boundary defines the actual area occupied by the construction object in three-dimensional space, including the left and right boundaries of the horizontal coordinate, the front and back boundaries of the vertical coordinate, and the lower and upper boundaries of the vertical height coordinate. When the starting node corresponds to the construction object with a horizontal coordinate of 2.0m, a vertical coordinate of 3.0m, and a vertical height of 0.0m, and the standard dimensions are 0.8m horizontal width, 0.8m vertical length, and 0.015m vertical height, the generated spatial boundary will have a left boundary of 2.0m and a right boundary of 2.8m, a front boundary of 3.0m and a back boundary of 3.8m, and a lower boundary of 0.0m and an upper boundary of 0.015m.

[0059] Extract the edge attributes of each incoming edge and read the temporal dependency type field stored in the edge attributes. Temporal dependency types are distinguished into bearer dependency and adjacency dependency, with different dependency types corresponding to different spatial inference directions and inference boundaries. For incoming edges of the bearer dependency type, extract the upper vertical boundary of the spatial boundary range of the construction object corresponding to the starting node as the vertical inference boundary. The vertical inference boundary represents the highest position of the preceding construction object in the vertical direction; the current construction object to be inferred should be located above this vertical inference boundary due to the bearer dependency. When the upper vertical height coordinate of the spatial boundary range of the construction object corresponding to the starting node is 0.015m, the corresponding vertical inference boundary is 0.015m. For incoming edges of the adjacency dependency type, extract the side horizontal boundary of the spatial boundary range of the construction object corresponding to the starting node as the horizontal inference boundary. The horizontal inference boundary represents the edge position of the preceding construction object in the horizontal direction; the current construction object to be inferred should be located on the adjacent side of this horizontal inference boundary due to the adjacency dependency. When the right boundary of the spatial boundary range of the construction object corresponding to the starting node is 2.8m, the corresponding horizontal inference boundary is also 2.8m. Collect all the vertical and horizontal inference boundaries corresponding to the incoming edges to form a set of inference boundaries.

[0060] Access the standard dimension database of the construction object. Query the standard dimension information of the construction object to be inferred based on its category identifier. The standard dimension information includes the standard width in the horizontal direction, the standard length in the vertical direction, and the standard height in the vertical direction. Perform spatial matching calculations between the standard dimension information of the construction object to be inferred and the vertical inference boundaries in the inference boundary set. For spatial locations where the vertical direction is missing, use the vertical inference boundary as the lower boundary of the vertical height coordinate of the construction object to be inferred. Superimpose the standard height of the construction object to this lower boundary to obtain the upper boundary of the vertical height coordinate. When the vertical inference boundary is 0.015m and the standard height of the construction object to be inferred is 0.3m, the lower boundary of the vertical height coordinate of the construction object to be inferred is 0.015m, and the upper boundary is 0.315m. Perform spatial matching calculations between the standard dimension information of the construction object to be inferred and the horizontal inference boundaries in the inference boundary set. For spatial locations where the horizontal coordinate is missing, use the horizontal inference boundary as the left boundary of the horizontal coordinate of the construction object to be inferred. Superimpose the standard width of the construction object to this left boundary to obtain the right boundary of the horizontal coordinate. When the horizontal inference boundary is 2.8m and the standard width of the construction object to be inferred is 0.3m, the left boundary of the horizontal coordinate of the construction object to be inferred is 2.8m and the right boundary is 3.1m. For spatial locations lacking vertical coordinates, a similar method is used to calculate the vertical coordinate boundary based on the horizontal inference boundary and the standard length. Combining the inference results in the vertical and horizontal directions, the complete spatial location of the construction object to be inferred is constructed. The complete spatial location includes complete three-dimensional coordinate information of horizontal coordinates, vertical coordinates, and vertical height coordinates.

[0061] The process involves locating the record entry corresponding to the construction object to be inferred within the semantic descriptor sequence, and updating the spatial location field of that record entry with the inferred complete spatial location. The update operation replaces the content of fields previously marked as incomplete spatial locations with the inferred complete 3D spatial coordinate information, and updates the spatial location marker status from incomplete to complete. This process is repeated for each construction object in the set of construction objects to be inferred, ensuring that the spatial location information of all construction objects in the semantic descriptor sequence is complete and accurate. The updated semantic descriptor sequence contains the complete spatial location information of all construction objects, providing a comprehensive spatial data foundation for subsequent spatial compliance checks and temporal dependency verification.

[0062] This invention achieves intelligent reasoning and completion for construction objects with incomplete spatial locations by analyzing the spatial boundary range and standard size information of preceding construction objects in temporal dependencies. This method fully utilizes the spatial constraint characteristics of carrying dependencies and adjacency dependencies, selecting appropriate reasoning boundaries and directions according to different dependency types, ensuring the spatial rationality and logical consistency of the reasoning results, and effectively solving the problem of incomplete data caused by missing spatial location information of construction objects.

[0063] Step 106: Match the updated semantic descriptor sequence with the extended constraint set, calculate the deviation degree of spatial location from the spatial constraint range, generate spatial compliance degree, generate anomaly judgment results based on spatial compliance degree, and output monitoring instructions.

[0064] In some embodiments of the present invention, step 106 may specifically include the following sub-steps: Sub-step 1061: Extract the complete spatial location of each construction object in the updated semantic descriptor sequence, and extract the corresponding spatial constraint range from the extended constraint set; Sub-step 1062: Identify the set of nodes forming a closed-loop path in the time-series dependency graph; for the construction object in the closed-loop path, calculate the deviation distance between the complete spatial location and the spatial constraint range; and accumulate the propagation deviation distance along the closed-loop path to obtain the closed-loop cumulative deviation value. Sub-step 1063: Calculate the closed-loop deviation amplification factor based on the closed-loop cumulative deviation value and the closed-loop path length. For nodes in the closed-loop path, multiply the closed-loop deviation amplification factor by the deviation distance to obtain the deviation degree. For nodes in the non-closed-loop path, use the deviation distance as the deviation degree. Sub-step 1064: Generate spatial compliance based on deviation, generate anomaly determination results based on spatial compliance, extract construction objects in closed-loop paths where the closed-loop deviation amplification coefficient exceeds the preset amplification threshold based on the anomaly determination results, generate monitoring instructions and output them.

[0065] Extract the complete spatial location information of each construction object from the updated semantic descriptor sequence. Iterate through all construction object records in the sequence, reading the category identifier and complete spatial location field for each object. The complete spatial location includes the object's horizontal, vertical, and height coordinates in 3D space. Access the extended constraint set and query the corresponding spatial constraint range based on the object's category identifier. The extended constraint set is stored as key-value pairs, where the key is the object's category identifier and the value is the spatial constraint range that the object must adhere to. The spatial constraint range defines the allowed boundaries of the construction object's location in 3D space, including the horizontal, vertical, and height coordinate ranges. For example, the wall decoration panel's spatial constraint range includes a horizontal coordinate range of 1.0m to 5.0m, a vertical coordinate range of 0.5m to 4.5m, and a height coordinate range of 0.5m to 2.8m. Associate the complete spatial location of each construction object with its corresponding spatial constraint range to construct a set of matching pairs between construction object spatial locations and constraint ranges.

[0066] Identify the set of nodes forming closed-loop paths in the temporal dependency graph. Traverse all nodes in the temporal dependency graph, performing a depth-first traversal along directed edges from each node, checking if there is a path returning to a visited node during the traversal. A closed-loop path is formed when a path originating from a node and returning to that node after traversing several directed edges. Record all nodes and edges contained in the closed-loop path; nodes in the closed-loop path represent construction objects with cyclic temporal dependencies. A closed-loop path may consist of a triangular loop with three nodes or a polygonal loop with four or more nodes. A closed-loop path is formed when node A points to node B, node B points to node C, and node C points to node A. Construct a closed-loop path set from all identified closed-loop paths, recording the node and edge sequences contained in each closed-loop path.

[0067] For each construction object in the closed-loop path, calculate the deviation distance between its complete spatial position and the corresponding spatial constraint range. Compare the complete spatial position coordinates of the construction object with the coordinate boundaries of the spatial constraint range to determine whether the spatial position is within the spatial constraint range. When the horizontal, vertical, and all three coordinates of the construction object are within the coordinate boundaries of the corresponding spatial constraint range, the deviation distance is 0m. When the spatial position coordinates of the construction object exceed the boundaries of the spatial constraint range, calculate the distance of the excess portion as the deviation distance. When the horizontal coordinate of the construction object is 5.5m and the right boundary of the horizontal coordinate of the spatial constraint range is 5.0m, the horizontal deviation distance is 0.5m. When the vertical coordinate of the construction object is 3.2m and the upper boundary of the vertical coordinate of the spatial constraint range is 2.8m, the vertical deviation distance is 0.4m. Combine the horizontal, vertical, and total deviation distances to calculate the total deviation distance in three-dimensional space. The total deviation distance is calculated using Euclidean distance, by taking the square root of the sum of the squares of the deviation distances in the three directions. When the lateral deviation is 0.5m, the longitudinal deviation is 0.3m, and the vertical deviation is 0.4m, the total spatial deviation is 0.71m.

[0068] The deviation distance of each node is accumulated and propagated along the closed-loop path in node order. Starting from the initial node of the closed-loop path, the deviation distance of that node is used as the initial value of the cumulative deviation. Moving to the next node in the closed-loop path, the deviation distance of that node is added to the current cumulative deviation value. This process is repeated for all nodes in the closed-loop path, accumulating the deviation distance of each node into the cumulative deviation value. After completing the full traversal of the closed-loop path, the cumulative deviation value is obtained. With three nodes in the closed-loop path, and deviation distances of 0.5m, 0.3m, and 0.4m respectively, the cumulative deviation value is 1.2m. The cumulative deviation value reflects the overall degree to which all construction objects in the closed-loop path deviate from the spatial constraints.

[0069] The number of edges in the closed-loop path is counted, representing the path length. The cumulative deviation value is divided by the path length to obtain the average single-edge deviation value. This average single-edge deviation value is compared with a preset baseline deviation value to calculate the closed-loop deviation amplification factor. The closed-loop deviation amplification factor is expressed as: Where α is the closed-loop deviation amplification factor, and D l The cumulative deviation of the closed loop is given by L, where L is the closed loop path length (i.e., the number of edges), and D is the cumulative deviation of the closed loop. j This is the preset baseline deviation value. The baseline deviation value is set to 0.2m, the closed-loop cumulative deviation value is 1.2m, and when the closed-loop path length is 3, the closed-loop deviation amplification factor is 3.0. A closed-loop deviation amplification factor greater than 1 indicates that there is a cumulative deviation effect in the closed-loop path; the larger the factor, the more severe the deviation amplification.

[0070] For each node in the closed-loop path, the deviation distance of that node is multiplied by the closed-loop deviation amplification factor to obtain the deviation degree of the construction object corresponding to that node. The deviation degree reflects the actual severity of the deviation of the construction object after considering the cumulative effect of the closed loop. When the deviation distance of a node is 0.5m and the closed-loop deviation amplification factor is 3.0, the deviation degree of that node is 1.5m. For nodes in non-closed-loop paths, the deviation distance of the node is directly used as the deviation degree without adjusting the amplification factor. Nodes in non-closed-loop paths do not have the cumulative effect of deviation caused by circular dependencies, so the deviation distance is the actual deviation degree. Traverse all nodes in the time-series dependency graph, calculate the corresponding deviation degree for each node, and construct a set of node deviation degrees.

[0071] Spatial compliance is calculated based on node deviation. Spatial compliance uses a normalized method to represent the degree to which the spatial position of the construction object conforms to the constraints. A maximum allowable deviation is set as a preset threshold, and the node deviation is compared with the maximum allowable deviation for calculation. Spatial compliance is expressed as: Where S is the spatial conformity, ranging from 0 to 1, and d is the node deviation. z This represents the preset maximum allowable deviation. The maximum allowable deviation is set to 2.0m. When the node deviation is 1.5m, the spatial compliance is 0.25. A spatial compliance close to 1 indicates that the spatial position and height of the construction object conform to the constraints, while a spatial compliance close to 0 indicates that the spatial position of the construction object deviates significantly from the constraints. The spatial compliance is calculated for each node, constructing a set of node spatial compliance.

[0072] A spatial compliance threshold is set, and the spatial compliance set of each node is traversed, comparing the spatial compliance of each node with the threshold. When the spatial compliance of a node is lower than the threshold, the construction object corresponding to that node is determined to have a spatial anomaly. The threshold is set to 0.7; when the spatial compliance of a node is 0.25, the node is marked as a spatial anomaly node. All spatial anomaly nodes are counted, and an anomaly determination result is generated. The anomaly determination result includes the identifier of the anomaly node, the category identifier of the corresponding construction object, its spatial location, deviation degree, and spatial compliance information.

[0073] The system iterates through all abnormal nodes in the anomaly detection results, identifying whether each node is located within a closed-loop path. For abnormal nodes within a closed-loop path, the corresponding closed-loop deviation amplification coefficient is extracted. A preset amplification threshold is set, and the closed-loop deviation amplification coefficient is compared to this threshold. With a preset amplification threshold of 2.5, a closed-loop path is marked as high-risk when the closed-loop deviation amplification coefficient is 3.0. All construction objects within the high-risk closed-loop path are extracted, and their category identifiers, spatial locations, and deviation information are encapsulated into monitoring instructions. These instructions include a list of construction objects requiring close monitoring, prompting construction personnel to adjust the positions and rectify any compliance issues. The generated monitoring instructions are output to the monitoring terminal device via a communication interface, enabling real-time early warning and guidance for spatial anomalies during the renovation process.

[0074] This invention achieves a quantitative assessment of the cumulative deviation effect in cyclic dependencies by identifying closed-loop paths in the time-series dependency graph and calculating the cumulative deviation value of the closed loop. The introduction of the closed-loop deviation amplification factor effectively amplifies the deviation of construction objects in the closed-loop path, highlighting the risk of deviation propagation and accumulation caused by cyclic dependencies, improving the accuracy and sensitivity of anomaly detection, and ensuring the pertinence and effectiveness of monitoring instructions.

[0075] like Figure 2 As shown, Figure 2 This is a schematic diagram of a monitoring system for the interior decoration process provided in an embodiment of the present invention. The system includes: The semantic recognition module 201 is used to acquire video data streams of the indoor decoration work area, identify construction objects through temporal segmentation and semantic segmentation, and extract category identifiers, spatial locations and time identifiers to generate a sequence of semantic descriptors; Dependency graph construction module 202 is used to construct a temporal dependency graph based on a sequence of semantic descriptors, where nodes represent construction objects and edges represent temporal dependencies between construction objects. The stage identification module 203 is used to identify the current construction stage by calculating the node mapping coverage and edge connectivity preservation degree of the time sequence dependency graph and the preset process flow diagram, and output the stage identifier. The constraint generation module 204 is used to extract construction specification constraints based on the stage identifier, propagate the construction specification constraints along the edges of the temporal dependency graph, adjust the spatial constraint range according to the temporal dependency relationship, and generate an extended constraint set. The location reasoning module 205 is used to analyze the category identifier and spatial location of construction objects in the semantic descriptor sequence, and to reason about the complete spatial location of construction objects marked as incomplete spatial locations and update the semantic descriptor sequence. The anomaly monitoring module 206 is used to match the updated semantic descriptor sequence with the extended constraint set, calculate the deviation degree of spatial location from the spatial constraint range, generate spatial compliance degree, generate anomaly judgment results based on spatial compliance degree, and output monitoring instructions.

[0076] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0077] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing a computer program, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0078] The specific embodiments described above are preferred embodiments of the present invention and are not intended to limit the specific scope of the present invention. The scope of the present invention includes, but is not limited to, these specific embodiments. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A method for monitoring the interior decoration process, characterized in that, Includes the following steps: The video data stream of the indoor decoration work area is acquired, and the construction objects are identified through temporal segmentation and semantic segmentation. The category identifier, spatial location and time identifier are extracted to generate a semantic descriptor sequence. A temporal dependency graph is constructed based on a sequence of semantic descriptors, where nodes represent construction objects and edges represent the temporal dependencies between construction objects. The current construction stage is identified by calculating the node mapping coverage and edge connectivity preservation of the temporal dependency graph and the preset process flow diagram, and the stage identifier is output. Extract construction specification constraints based on stage identifiers, propagate these constraints along the edges of the temporal dependency graph, adjust the spatial constraint range based on temporal dependencies, and generate an extended constraint set. Analyze the category identifiers and spatial locations of construction objects in the semantic descriptor sequence, infer the complete spatial locations for construction objects marked with incomplete spatial locations, and update the semantic descriptor sequence; The updated semantic descriptor sequence is matched with the extended constraint set, the deviation of the spatial location from the spatial constraint range is calculated, the spatial compliance is generated, the anomaly judgment result is generated based on the spatial compliance, and the monitoring instruction is output.

2. The method according to claim 1, characterized in that, Construction objects are identified through temporal and semantic segmentation, and a sequence of semantic descriptors is generated by extracting category identifiers, spatial locations, and temporal identifiers, including: The video data stream is parsed into a sequence of frames arranged in chronological order. The frame sequence is segmented temporally to identify changes in the motion state of construction objects and scene content between consecutive frames, determine the start and end frames of the operation, and segment the frame sequence into multiple operation segments based on the start and end frames. Semantic segmentation is performed on frames in the work segment to identify construction objects, and pixel regions of construction objects are extracted and labeled with category identifiers of construction objects. Calculate the boundary range of the pixel region and convert the boundary range into the spatial location of the construction object; Detect the occlusion situation in the pixel area of ​​the construction object and count the number of occluded pixels. Calculate the occlusion ratio of the number of occluded pixels to the total number of pixels in the pixel area. Mark the spatial position of the construction object as a complete spatial position or an incomplete spatial position according to the occlusion ratio. Generate a sequence of semantic descriptors by identifying the category, spatial location, and time of the construction object.

3. The method according to claim 1, characterized in that, Constructing a temporal dependency graph based on a sequence of semantic descriptors includes: Based on the time identifier of the construction object in the semantic descriptor sequence, the construction object is mapped to the node of the temporal dependency graph, and the spatial location of the construction object corresponding to the temporally adjacent node is extracted. Calculate the horizontal projection area of ​​the spatial positions of the construction object corresponding to the previous node and the construction object corresponding to the subsequent node, and identify the temporal dependency relationship type between the construction objects by the overlap state of the horizontal projection area; When the horizontal projection areas overlap, the vertical height difference is extracted and the temporal dependency type is marked as a bearing dependency. When the horizontal projection areas do not overlap, the horizontal spacing value is extracted and the temporal dependency type is marked as an adjacency dependency. A vertical constraint propagation path is constructed based on the vertical height difference of the bearing dependency, and a horizontal constraint isolation region is constructed based on the horizontal spacing of the adjacency dependency. In the temporal dependency graph, edges are created from the previous node to the next node. The temporal dependency type, vertical constraint propagation path, or horizontal constraint isolation region are stored as edge attributes in the edges to complete the construction of the temporal dependency graph.

4. The method according to claim 1, characterized in that, The current construction stage is identified by calculating the node mapping coverage and edge connectivity preservation between the temporal dependency graph and the preset process flow diagram. The output stage identifier includes: Extract the category identifiers and spatial location markers of nodes in the temporal dependency graph, and extract the standard node category set and standard edge connection relationship set corresponding to each construction stage in the preset process flow diagram; For nodes whose spatial location is marked as incomplete spatial location, the corresponding construction stage is searched in the standard edge connection relationship set as a candidate construction stage based on the category identifier of the starting node of the edge pointing to the node with incomplete spatial location in the temporal dependency graph. Calculations are performed for each construction stage in the preset process flow diagram, and the number of matching nodes is obtained by matching the category identifier of the complete spatial location node with the standard node category set. Incomplete spatial location nodes that are the same as the candidate construction stage and the current construction stage are included in the number of matching nodes. The proportion of the number of matching nodes to the total number of nodes in the standard node category set is calculated as the node mapping coverage. Constraint propagation is performed along the edges based on the edge attributes of the temporal dependency graph. The number of edge connections covered by the propagation that match the standard edge connection set is counted. The proportion of the number of matches to the total number of edge connections in the standard edge connection set is calculated as the edge connection preservation degree. The comprehensive matching degree of each construction stage is calculated based on the node mapping coverage and edge connectivity preservation degree. The construction stage with the highest comprehensive matching degree is selected as the current construction stage and the stage identifier is output.

5. The method according to claim 1, characterized in that, Construction specification constraints are extracted based on stage identifiers, propagated along the edges of the temporal dependency graph, and the spatial constraint range is adjusted according to the temporal dependencies to generate an extended constraint set including: Based on the stage identifier, extract the construction specification constraints corresponding to the current construction stage from the preset construction specification library. The construction specification constraints include the spatial constraint range of each type of construction object. Extract the edge attributes of each edge in the temporal dependency graph and identify the types of temporal dependencies in the edge attributes; For edges with a bearing dependency relationship, extract the vertical constraint propagation path and vertical height difference from the edge attributes, propagate the spatial constraint range corresponding to the starting node along the vertical constraint propagation path to the ending node, and superimpose the vertical height difference onto the boundary coordinates of the spatial constraint range corresponding to the ending node in the vertical direction to generate the vertically adjusted spatial constraint range. For edges with adjacency dependency relationships, extract the horizontal constraint isolation region and horizontal spacing value from the edge attributes. Based on the horizontal spacing value, adjust the horizontal spacing between the spatial constraint ranges corresponding to the start node and the end node to generate the horizontally adjusted spatial constraint range. For cases where multiple edges point to the same termination node, the adjusted spatial constraint ranges propagated from each edge to the termination node are extracted, and the intersection of the adjusted spatial constraint ranges is calculated as the spatial constraint range corresponding to the termination node. The adjusted spatial constraint ranges for each node are aggregated to generate an extended constraint set.

6. The method according to claim 1, characterized in that, Analyzing the category identifiers and spatial locations of construction objects in the semantic descriptor sequence, inferring the complete spatial locations for construction objects marked with incomplete spatial locations, and updating the semantic descriptor sequence includes: Extract construction objects whose spatial location is marked as incomplete spatial location from the semantic descriptor sequence, and extract each incoming edge of the node corresponding to the construction object with incomplete spatial location in the temporal dependency graph; Extract the category identifier and spatial position of the construction object corresponding to the starting node of each edge, obtain the corresponding standard size information based on the category identifier, and superimpose the standard size information onto the spatial position of the construction object corresponding to the starting node to generate the spatial boundary range of the construction object corresponding to the starting node. Extract the temporal dependency type from the edge attributes of each edge. For the carrying dependency type, extract the upper boundary of the spatial boundary in the vertical direction as the vertical inference boundary. For the adjacency dependency type, extract the side boundary of the spatial boundary in the horizontal direction as the horizontal inference boundary. Obtain the standard dimension information corresponding to the construction object with incomplete spatial location, and perform spatial matching between the standard dimension information and the vertical and horizontal reasoning boundaries to calculate the complete spatial location of the construction object with incomplete spatial location. Update the complete spatial location of the construction object whose spatial location is incomplete to the semantic descriptor sequence.

7. The method according to claim 1, characterized in that, The updated semantic descriptor sequence is matched with the extended constraint set to calculate the deviation of spatial location from the spatial constraint range, generate spatial compliance, and generate anomaly judgment results based on the spatial compliance, and output monitoring instructions including: Extract the complete spatial location of each construction object from the updated semantic descriptor sequence, and extract the corresponding spatial constraint range from the extended constraint set; In the time-dependent graph, identify the set of nodes that form a closed-loop path. For the construction objects in the closed-loop path, calculate the deviation distance between the complete spatial location and the spatial constraint range. Accumulate the propagation deviation distance along the closed-loop path to obtain the closed-loop cumulative deviation value. The closed-loop deviation amplification factor is calculated based on the cumulative deviation value and the closed-loop path length. For nodes in the closed-loop path, the deviation amplification factor is multiplied by the deviation distance to obtain the deviation degree. For nodes in the non-closed-loop path, the deviation distance is used as the deviation degree. Based on the deviation, spatial compliance is generated; based on the spatial compliance, anomaly judgment results are generated; based on the anomaly judgment results, construction objects in the closed-loop path where the closed-loop deviation amplification coefficient exceeds the preset amplification threshold are extracted; monitoring instructions are generated and output.

8. A monitoring system for the interior decoration process, used to implement the method described in any one of claims 1-7, characterized in that, The system includes: The semantic recognition module is used to acquire video data streams of the indoor decoration work area, identify construction objects through temporal segmentation and semantic segmentation, and extract category identifiers, spatial locations and time identifiers to generate a sequence of semantic descriptors; The dependency graph construction module is used to construct a temporal dependency graph based on a sequence of semantic descriptors, where nodes represent construction objects and edges represent the temporal dependencies between construction objects. The stage identification module is used to identify the current construction stage by calculating the node mapping coverage and edge connectivity preservation degree between the time-series dependency graph and the preset process flow diagram, and output the stage identifier. The constraint generation module is used to extract construction specification constraints based on stage identifiers, propagate the construction specification constraints along the edges of the temporal dependency graph, adjust the spatial constraint range according to the temporal dependency relationship, and generate an extended constraint set. The location reasoning module is used to analyze the category identifier and spatial location of construction objects in the semantic descriptor sequence, and to reason about the complete spatial location of construction objects marked with incomplete spatial locations and update the semantic descriptor sequence. The anomaly monitoring module is used to match the updated semantic descriptor sequence with the extended constraint set, calculate the deviation degree of spatial location from the spatial constraint range, generate spatial compliance degree, generate anomaly judgment results based on spatial compliance degree, and output monitoring instructions.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 7.

Images