Production scheduling management system and method for fabricated building

Abstract

The invention discloses a production scheduling management system and method for a prefabricated building. The method comprises the following steps: acquiring a monitoring image of a site constructor through a camera; performing image analysis on the monitoring image of the on-site construction personnel to obtain personnel head features; and determining whether an early warning prompt is generated or not based on the head features of the person. Therefore, the problems of low efficiency and low precision of manual inspection can be avoided, so that the supervision efficiency and the supervision level of a using unit are optimized, and the construction safety is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent management technology, and in particular to a production scheduling management system and method for prefabricated buildings. Background Art

[0002] Prefabricated construction is a fast and efficient construction method that reduces construction time and improves building quality by prefabricating components in a factory environment and assembling them on site. However, the safety of construction workers remains an important concern during the prefabricated construction process.

[0003] In order to ensure the safety of construction workers, it is necessary to monitor and manage whether they are wearing helmets. However, traditional management systems usually rely on manual inspections to monitor the safety of construction sites, requiring dedicated inspectors to conduct frequent inspections. This method is inefficient and prone to errors, and the investment in human resources may be high for large construction sites. Not only that, manual inspections are easily affected by human factors, such as fatigue and lack of concentration, which may lead to a decrease in the accuracy of monitoring results. In addition, traditional systems can usually only monitor areas that inspectors can reach, and cannot effectively monitor and manage some difficult-to-reach or blind locations. Especially on large-scale construction sites, it is difficult to ensure that every construction worker can be accurately detected.

[0004] Some existing management systems use cameras to monitor and manage construction workers, but the deployment of cameras also has blind spots. In addition, the system often cannot provide real-time monitoring and early warning functions. Once a safety problem occurs, it may take some time to discover it in time and take corresponding measures, which increases the risk of accidents.

[0005] Therefore, an optimized production scheduling and management system for prefabricated buildings is desired. Summary of the invention

[0006] The embodiment of the present invention provides a production scheduling management system and method for prefabricated buildings, which collects monitoring images of on-site construction personnel through cameras; performs image analysis on the monitoring images of the on-site construction personnel to obtain personnel head features; and determines whether to generate an early warning prompt based on the personnel head features. In this way, the problems of low efficiency and low accuracy of manual inspections can be avoided, thereby optimizing the supervision efficiency and supervision level of the user unit and improving construction safety.

[0007] The embodiment of the present invention further provides a production scheduling management system for prefabricated buildings, which includes:

[0008] A monitoring image acquisition module is used to collect monitoring images of on-site construction personnel through cameras;

[0009] A monitoring image analysis module, used for performing image analysis on the monitoring image of the on-site construction personnel to obtain the personnel's head features;

[0010] The warning module is used to determine whether to generate a warning prompt based on the head features of the person.

[0011] The embodiment of the present invention also provides a production scheduling management method for prefabricated buildings, which includes:

[0012] Collect monitoring images of on-site construction workers through cameras;

[0013] Performing image analysis on the monitoring image of the on-site construction personnel to obtain personnel head features;

[0014] Based on the head features of the person, determine whether to generate a warning prompt. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0016] Figure 1 A block diagram of a production scheduling and management system for prefabricated buildings provided in an embodiment of the present invention.

[0017] Figure 2 The present invention provides a block diagram of a monitoring image analysis module in a production scheduling and management system for prefabricated buildings according to an embodiment of the present invention.

[0018] Figure 3 The present invention provides a flowchart of a production scheduling management method for prefabricated buildings in an embodiment of the present invention.

[0019] Figure 4 A schematic diagram of a system architecture of a production scheduling management method for prefabricated buildings provided in an embodiment of the present invention.

[0020] Figure 5 This is an application scenario diagram of a production scheduling management system for prefabricated buildings provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0022] Unless otherwise specified, all technical and scientific terms used in the embodiments of the present application have the same meaning as those commonly understood by those skilled in the art of the present application. The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit the scope of this application.

[0023] In the embodiments of the present application, it should be noted that, unless otherwise specified and limited, the term "connection" should be understood in a broad sense. For example, it can be an electrical connection or a connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0024] It should be noted that the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that the specific order or sequence of "first\second\third" can be interchanged where permitted. It should be understood that the objects distinguished by "first\second\third" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0025] It should be understood that prefabricated construction is a construction method that prefabricates building components in a factory and then transports them to the site for assembly and installation. Compared with traditional on-site construction methods, prefabricated construction has the following characteristics and advantages:

[0026] 1. Short construction period: As prefabricated buildings are prefabricated in factories and assembled on site, the construction time can be greatly shortened. Prefabricated components are processed in the factory at the same time, while the foundation and base are constructed on the construction site. The two can be carried out in parallel, thus reducing the time of the entire project.

[0027] 2. Controllable quality: Components prefabricated in the factory are subject to strict quality control to ensure the accuracy and quality of the components. Compared with on-site construction, the components of prefabricated buildings are more accurate and can provide higher quality assurance.

[0028] 3. Resource conservation: Prefabricated buildings adopt modular design and production methods, which can effectively utilize materials and resources. When prefabricated in the factory, material waste can be minimized and materials can be recycled.

[0029] 4. Environmentally friendly: As prefabricated buildings can reduce the time and noise of on-site construction, they have less impact on the surrounding environment. In addition, environmentally friendly materials can be used in the prefabrication process, and waste materials can be recycled and reused, reducing the consumption of natural resources.

[0030] 5. Flexibility and reusability: The components of prefabricated buildings can be standardized and modularized, making the building design more flexible and can be combined and adjusted according to needs. In addition, the components of prefabricated buildings can be disassembled and reassembled, and have high reusability.

[0031] Prefabricated buildings are widely used in residential, commercial, office and public facilities. It can not only improve the efficiency and quality of construction, but also meet the needs of rapid construction and sustainable development. With the continuous advancement and innovation of technology, prefabricated buildings will continue to develop in the future and become one of the important trends in the construction industry.

[0032] Monitoring and management of construction workers wearing hard hats is crucial to the safety of construction sites. Hard hats are important protective equipment for construction workers to protect their heads from falling objects, collisions, impacts, etc. Wearing hard hats can effectively reduce the risk of head injuries and protect the life safety and health of construction workers.

[0033] In many countries and regions, wearing hard hats on construction sites is a legal requirement. Monitoring and managing whether construction workers wear hard hats can ensure that companies and construction parties comply with relevant laws and regulations and avoid fines, penalties or legal disputes due to violations.

[0034] At the construction site, there may be dangerous factors such as falling objects, high-altitude operations, and mechanical equipment operation. Wearing a hard hat can reduce the risk of head injuries and the probability of accidents. Monitoring and managing whether construction workers are wearing hard hats can timely detect the situation of not wearing hard hats and take measures to prevent accidents.

[0035] The traditional manual inspection method requires special inspectors to conduct frequent inspections, which is inefficient and prone to errors. Through the system that monitors and manages whether construction workers are wearing helmets, automated monitoring and management can be achieved, improving the efficiency and accuracy of construction management.

[0036] Monitoring and managing whether construction workers wear helmets can strengthen the cultivation of safety awareness and safety culture. By enforcing the wearing of helmets and monitoring them, construction workers can be reminded to always pay attention to safety and form good safety behavior habits.

[0037] Monitoring and managing whether construction workers wear hard hats is crucial to ensuring the safety of construction sites. It can protect the life and health of construction workers, comply with laws and regulations, reduce accident risks, improve construction management efficiency, and shape a good safety culture.

[0038] In one embodiment of the present invention, Figure 1 FIG. 1 is a block diagram of a production scheduling management system for prefabricated buildings provided in an embodiment of the present invention. Figure 1 As shown, a production scheduling and management system 100 for prefabricated buildings according to an embodiment of the present invention includes: a monitoring image acquisition module 110, which is used to acquire monitoring images of on-site construction personnel through a camera; a monitoring image analysis module 120, which is used to perform image analysis on the monitoring images of the on-site construction personnel to obtain personnel head features; and an early warning module 130, which is used to determine whether to generate an early warning prompt based on the personnel head features.

[0039] In the monitoring image acquisition module 110, ensure that the position and angle of the camera can fully cover the construction site to obtain accurate monitoring images. At the same time, consider the waterproof, dustproof and durability of the camera to adapt to various construction environments. Through the monitoring image acquisition module, the monitoring images of the construction workers can be obtained in real time, providing a data basis for subsequent image analysis. In this way, real-time monitoring and management of the construction workers wearing safety helmets can be achieved.

[0040] In the monitoring image analysis module 120, the image processing and analysis algorithm is designed and optimized to accurately identify and extract the head features of the construction workers, such as faces, helmets, etc. The algorithm is trained and optimized considering the influence of factors such as lighting, angles, and occlusions on image analysis. Through the monitoring image analysis module, the monitoring images of the construction workers can be processed and analyzed to extract the head features. In this way, the automatic identification and monitoring of the wearing of safety helmets can be realized.

[0041] In the early warning module 130, the triggering conditions and prompting methods of the early warning are determined according to the image analysis results. The early warning prompt can be carried out in the form of sound, light, alarm, etc., and its timeliness and effectiveness need to be considered so that the construction personnel can take timely actions. The early warning module can determine whether there are construction personnel who are not wearing safety helmets based on the analysis results of the personnel's head features, and generate corresponding early warning prompts. In this way, construction personnel can be reminded to wear safety helmets in time to reduce the risk of accidents.

[0042] The production scheduling management system 100 for prefabricated buildings can monitor and manage the wearing of helmets by construction workers in real time, improve safety awareness and sense of responsibility, and reduce the risk of accidents. Automated image processing and analysis can be performed to improve the accuracy and efficiency of monitoring and reduce human resources and time costs. Construction workers who are not wearing helmets can be warned and warned in time to prompt them to correct their behavior in time and further reduce the possibility of accidents. Data recording and analysis can be performed to help improve construction management and safety measures and improve the safety level of the construction site. It can also shape a safety culture, strengthen the cultivation of safety awareness and safety behavior, and promote the formation of safety behavior habits among construction workers.

[0043] Specifically, the monitoring image acquisition module 110 is used to collect monitoring images of on-site construction personnel through a camera. In response to the above technical problems, the technical concept of the present application is to collect monitoring images of on-site construction personnel through a camera, and use image processing and analysis algorithms at the back end to analyze the monitoring images of on-site construction personnel to automatically monitor and manage whether safety helmets are worn in real time, and to promptly warn construction personnel who do not wear safety helmets to ensure the safety of construction personnel. In this way, the problems of low efficiency and low accuracy of manual inspections can be avoided, thereby optimizing the supervision efficiency and supervision level of the user unit and improving construction safety.

[0044] In this application, through image processing and analysis algorithms, it is possible to monitor in real time whether construction workers are wearing helmets. The system can analyze the monitoring images in real time, identify construction workers who are not wearing helmets, and issue timely warnings, which can greatly improve the real-time and accuracy of monitoring and detect violations of wearing helmets in a timely manner.

[0045] Using image processing and analysis algorithms to monitor helmet wearing can achieve automated monitoring and management. Compared with traditional manual inspection methods, automated monitoring can not only improve efficiency, but also reduce human resources and time costs. The system can monitor around the clock, all day, without time and manpower restrictions. Image processing and analysis algorithms can accurately identify and analyze helmets in monitoring images. By training and optimizing algorithms, the accuracy and reliability of identification can be improved, and misjudgment and missed detection can be reduced. This ensures accurate monitoring and management of helmet wearing requirements.

[0046] Once the monitoring system identifies a construction worker who is not wearing a helmet, it can immediately issue an early warning, such as sound, light or alarm, to remind the construction worker to wear a helmet. This real-time early warning and warning can help construction workers correct their behavior in a timely manner and reduce the risk of accidents. The monitoring system can record and store monitoring images and related data, including the wearing of helmets, the number and frequency of violations, etc. These data can be used for post-analysis and statistics to help construction parties and management departments understand the safety behaviors and trends of construction workers and further improve construction management and safety measures.

[0047] This automated monitoring system can become an important auxiliary tool for construction site safety management, effectively reducing accident risks and improving construction safety levels.

[0048] Specifically, in the technical solution of the present application, first, a monitoring image of the on-site construction personnel collected by a camera is obtained. The monitoring image of the on-site construction personnel can provide the following useful information:

[0049] Personnel head features, monitoring images can capture the construction workers' head features, such as face, helmet, etc. These head features are an important basis for judging whether a helmet is worn.

[0050] Safety helmet recognition: Through monitoring image analysis, the head features of construction workers can be identified and analyzed to determine whether they are wearing safety helmets. If the monitoring image shows that the construction worker is not wearing a safety helmet, an early warning prompt can be triggered.

[0051] Location and behavior information: Monitoring images can provide location and behavior information of construction workers. For example, if a construction worker is found working in a dangerous area without a helmet, an early warning can be generated.

[0052] Abnormal situation detection, monitoring images can be used to detect abnormal situations, such as construction workers suddenly taking off their helmets or wearing the wrong helmets. These abnormal situations can also trigger early warning prompts.

[0053] Monitoring images can provide useful information about construction workers' head features, helmet wearing conditions, location and behavior information, and abnormal conditions. After image analysis and processing, this information can be used to determine whether to generate an early warning. By analyzing monitoring images, the system can monitor the helmet wearing conditions of construction workers in real time and generate early warnings as needed to improve the safety of the construction site.

[0054] Furthermore, the monitoring images of construction workers on site collected by the camera play a key role in determining whether to generate an early warning prompt. The monitoring images provide real-time construction site information, and the image analysis module processes and analyzes these images to determine whether the construction workers are wearing safety helmets. Based on the results of the image analysis, the early warning module can determine whether there are construction workers who are not wearing safety helmets and generate corresponding early warning prompts.

[0055] Specifically, the monitoring image acquisition module uses the camera to obtain the monitoring images of the construction workers in real time, including their head features. These images are transmitted to the monitoring image analysis module for processing and analysis. The image analysis module uses image processing and analysis algorithms to identify and extract the head features of the construction workers and determine whether they are wearing safety helmets.

[0056] Based on the results of the image analysis, the early warning module can determine whether to generate an early warning prompt. If the analysis results show that a construction worker is not wearing a helmet, the early warning module will trigger the early warning mechanism and generate a corresponding early warning prompt, such as sound, light or alarm, to remind the construction worker to wear a helmet. If the analysis results show that all construction workers are wearing helmets, the early warning module will not generate an early warning prompt.

[0057] Therefore, the surveillance images collected by the camera are an important basis for judging whether the construction workers are wearing helmets, and the image analysis module processes and analyzes these images to ultimately decide whether to generate an early warning. This analysis and judgment based on surveillance images can achieve real-time monitoring and management of the wearing of helmets by construction workers, improving the safety of the construction site.

[0058] Specifically, the monitoring image analysis module 120 is used to perform image analysis on the monitoring image of the on-site construction personnel to obtain the personnel's head features. Figure 2 : is a block diagram of the monitoring image analysis module in a production scheduling management system for prefabricated buildings provided in an embodiment of the present invention, such as Figure 2 As shown, the monitoring image analysis module 120 includes: a person head target framing unit 121, which is used to pass the monitoring image through a person head area target detection network to obtain an image of the person head area of ​​interest; a person head local feature extraction unit 122, which is used to pass the image of the person head area of ​​interest through an image feature extractor based on a backbone network to obtain a person head local feature map; a receptive field expansion unit 123, which is used to perform receptive field expansion processing on the person head local feature map to obtain a person head global feature map as the person head feature.

[0059] First, the surveillance image is processed through the person's head area target detection network to locate and extract the image of the area of ​​interest of the person's head. In this way, the position of the person's head is accurately found through the target detection technology, providing accurate input for the subsequent head feature extraction.

[0060] Then, the image of the region of interest of the person's head is processed by the image feature extractor based on the backbone network to extract the local feature map of the person's head. In this way, local features related to the head features, such as the face outline, helmet and other information, can be extracted from the image of the region of interest of the head.

[0061] Next, the receptive field expansion process is performed on the local feature map of the person's head to obtain the global feature map of the person's head. Receptive field expansion is an image processing technology that can capture broader contextual information by introducing a larger receptive field in the local feature map. In this way, the contextual information of the head features can be enhanced and the global expression ability of the features can be improved.

[0062] The monitoring image analysis module can extract the characteristic information of the person's head from the monitoring image, including the position, local features and global features of the head, which are very important for judging the wearing of the helmet and detecting abnormal conditions. Through precise target positioning, local feature extraction and global feature enhancement, the accuracy and robustness of the system for helmet wearing can be improved, thereby providing a reliable basis for generating early warning prompts.

[0063] Among them, the person head area target detection network is an anchor window-based target detection network, and the anchor window-based target detection network is Fast R-CNN, Faster R-CNN or RetinaNet.

[0064] Next, considering that when actually detecting whether the on-site construction workers are wearing helmets, the focus should be on the implicit feature information of the construction workers' head area. Therefore, if it is possible to filter out useless interference features in the area when detecting whether the on-site construction workers are wearing helmets, it is obviously possible to improve the accuracy of the detection. Based on this, in the technical solution of the present application, it is necessary to further pass the monitoring image through a personnel head area target detection network to obtain an image of the personnel head area of ​​interest, so as to frame the area of ​​interest of the construction workers' heads in the monitoring image of the on-site construction workers, thereby facilitating the subsequent detection of the construction workers wearing helmets. Accuracy.

[0065] It should be understood that the human head area target detection network is a deep learning model that aims to detect the human head area in the monitoring image and frame it as the area of ​​interest, so as to improve the accuracy of subsequent detection of construction workers wearing safety helmets.

[0066] This object detection network is usually based on the architecture of Convolutional Neural Network (CNN), such as Faster R-CNN, YOLO (You Only Look Once) or SSD (Single Shot MultiBoxDetector). These network structures have powerful object detection capabilities and can effectively locate and identify objects in images.

[0067] The head area target detection network is trained with a large amount of labeled data to learn the feature representation and positioning information of the head. It can automatically identify the head in the image and generate a bounding box to frame the head area as the region of interest. This region of interest image can be further used to extract local and global features of the head to determine whether the construction worker is wearing a helmet.

[0068] The head area target detection network requires a large number of labeled image datasets during training, which include the labeled information of the head area of ​​people at the construction site. Through repeated iterative training, the network can gradually optimize its own parameters to improve the ability to accurately detect people's heads.

[0069] The head area target detection network is a deep learning model used to detect the head area of ​​a person in a surveillance image. It can learn the feature representation and positioning information of a person's head through training, and generate images of the area of ​​interest, providing accurate input for subsequent helmet wearing detection.

[0070] Furthermore, the backbone network is a convolutional neural network model. Then, a convolutional neural network model with excellent performance in extracting local implicit features of images is used to perform feature mining of the image of the region of interest of the person's head. That is to say, specifically, the image of the region of interest of the person's head is passed through an image feature extractor based on a backbone network to obtain a local feature map of the person's head. In particular, here, the backbone network is a convolutional neural network model, so as to extract the local detail implicit feature distribution information about the construction worker's head in the image of the region of interest of the person's head, such as the contour, color, shape and texture of the head, which is conducive to fully expressing the characteristics of whether the construction worker's head is wearing a safety helmet.

[0071] Among them, the image feature extractor based on the backbone network is a deep learning model that can extract high-level feature representations in images after training. By inputting the image of the area of ​​interest of the person's head into the feature extractor, local features of the head area can be extracted, such as the face contour, helmet, etc. These local features are very important for judging whether the construction workers are wearing safety helmets.

[0072] The image feature extractor based on the backbone network has strong expressive power and can convert the input image into a high-dimensional feature representation. The local feature map of the person's head can capture the detailed information of the head area, such as texture, shape, etc. These features are very useful for subsequent feature analysis and classification tasks.

[0073] By extracting the local feature map of the person's head through the backbone network-based image feature extractor, the dimension of the original image data can be reduced to a lower-dimensional feature representation. This helps reduce the computational complexity of subsequent processing and improves computational efficiency.

[0074] That is, by processing the image of the area of ​​interest of the person's head through the backbone network-based image feature extractor, the local feature map of the head can be extracted. The local feature map contains important information of the head area, which is helpful for the subsequent accurate analysis and processing of helmet wearing detection and other related tasks.

[0075] In one embodiment of the present application, the receptive field expansion unit 123 is used to: pass the local feature map of the person's head through an image feature pixel global sensor based on a non-local neural network model to obtain the global feature map of the person's head.

[0076] Furthermore, considering that convolution is a typical local operation, it can only extract local features of the image, but cannot pay attention to the overall situation, which will affect the accuracy of detection. Therefore, in the technical solution of the present application, in order to more accurately detect whether the construction workers are wearing helmets, a non-local neural network is used to further enhance the image feature receptive field. That is, the local feature map of the head of the person is passed through the global sensor of the image feature pixels based on the non-local neural network model to expand the feature receptive field through the non-local neural network model, thereby obtaining a global feature map of the head of the person. In particular, here, the non-local neural network model captures hidden dependency information by calculating the similarity between the implicit feature distribution information of each local head region in the image of the region of interest of the head of the person, and then models contextual features, so that the network pays attention to the global overall content between the features of each local head region, thereby improving the feature extraction capability of the backbone network in classification and detection tasks.

[0077] In this application, the non-local neural network model is a deep learning model for image processing and computer vision tasks, whose main goal is to achieve global feature perception and extraction. Compared with the traditional local perception model, the non-local neural network model can capture broader semantic relationships through global context information, thereby improving the expressiveness and discrimination ability of features.

[0078] The core idea of ​​the non-local neural network model is to introduce non-local operations to model the relationship between different positions in the image. This operation can perform adaptive weight calculations between local features, so that each feature pixel can perceive global information.

[0079] Specifically, the non-local neural network model uses non-local blocks as the basic components of the network. The non-local block includes two main steps: calculating similarity and weighted summation. In the similarity calculation step, the degree of association between two feature pixels is measured by calculating the similarity between them. Then, in the weighted summation step, all feature pixels are weighted summed according to the results of the similarity calculation to obtain a global feature representation.

[0080] The advantage of the non-local neural network model is that it can capture global context information, thereby improving the expressiveness of features. For the processing of the local feature map of the person's head, the relationship between each pixel in the local feature map and other pixels can be modeled through the global perceptron of image feature pixels based on the non-local neural network model, thereby obtaining a global feature map of the person's head. This global feature map can better represent the overall features of the head, including the shape, texture and other information of the head, which is helpful for subsequent feature analysis and helmet wearing detection tasks.

[0081] Specifically, the warning module 130 is used to determine whether to generate a warning prompt based on the head features of the personnel. It includes: a feature distribution optimization unit, which is used to optimize the feature distribution of the global feature map of the personnel head to obtain an optimized global feature map of the personnel head; a safety helmet wearing detection unit, which is used to pass the optimized global feature map of the personnel head through a classifier to obtain a classification result, and the classification result is used to indicate whether the on-site construction personnel wear a safety helmet; a warning prompt generation determination unit, which is used to determine whether to generate a warning prompt based on the classification result.

[0082] The process of optimizing the feature distribution of the global feature map of the person's head aims to improve the discriminability and expressiveness of the features to better distinguish features of different categories. By optimizing the feature distribution, the features related to the helmet wearing status can be made more obvious and prominent, which is helpful for the subsequent helmet wearing detection task.

[0083] The optimized global feature map of the head of the person is input into the classifier, and the discrimination ability of the classifier can be used to determine whether the construction personnel on site are wearing a helmet. The classifier can be a model based on machine learning or deep learning. After training, it can learn the relationship between different features and the wearing status of the helmet, and perform classification prediction on the input feature map.

[0084] Based on the classification results, it can be determined whether to generate an early warning prompt. If the classification results indicate that a construction worker is not wearing a safety helmet, the system can trigger a corresponding early warning mechanism, such as issuing a sound alarm, sending a notification to relevant personnel, etc., to remind and warn on-site managers and workers to take timely measures to ensure the safety of the construction site. In particular, in the technical solution of the present application, when the local feature map of the person's head is passed through a global sensor of image feature pixels based on a non-local neural network model, it is expected that the global feature map of the person's head, while expressing the global image semantic features based on the local feature map of the person's head, still has a good expression of the local image semantic features of the image of the region of interest of the person's head extracted by the image feature extractor based on the backbone network in the classification scenario. Therefore, it is necessary to correct the global feature map of the person's head based on the local image semantic feature representation of the local feature map of the person's head.

[0085] Based on this, the applicant of the present application performs smooth response parameter decoupling and fusion on the local feature map of the person's head, for example, denoted as F1, and the global feature map of the person's head, for example, denoted as F2, to obtain an optimized global feature map of the person's head, for example, denoted as F2′, specifically: the local feature map of the person's head and the global feature map of the person's head are subjected to smooth response parameter decoupling and fusion using the following fusion formula to obtain the optimized global feature map of the person's head; wherein the fusion formula is:

[0086]

[0087] Wherein, F1 and F2 are the local feature map of the person's head and the global feature map of the person's head, respectively; cov(F1, F2) represents the cosine distance between the local feature map of the person's head and the global feature map of the person's head; log represents the logarithmic function value with base 2; exp(·) represents the exponential operation of the feature map, and the exponential operation of the feature map represents the natural exponential function value raised to the power of the feature value of each position in the feature map; represents positional subtraction, represents positional addition, ⊙ represents positional dot multiplication, and F2′ is the global feature map of the optimized person’s head.

[0088] Here, the smooth response parameterized decoupling fusion compiles the point-by-point embedding between the features of the local feature map F1 of the personnel head and the global feature map F2 of the personnel head based on the non-negative symmetry of the cosine distance between the local feature map F1 of the personnel head and the global feature map F2 of the personnel head by using the decoupling principle of the smooth parameterized function, so as to infer the information distribution shift between the local feature map F1 of the personnel head and the global feature map F2 of the personnel head by the spatial transformation between the features, so as to express the information structured fusion of the smooth response between the features under the classification rules, so as to improve the expression effect of the optimized global feature map F2′ of the personnel head for the local image semantic features of the local feature map of the personnel head based on the classification rules, so as to improve the accuracy of the classification results obtained by the optimized global feature map of the personnel head through the classifier. In this way, it is possible to monitor and manage whether the on-site construction personnel wear safety helmets in real time, and to generate early warning prompts for construction personnel who do not wear safety helmets in time to ensure the safety of construction personnel, thereby optimizing the supervision efficiency and supervision level of the user unit and improving construction safety.

[0089] In one embodiment of the present application, the helmet wearing detection unit includes: a matrix expansion subunit, used to expand the optimized personnel head global feature map into a classification feature vector according to a row vector or a column vector; a fully connected encoding subunit, used to use multiple fully connected layers of the classifier to fully connect the classification feature vector to obtain an encoded classification feature vector; and a classification subunit, used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0090] Then, the global feature map of the personnel's head is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the on-site construction personnel are wearing safety helmets. In other words, the global associated feature information of the construction personnel's head is used for classification processing, so as to automatically monitor and manage whether safety helmets are worn in real time, and to provide timely early warning prompts to construction personnel who are not wearing safety helmets to ensure the safety of construction personnel. That is, in the technical solution of the present application, it is determined whether to generate an early warning prompt based on the classification result. In this way, the problems of low efficiency and low accuracy of manual inspections can be avoided, thereby optimizing the supervision efficiency and supervision level of the user unit and improving construction safety.

[0091] In summary, a production scheduling and management system 100 for prefabricated buildings based on an embodiment of the present invention is explained. Monitoring images of on-site construction workers are collected by cameras, and image processing and analysis algorithms are used at the back end to perform monitoring image analysis of on-site construction workers to automatically perform real-time monitoring and management of whether safety helmets are worn, and timely early warning prompts are given to construction workers who are not wearing safety helmets to ensure the safety of construction workers. In this way, the problems of low efficiency and low accuracy of manual inspections can be avoided, thereby optimizing the supervision efficiency and level of the user unit and improving construction safety.

[0092] As described above, the production scheduling management system 100 for prefabricated buildings according to an embodiment of the present invention can be implemented in various terminal devices, such as a server for production scheduling management of prefabricated buildings. In one example, the production scheduling management system 100 for prefabricated buildings according to an embodiment of the present invention can be integrated into a terminal device as a software module and / or a hardware module. For example, the production scheduling management system 100 for prefabricated buildings can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the production scheduling management system 100 for prefabricated buildings can also be one of the many hardware modules of the terminal device.

[0093] Alternatively, in another example, the production scheduling and management system 100 for prefabricated buildings and the terminal device may also be separate devices, and the production scheduling and management system 100 for prefabricated buildings may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0094] Figure 3 The present invention provides a flowchart of a production scheduling management method for prefabricated buildings in an embodiment of the present invention. Figure 4 Schematic diagram of a system architecture of a production scheduling management method for prefabricated buildings provided in an embodiment of the present invention. Figure 3 and Figure 4 As shown, a production scheduling management method for prefabricated buildings includes: 210, collecting monitoring images of on-site construction personnel through a camera; 220, performing image analysis on the monitoring images of the on-site construction personnel to obtain personnel head features; 230, determining whether to generate an early warning prompt based on the personnel head features.

[0095] Specifically, in the production scheduling management method for prefabricated buildings, image analysis is performed on the monitoring image of the on-site construction personnel to obtain the personnel head features, including: passing the monitoring image through a personnel head area target detection network to obtain an image of the personnel head area of ​​interest; passing the image of the personnel head area of ​​interest through an image feature extractor based on a backbone network to obtain a local feature map of the personnel head; and performing receptive field expansion processing on the local feature map of the personnel head to obtain a global feature map of the personnel head as the personnel head features.

[0096] Those skilled in the art will appreciate that the specific operations of each step in the above-mentioned production scheduling management method for prefabricated buildings have been described in detail above. Figure 1 to Figure 2 The description of the production scheduling management system for prefabricated buildings has been introduced in detail, and therefore, its repeated description will be omitted.

[0097] Figure 5 FIG. 1 is an application scenario diagram of a production scheduling management system for prefabricated buildings provided in an embodiment of the present invention. Figure 5 As shown, in this application scenario, first, a monitoring image of the on-site construction workers is collected by a camera (for example, Figure 5 Then, the acquired monitoring image is input to a server (for example, Figure 5 In S) shown in , the server is capable of processing the monitoring image based on a production scheduling management algorithm for prefabricated buildings to determine whether to generate an early warning prompt.

[0098] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A production scheduling management system for prefabricated buildings, characterized in that: include: A monitoring image acquisition module is used to collect monitoring images of on-site construction personnel through cameras; A monitoring image analysis module, used for performing image analysis on the monitoring image of the on-site construction personnel to obtain the personnel's head features; The warning module is used to determine whether to generate a warning prompt based on the head features of the person.

2. The production scheduling and management system for prefabricated buildings according to claim 1, characterized in that: The monitoring image analysis module comprises: A person head target framing unit, used for passing the monitoring image through a person head area target detection network to obtain a person head area of ​​interest image; A person's head local feature extraction unit, used for passing the image of the person's head region of interest through an image feature extractor based on a backbone network to obtain a person's head local feature map; The receptive field expansion unit is used to perform receptive field expansion processing on the local feature map of the person's head to obtain a global feature map of the person's head as the person's head feature.

3. The production scheduling and management system for prefabricated buildings according to claim 2, characterized in that: The person head area target detection network is an anchor window-based target detection network, and the anchor window-based target detection network is Fast R-CNN, Faster R-CNN or RetinaNet.

4. The production scheduling and management system for prefabricated buildings according to claim 3, characterized in that: The backbone network is a convolutional neural network model.

5. The production scheduling and management system for prefabricated buildings according to claim 4, characterized in that: The receptive field expansion unit is used to: obtain the global feature map of the person's head by passing the local feature map of the person's head through an image feature pixel global sensor based on a non-local neural network model.

6. The production scheduling and management system for prefabricated buildings according to claim 5, characterized in that: The early warning module comprises: A feature distribution optimization unit, used for optimizing the feature distribution of the global feature map of the person's head to obtain an optimized global feature map of the person's head; A safety helmet wearing detection unit, used for passing the optimized personnel head global feature map through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the on-site construction personnel are wearing safety helmets; The early warning prompt generation determination unit is used to determine whether to generate an early warning prompt based on the classification result.

7. The production scheduling and management system for prefabricated buildings according to claim 6, characterized in that: The feature distribution optimization unit is used to: perform smooth response parameter decoupling and fusion on the local feature map of the person's head and the global feature map of the person's head according to the following fusion formula to obtain the optimized global feature map of the person's head; Among them, the fusion formula is: Wherein, F1 and F2 are the local feature map of the person's head and the global feature map of the person's head, respectively; cov(F1, F2) represents the cosine distance between the local feature map of the person's head and the global feature map of the person's head; log represents the logarithmic function value with base 2; exp(·) represents the exponential operation of the feature map, and the exponential operation of the feature map represents the natural exponential function value raised to the power of the feature value of each position in the feature map; represents positional subtraction, represents positional addition, ⊙ represents positional dot multiplication, and F2′ is the global feature map of the optimized person’s head.

8. The production scheduling and management system for prefabricated buildings according to claim 7, characterized in that: The helmet wearing detection unit comprises: A matrix expansion subunit, used for expanding the optimized person's head global feature map into a classification feature vector according to a row vector or a column vector; a fully connected encoding subunit, configured to perform fully connected encoding on the classification feature vector using a plurality of fully connected layers of the classifier to obtain an encoded classification feature vector; and The classification subunit is used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

9. A production scheduling management method for prefabricated buildings, characterized in that: include: Collect monitoring images of on-site construction workers through cameras; Performing image analysis on the monitoring image of the on-site construction personnel to obtain personnel head features; Based on the head features of the person, determine whether to generate a warning prompt.

10. The production scheduling management method for prefabricated buildings according to claim 9, characterized in that: Performing image analysis on the monitoring image of the on-site construction personnel to obtain personnel head features includes: The monitoring image is passed through a person's head area target detection network to obtain an image of a region of interest in the person's head; The image of the region of interest of the person's head is passed through an image feature extractor based on a backbone network to obtain a local feature map of the person's head; The receptive field expansion process is performed on the local feature map of the person's head to obtain a global feature map of the person's head as the person's head feature.

Images