CN115861904A - Method and system for generating slag car roof fall detection model - Google Patents
Method and system for generating slag car roof fall detection model Download PDFInfo
- Publication number
- CN115861904A CN115861904A CN202310154326.9A CN202310154326A CN115861904A CN 115861904 A CN115861904 A CN 115861904A CN 202310154326 A CN202310154326 A CN 202310154326A CN 115861904 A CN115861904 A CN 115861904A
- Authority
- CN
- China
- Prior art keywords
- detection model
- data set
- muck
- car
- preset
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 177
- 239000002893 slag Substances 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000012549 training Methods 0.000 claims abstract description 72
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 15
- SAZUGELZHZOXHB-UHFFFAOYSA-N acecarbromal Chemical compound CCC(Br)(CC)C(=O)NC(=O)NC(C)=O SAZUGELZHZOXHB-UHFFFAOYSA-N 0.000 claims abstract description 8
- 238000005520 cutting process Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 7
- 238000013519 translation Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 230000009471 action Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000004927 fusion Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Refinement Of Pig-Iron, Manufacture Of Cast Iron, And Steel Manufacture Other Than In Revolving Furnaces (AREA)
Abstract
The embodiment of the application provides a method and a system for generating a muck vehicle roof fall detection model, and relates to the technical field of target detection. The method for generating the slag car roof fall detection model comprises the following steps: acquiring training video data; generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate; constructing a preset detection model, wherein the preset detection model is a two-stage detection model, rough target detection is carried out through a preset yolov5 target detection algorithm in one stage, and fine classification is carried out by inputting a detection target into a resnet18 in the second stage; and performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model. The method for generating the slag car roof fall detection model can achieve the technical effect of improving the slag car roof fall detection efficiency.
Description
Technical Field
The application relates to the technical field of target detection, in particular to a method and a system for generating a detection model for roof fall of a muck truck, an electronic device and a computer-readable storage medium.
Background
At present, the cleaning and transportation of construction waste is a problem that the urban content management is not negligible, a muck truck is the main force for the transportation of the construction waste and makes an important contribution to the urban muck transportation, however, some unavoidable problems of the management of the muck truck exist, a large number of muck trucks are not washed and directly driven out of a construction site, and a large amount of soil brought by wheels brings about no small pollution to urban roads. Some muck trucks are overloaded seriously, the top ends of the muck trucks are not sealed, and the muck falls on the ground along with the bumping of one road. In addition, the construction side does not have a question about the overload phenomenon of the muck car, and the muck car is released and flows, so that the muck car is more likely to be pulled and run quickly, the transportation is illegal, and the urban environment is seriously polluted. The muck truck roof fall is sharp, roof canopy cover is not tight always to be the common problem in the muck truck management, some muck truck drivers just fill up the inside with sand and stone earth for a few trips of pulling a little more at every turn, open and spill all the way, cause environmental pollution serious, it is not in place by or muck truck management to pursue its final source in fact, current muck truck roof fall management relies on artifical management and control entirely, do not have any automatic means, have serious undetected, the false retrieval phenomenon, the action phenomenon of muck truck roof fall can't be solved. The existing slag car roof fall detection is against manual visual identification, and a lot of false reports and false reports exist for colleagues who consume time and labor.
Disclosure of Invention
An object of the embodiments of the present application is to provide a method, a system, an electronic device, and a computer-readable storage medium for generating a muck truck roof fall detection model, which can achieve the technical effect of improving the detection efficiency of the muck truck roof fall.
In a first aspect, an embodiment of the present application provides a method for generating a muck truck roof fall detection model, including:
acquiring training video data;
generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate;
constructing a preset detection model, wherein the preset detection model is a two-stage detection model, rough target detection is carried out through a preset yolov5 target detection algorithm in one stage, and fine classification is carried out by inputting a detection target into a resnet18 in the second stage;
and performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model.
In the implementation process, the method for generating the slag car roof fall detection model is based on multi-algorithm fusion, the targets are roughly detected through a preset yolov5 target detection algorithm in one stage, the detection targets are input into the resnet18 in the second stage to be classified in a subdivided mode, the model training of the preset detection model can be completed under a small number of samples, and the slag car roof fall detection model is obtained; therefore, the method for generating the slag car roof fall detection model can achieve the technical effect of improving the slag car roof fall detection efficiency.
Further, before the step of constructing the preset detection model, the method further comprises:
and performing data enhancement on the image data set of the slag car through a preset generated countermeasure network GAN to obtain an expanded image data set of the slag car.
In the implementation process, the preset generation confrontation network GAN is used for data enhancement, and the number of pictures of the muck car picture data set can be expanded, so that the model training efficiency and the training accuracy are improved.
Further, the step of performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model comprises:
adjusting hue and saturation of the expanded muck car picture data set to obtain a first expanded muck car picture data set;
adding one or more operations of random zooming, cutting, translation, cutting and rotation to the first expansion muck car picture data set to obtain a second expansion muck car picture data set;
and performing iterative training on the preset detection model through the second expansion slag car picture data set, wherein the epoch of the preset detection model reaches a preset value, and obtaining a slag car roof fall detection model.
Further, before the step of performing iterative training on the preset detection model through the second expansion slag car picture data set, and obtaining a slag car roof fall detection model when the epoch of the preset detection model reaches a preset value, the method further includes:
and randomly selecting two samples from the second expansion slag car image data set for random weighted summation, wherein the labels of the samples correspond to the weighted summation.
Further, after the step of performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model, the method further includes:
extracting frame rate information of a video to be detected and carrying out data preprocessing to obtain a data set of a picture to be detected;
and detecting the data set of the picture to be detected according to the slag car roof fall detection model to obtain a prediction reasoning result.
In a second aspect, an embodiment of the present application further provides a system for generating a slag car roof fall detection model, including:
the training video acquisition module is used for acquiring training video data;
the training picture generation module is used for generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate;
the model construction module is used for constructing a preset detection model, the preset detection model is a two-stage detection model, rough target detection is carried out through a preset yolov5 target detection algorithm in one stage, and fine classification is carried out by inputting detection targets into a resnet18 in the second stage;
and the model training module is used for performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model.
Further, the generation system of the slag car roof fall detection model further comprises:
and the data enhancement module is used for performing data enhancement on the image data set of the muck car through a preset generated confrontation network GAN to obtain an expanded image data set of the muck car.
Further, the training module comprises:
the first processing unit is used for adjusting the hue and the saturation of the expansion muck car picture data set to obtain a first expansion muck car picture data set;
the second processing unit is used for adding one or more operations of random zooming, cutting, translation, cutting and rotation to the first expansion muck car picture data set to obtain a second expansion muck car picture data set;
and the training unit is used for performing iterative training on the preset detection model through the second expansion slag car picture data set, and acquiring a slag car roof fall detection model when the epoch of the preset detection model reaches a preset value.
Further, the training module further comprises:
and the third processing unit is used for randomly selecting two samples from the second expansion slag car picture data set for random weighted summation, and labels of the samples correspond to the weighted summation.
Further, the generation system of the slag car roof fall detection model further comprises a prediction and inference module, and the prediction and inference module is used for: extracting frame rate information of a video to be detected and carrying out data preprocessing to obtain a data set of a picture to be detected; and detecting the image data set to be detected according to the slag car roof fall detection model to obtain a prediction inference result.
In a third aspect, an electronic device provided in an embodiment of the present application includes: memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the method according to any of the first aspect when executing the computer program.
In a fourth aspect, embodiments of the present application provide a computer-readable storage medium having instructions stored thereon, which when executed on a computer, cause the computer to perform the method according to any one of the first aspect.
In a fifth aspect, embodiments of the present application provide a computer program product, which when run on a computer, causes the computer to perform the method according to any one of the first aspect.
Additional features and advantages of the disclosure will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the above-described techniques.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a schematic flow chart of a method for generating a slag car roof fall detection model according to an embodiment of the present application;
FIG. 2 is a schematic flow chart illustrating another method for generating a model for detecting roof fall of a slag car according to an embodiment of the present disclosure;
fig. 3 is a block diagram of a system for generating a detection model for roof fall of a muck truck according to an embodiment of the present disclosure;
fig. 4 is a block diagram of an electronic device according to an embodiment of the present disclosure.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.
The embodiment of the application provides a method and a system for generating a slag car roof fall detection model, electronic equipment and a computer readable storage medium, which can be applied to the roof fall detection process of a slag car; the method for generating the slag car roof fall detection model is based on multi-algorithm fusion, the rough detection target is carried out through a preset yolov5 target detection algorithm in one stage, the detection target is input into the resnet18 for fine classification in the second stage, the model training of the preset detection model can be completed under a small amount of samples, and the slag car roof fall detection model is obtained; therefore, the method for generating the slag car roof fall detection model can achieve the technical effect of improving the slag car roof fall detection efficiency.
Referring to fig. 1, fig. 1 is a schematic flow chart of a method for generating a slag car roof fall detection model according to an embodiment of the present application, where the method for generating the slag car roof fall detection model includes the following steps:
s100: training video data is acquired.
Illustratively, the training video data may be video data collected over a period of time on a worksite surveillance camera; alternatively, the training video data may be video data of about 5G size collected within one week.
S200: and generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate.
Illustratively, in the step of generating the image data set of the dregs car according to the training video data, invalid image data caused by false touch needs to be removed so as to make the training video data into the image data set of the dregs car; the muck car image data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate, and subsequent model training is facilitated.
S300: and constructing a preset detection model, wherein the preset detection model is a two-stage detection model, one stage carries out rough target detection through a preset yolov5 target detection algorithm, and the two stages carry out fine classification by inputting the detection target into a resnet 18.
Exemplarily, on the premise of target detection with few samples, a preset detection model is constructed in the embodiment of the application, a two-stage detection model is selected, and rough target detection is performed in one stage through a preset yolov5 target detection algorithm to obtain a detection target; the second stage is to input the detection target into the resnet18 for fine classification so as to complete the final classification process; therefore, by the mode, the training of the muck car roof fall detection model can be completed on the premise of a small amount of samples, and the accuracy of the model can be ensured.
S400: and performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model.
In some embodiments, the method for generating the slag car roof fall detection model is based on multi-algorithm fusion, the targets are roughly detected through a preset yolov5 target detection algorithm in one stage, the detection targets are input into the resnet18 for fine classification in the second stage, and model training of the preset detection model can be completed under a small amount of samples to obtain the slag car roof fall detection model; therefore, the method for generating the slag car roof fall detection model can achieve the technical effect of improving the slag car roof fall detection efficiency.
Referring to fig. 2, fig. 2 is a schematic flow chart of another method for generating a slag car roof fall detection model according to an embodiment of the present application.
Exemplarily, at S300: before the step of constructing the preset detection model, the method further comprises the following steps:
s210: and performing data enhancement on the image data set of the muck car through a preset generated countermeasure network GAN to obtain an expanded muck car image data set.
Illustratively, the preset generation countermeasure network GAN is used for data enhancement, and the number of pictures of the muck car picture data set can be expanded, so that the model training efficiency and the training accuracy are improved; for example, the number of the images in the image data set of the muck truck is 2500, and the 2500 pieces of data can be expanded to 10000 pieces of data through the preset generation countermeasure network GAN.
Exemplarily, S400: carrying out iterative training on a preset detection model through a muck car picture data set to obtain a muck car roof fall detection model, comprising the following steps of:
s410: adjusting hue and saturation of the expanded muck car picture data set to obtain a first expanded muck car picture data set;
s420: adding one or more operations of random zooming, cutting, translation, shearing and rotation to the first expansion muck car picture data set to obtain a second expansion muck car picture data set;
s440: and performing iterative training on the preset detection model through the second expansion slag car picture data set, and obtaining the slag car roof fall detection model when the epoch of the preset detection model reaches a preset value.
Illustratively, at S440: before the step of performing iterative training on the preset detection model through the second expansion slag car picture data set and enabling the epoch of the preset detection model to reach the preset value and obtaining the slag car roof fall detection model, the method further comprises the following steps of:
s430: and randomly selecting two samples from the second expansion slag car image data set for random weighted summation, wherein labels of the samples correspond to weighted summation.
Exemplarily, at S400: after the step of iteratively training the preset detection model through the muck car picture data set to obtain the muck car roof fall detection model, the method further comprises the following steps of:
s510: extracting frame rate information of a video to be detected and carrying out data preprocessing to obtain a data set of a picture to be detected;
s520: and detecting the data set of the picture to be detected according to the slag car roof fall detection model to obtain a prediction reasoning result.
Illustratively, the preset yolov5 target detection algorithm provided by the embodiment of the application adds a prediction head for detecting a tiny object on the basis of a yolov5 original edition. In combination with other 3 measuring probes, the 4-head structure can relieve the negative influence caused by severe target scale change. Optionally, the added prediction header (Head 1) is generated by a low-level, high-resolution feature map, which is more sensitive to tiny objects. After the detection head is added, although the calculation and storage cost is increased, the detection performance of the micro object is greatly improved.
Illustratively, the partial volume blocks and CSP bootleneck blocks in yolov5 master may be replaced by a Transformer encoder block. Compared with the original bottleeck blocks in the CSPDarknet53, the Transformer encoder block can capture global information and rich context information.
Illustratively, each Transformer encoder block contains 2 sub-layers. The 1 st sublayer is a multi-head integration layer, and the 2 nd sublayer (MLP) is a fully connected layer. Residual connections are used between each sub-layer. The Transformer encoder block increases the ability to capture different local information. It can also exploit the feature characterization potential with a self-attention mechanism. Based on yolov5, a transducer encoder block was applied only to the head portion to form a Transducer Prediction Head (TPH) and a backbone end. Because the feature map resolution at the end of the network is lower. Applying TPH to low resolution profiles may reduce computation and storage costs. Furthermore, some of the TPH blocks of the earlier layers may be selected for removal when magnifying the resolution of the input image to make the training process available.
Exemplarily, with reference to fig. 1 to fig. 2, a specific flow step of the method for generating a detection model for roof fall of a muck truck provided in the embodiment of the present application is as follows:
step 1, video data with the size of about 5G per week are collected on a construction site supervision camera, invalid picture data caused by false touch are removed, a muck car picture data set is manufactured, the muck car picture data set comprises about 2 pieces of picture data, the collected image resolution is 1X 720, the format is png, and the picture data set is divided into full package, half package, no package and empty car according to the condition of top coverage;
step 2, expanding 2500 pieces of data to 10000 pieces of data by utilizing the generated countermeasure network GAN data enhancement;
step 3, constructing a model: because the invention provides target detection with few samples, a two-stage detection model is selected in the stage, yolov5 is used for rough target detection in the stage, and the detected target is input into the resnet18 for fine classification;
step 4, model training: in the training stage, the invention carries out data enhancement means on the data:
4.1, photometric adjusts the hue and saturation of the image;
step 4.2, the geometric is to add random zooming, clipping, translation, clipping and rotation;
4.3, selecting 2 samples from the training image randomly by the MixUp to carry out random weighted summation, wherein labels of the samples also correspond to the weighted summation;
step 4.4, setting parameters, namely training the model by utilizing 4 2080Ti, setting img _ size to be 640x640, setting batch _sizeto be 16, setting epoch to be 200, and setting various super parameters;
4.5, after the steps are finished, starting iterative training, and outputting and storing the model after the epoch reaches a set value;
and 5, in the stage of prediction inference, the model stored in the step is utilized to perform video inference after the corresponding data preprocessing is performed on the video extraction frame rate.
Referring to fig. 3, fig. 3 is a block diagram of a system for generating a slag car roof fall detection model according to an embodiment of the present application, where the system for generating a slag car roof fall detection model includes:
a training video acquisition module 100, configured to acquire training video data;
the training picture generation module 200 is used for generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate;
the model building module 300 is used for building a preset detection model, the preset detection model is a two-stage detection model, rough target detection is carried out through a preset yolov5 target detection algorithm in one stage, and fine classification is carried out by inputting detection targets into the resnet18 in the second stage;
and the model training module 400 is used for performing iterative training on a preset detection model through the muck car picture data set to obtain a muck car roof fall detection model.
Illustratively, the generation system of the muck truck roof fall detection model further comprises:
and the data enhancement module is used for performing data enhancement on the image data set of the muck car through a preset generated confrontation network GAN to obtain an expanded muck car image data set.
Illustratively, the training module 400 includes:
the first processing unit is used for adjusting hue and saturation of the expanded muck car picture data set to obtain a first expanded muck car picture data set;
the second processing unit is used for adding one or more operations of random zooming, cutting, translation, cutting and rotation to the first expansion muck car picture data set to obtain a second expansion muck car picture data set;
and the training unit is used for carrying out iterative training on the preset detection model through the second expansion slag car picture data set, and obtaining the slag car roof fall detection model when the epoch of the preset detection model reaches a preset value.
Illustratively, the training module 400 further comprises:
and the third processing unit is used for randomly selecting two samples from the second expansion slag car image data set for random weighted summation, and labels of the samples correspond to the weighted summation.
Illustratively, the generation system of the muck vehicle roof fall detection model further comprises a prediction and inference module, wherein the prediction and inference module is used for: extracting frame rate information of a video to be detected and carrying out data preprocessing to obtain a data set of a picture to be detected; and detecting the data set of the picture to be detected according to the slag car roof fall detection model to obtain a prediction reasoning result.
It should be noted that the generation system of the muck vehicle roof fall detection model provided in the embodiment of the present application corresponds to the method embodiment shown in fig. 1 to 2, and is not described herein again to avoid repetition.
Fig. 4 shows a block diagram of an electronic device according to an embodiment of the present disclosure, where fig. 4 is a block diagram of the electronic device. The electronic device may include a processor 510, a communication interface 520, a memory 530, and at least one communication bus 540. Wherein the communication bus 540 is used for realizing direct connection communication of these components. In this embodiment, the communication interface 520 of the electronic device is used for performing signaling or data communication with other node devices. Processor 510 may be an integrated circuit chip having signal processing capabilities.
The processor 510 may be a general-purpose processor including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor 510 may be any conventional processor or the like.
The Memory 530 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Read Only Memory (EPROM), an electrically Erasable Read Only Memory (EEPROM), and the like. The memory 530 stores computer readable instructions, which when executed by the processor 510, enable the electronic device to perform the steps involved in the method embodiments of fig. 1-2 described above.
Optionally, the electronic device may further include a memory controller, an input output unit.
The memory 530, the memory controller, the processor 510, the peripheral interface, and the input/output unit are electrically connected to each other directly or indirectly, so as to implement data transmission or interaction. For example, these elements may be electrically coupled to each other via one or more communication buses 540. The processor 510 is used to execute executable modules stored in the memory 530, such as software functional modules or computer programs included in the electronic device.
The input and output unit is used for providing a task for a user to create and start an optional time period or preset execution time for the task creation so as to realize the interaction between the user and the server. The input/output unit may be, but is not limited to, a mouse, a keyboard, and the like.
It will be appreciated that the configuration shown in fig. 4 is merely illustrative and that the electronic device may include more or fewer components than shown in fig. 4 or may have a different configuration than shown in fig. 4. The components shown in fig. 4 may be implemented in hardware, software, or a combination thereof.
The embodiment of the present application further provides a storage medium, where the storage medium stores instructions, and when the instructions are run on a computer, when the computer program is executed by a processor, the method in the method embodiment is implemented, and in order to avoid repetition, details are not repeated here.
The present application also provides a computer program product which, when run on a computer, causes the computer to perform the method of the method embodiments.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative, and for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising one of 8230; \8230;" 8230; "does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. A method for generating a slag car roof fall detection model is characterized by comprising the following steps:
acquiring training video data;
generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate;
constructing a preset detection model, wherein the preset detection model is a two-stage detection model, rough target detection is carried out through a preset yolov5 target detection algorithm in one stage, and fine classification is carried out by inputting a detection target into a resnet18 in the second stage;
and performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model.
2. The method for generating a muck truck roof fall detection model according to claim 1, wherein before the step of constructing a preset detection model, the method further comprises:
and performing data enhancement on the image data set of the slag car through a preset generated countermeasure network GAN to obtain an expanded image data set of the slag car.
3. The method for generating the slag car roof fall detection model according to claim 2, wherein the step of iteratively training the preset detection model through the slag car picture data set to obtain the slag car roof fall detection model comprises:
adjusting hue and saturation of the expanded muck car picture data set to obtain a first expanded muck car picture data set;
adding one or more operations of random zooming, cutting, translation, cutting and rotation to the first expansion muck car picture data set to obtain a second expansion muck car picture data set;
and performing iterative training on the preset detection model through the second expansion slag car picture data set, wherein the epoch of the preset detection model reaches a preset value, and obtaining a slag car roof fall detection model.
4. The method for generating the muck car roof fall detection model according to claim 3, wherein before the step of obtaining the muck car roof fall detection model by performing iterative training on the preset detection model through the second expanded muck car picture data set and an epoch of the preset detection model reaching a preset value, the method further comprises:
and randomly selecting two samples from the second expansion slag car image data set for random weighted summation, wherein the labels of the samples correspond to the weighted summation.
5. The method for generating a muck car roof fall detection model according to claim 1, wherein after the step of iteratively training the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model, the method further comprises:
extracting frame rate information of a video to be detected and carrying out data preprocessing to obtain a data set of a picture to be detected;
and detecting the data set of the picture to be detected according to the slag car roof fall detection model to obtain a prediction reasoning result.
6. A system for generating a slag car roof fall detection model is characterized by comprising:
the training video acquisition module is used for acquiring training video data;
the training picture generation module is used for generating a muck car picture data set according to the training video data, wherein the muck car picture data set is divided into four categories of full wrapping, half wrapping, no wrapping and empty car based on the top coverage rate;
the model construction module is used for constructing a preset detection model, the preset detection model is a two-stage detection model, rough target detection is carried out through a preset yolov5 target detection algorithm in one stage, and fine classification is carried out by inputting detection targets into a resnet18 in the second stage;
and the model training module is used for performing iterative training on the preset detection model through the muck car picture data set to obtain a muck car roof fall detection model.
7. The system for generating a slag car roof fall detection model according to claim 6, further comprising:
and the data enhancement module is used for performing data enhancement on the image data set of the muck car through a preset generated confrontation network GAN to obtain an expanded image data set of the muck car.
8. The system of claim 7, wherein the training module comprises:
the first processing unit is used for adjusting the hue and the saturation of the expanded muck car picture data set to obtain a first expanded muck car picture data set;
the second processing unit is used for adding one or more operations of random zooming, cutting, translation, cutting and rotation to the first expansion muck car picture data set to obtain a second expansion muck car picture data set;
and the training unit is used for performing iterative training on the preset detection model through the second expansion slag car picture data set, and acquiring a slag car roof fall detection model when the epoch of the preset detection model reaches a preset value.
9. An electronic device, comprising: a memory, a processor and a computer program stored in the memory and executable on the processor, the processor when executing the computer program implementing the steps of the method of generating a muck truck roof fall detection model according to any one of claims 1 to 5.
10. A computer-readable storage medium having stored thereon instructions which, when executed on a computer, cause the computer to perform a method of generating a muck truck roof fall detection model according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310154326.9A CN115861904A (en) | 2023-02-23 | 2023-02-23 | Method and system for generating slag car roof fall detection model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310154326.9A CN115861904A (en) | 2023-02-23 | 2023-02-23 | Method and system for generating slag car roof fall detection model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115861904A true CN115861904A (en) | 2023-03-28 |
Family
ID=85658740
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310154326.9A Pending CN115861904A (en) | 2023-02-23 | 2023-02-23 | Method and system for generating slag car roof fall detection model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115861904A (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110766039A (en) * | 2019-09-02 | 2020-02-07 | 厦门卫星定位应用股份有限公司 | Muck truck transportation state identification method, medium, equipment and muck truck |
| CN111209923A (en) * | 2020-04-23 | 2020-05-29 | 北京慧智数据科技有限公司 | Deep learning technology-based muck truck cover or uncover identification method |
| CN111401162A (en) * | 2020-03-05 | 2020-07-10 | 上海眼控科技股份有限公司 | Illegal auditing method for muck vehicle, electronic device, computer equipment and storage medium |
| CN111950368A (en) * | 2020-07-09 | 2020-11-17 | 深圳神目信息技术有限公司 | Freight vehicle monitoring method, device, electronic equipment and medium |
| CN113723833A (en) * | 2021-09-02 | 2021-11-30 | 湖南三乔林业有限公司 | Method and system for evaluating afforestation actual performance quality, terminal equipment and storage medium |
| CN113947108A (en) * | 2021-10-15 | 2022-01-18 | 福州大学 | Player tracking detection method based on YOLO V5 |
| CN114445780A (en) * | 2022-02-10 | 2022-05-06 | 青岛熙正数字科技有限公司 | Detection method and device for bare soil covering, and training method and device for recognition model |
| CN114677596A (en) * | 2022-05-26 | 2022-06-28 | 之江实验室 | Remote sensing image ship detection method and device based on attention model |
| CN114972789A (en) * | 2022-05-27 | 2022-08-30 | 无锡沃格自动化科技股份有限公司 | Large-size light guide plate defect visual detection method and system and storage medium |
| CN115311688A (en) * | 2022-08-11 | 2022-11-08 | 睿云奇智(青岛)科技有限公司 | Pedestrian detection method and device, electronic equipment and storage medium |
| CN115511879A (en) * | 2022-11-07 | 2022-12-23 | 山东大学 | Detection system and detection method for detecting whether muck truck is covered with no cover based on computer vision |
| CN115631433A (en) * | 2022-10-17 | 2023-01-20 | 昭通亮风台信息科技有限公司 | Target detection method, device, equipment and medium |
| CN115690627A (en) * | 2022-11-03 | 2023-02-03 | 安徽大学 | Method and system for detecting aerial image rotating target |
-
2023
- 2023-02-23 CN CN202310154326.9A patent/CN115861904A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110766039A (en) * | 2019-09-02 | 2020-02-07 | 厦门卫星定位应用股份有限公司 | Muck truck transportation state identification method, medium, equipment and muck truck |
| CN111401162A (en) * | 2020-03-05 | 2020-07-10 | 上海眼控科技股份有限公司 | Illegal auditing method for muck vehicle, electronic device, computer equipment and storage medium |
| CN111209923A (en) * | 2020-04-23 | 2020-05-29 | 北京慧智数据科技有限公司 | Deep learning technology-based muck truck cover or uncover identification method |
| CN111950368A (en) * | 2020-07-09 | 2020-11-17 | 深圳神目信息技术有限公司 | Freight vehicle monitoring method, device, electronic equipment and medium |
| CN113723833A (en) * | 2021-09-02 | 2021-11-30 | 湖南三乔林业有限公司 | Method and system for evaluating afforestation actual performance quality, terminal equipment and storage medium |
| CN113947108A (en) * | 2021-10-15 | 2022-01-18 | 福州大学 | Player tracking detection method based on YOLO V5 |
| CN114445780A (en) * | 2022-02-10 | 2022-05-06 | 青岛熙正数字科技有限公司 | Detection method and device for bare soil covering, and training method and device for recognition model |
| CN114677596A (en) * | 2022-05-26 | 2022-06-28 | 之江实验室 | Remote sensing image ship detection method and device based on attention model |
| CN114972789A (en) * | 2022-05-27 | 2022-08-30 | 无锡沃格自动化科技股份有限公司 | Large-size light guide plate defect visual detection method and system and storage medium |
| CN115311688A (en) * | 2022-08-11 | 2022-11-08 | 睿云奇智(青岛)科技有限公司 | Pedestrian detection method and device, electronic equipment and storage medium |
| CN115631433A (en) * | 2022-10-17 | 2023-01-20 | 昭通亮风台信息科技有限公司 | Target detection method, device, equipment and medium |
| CN115690627A (en) * | 2022-11-03 | 2023-02-03 | 安徽大学 | Method and system for detecting aerial image rotating target |
| CN115511879A (en) * | 2022-11-07 | 2022-12-23 | 山东大学 | Detection system and detection method for detecting whether muck truck is covered with no cover based on computer vision |
Non-Patent Citations (2)
| Title |
|---|
| XINGKUI ZHU 等: "TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarios", 《2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW)》 * |
| 涂超 等: "基于多源信息融合的交通要素识别方法研究", 《农业装备与车辆工程》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xiao et al. | Development of an image data set of construction machines for deep learning object detection | |
| CN108319909B (en) | Driving behavior analysis method and system | |
| CN114049356B (en) | Method, device and system for detecting structure apparent crack | |
| Wu et al. | Applying deep convolutional neural network with 3D reality mesh model for water tank crack detection and evaluation | |
| CN115019314A (en) | Commodity price identification method, device, equipment and storage medium | |
| CN112613434A (en) | Road target detection method, device and storage medium | |
| CN114120284A (en) | Deep learning highway lane structuring method, storage medium and device | |
| EP3767590A1 (en) | Device and method for training a generative model | |
| Lee et al. | Development of a car-free street mapping model using an integrated system with unmanned aerial vehicles, aerial mapping cameras, and a deep learning algorithm | |
| CN115861904A (en) | Method and system for generating slag car roof fall detection model | |
| CN112712098B (en) | Image data processing method and device | |
| CN112597996A (en) | Task-driven natural scene-based traffic sign significance detection method | |
| US20230281999A1 (en) | Infrastructure analysis using panoptic segmentation | |
| CN112837326B (en) | Method, device and equipment for detecting carryover | |
| Fei et al. | Adapting public annotated data sets and low-quality dash cameras for spatiotemporal estimation of traffic-related air pollution: A transfer-learning approach | |
| CN116229443A (en) | Vehicle license plate detection method, device, computer equipment and storage medium | |
| CN114648718B (en) | Road occupation detection method, system, electronic equipment and readable storage medium | |
| CN115272853A (en) | Industrial wasteland identification method and product based on artificial intelligence technology and big data | |
| CN114627400A (en) | Lane congestion detection method and device, electronic equipment and storage medium | |
| Yang et al. | Sewer pipe defects diagnosis assessment using multivariate analysis on CCTV video imagery | |
| CN116469013B (en) | Road ponding prediction method, device, computer equipment and storage medium | |
| JP2003030649A (en) | Image processing method and image processor | |
| Muhammad | Detection parking Spaces by using the ResNet50 Algorithm | |
| EP4090016A2 (en) | Apparatus, method and program | |
| WO2023152882A1 (en) | Model training device, model training method, and non-transitory computer-readable medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20230328 |