CN112800912A - Dynamic feature based label-based migration feature neural network training method - Google Patents
Dynamic feature based label-based migration feature neural network training method Download PDFInfo
- Publication number
- CN112800912A CN112800912A CN202110074093.2A CN202110074093A CN112800912A CN 112800912 A CN112800912 A CN 112800912A CN 202110074093 A CN202110074093 A CN 202110074093A CN 112800912 A CN112800912 A CN 112800912A
- Authority
- CN
- China
- Prior art keywords
- feature map
- classification
- neural network
- feature
- function value
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 109
- 238000012549 training Methods 0.000 title claims abstract description 86
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000013508 migration Methods 0.000 title claims abstract description 61
- 230000005012 migration Effects 0.000 title claims abstract description 61
- 238000012544 monitoring process Methods 0.000 claims abstract description 18
- 230000006870 function Effects 0.000 claims description 144
- 238000001514 detection method Methods 0.000 claims description 42
- 238000013527 convolutional neural network Methods 0.000 claims description 28
- 230000004927 fusion Effects 0.000 claims description 19
- 230000003068 static effect Effects 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000010276 construction Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 23
- 238000003860 storage Methods 0.000 description 11
- 241000276425 Xiphophorus maculatus Species 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000009435 building construction Methods 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000003252 repetitive effect Effects 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
Abstract
The application relates to intelligent construction safety monitoring in the field of intelligent buildings, and particularly discloses a training method of a neural network based on migration features of dynamic features based on labels.
Description
Technical Field
The present invention relates to intelligent operation state control in the field of intelligent manufacturing, and more particularly, to a method for training a neural network based on a dynamic feature tag-based migration feature, a method for detecting stability based on a deep neural network, a system for training a neural network based on a dynamic feature tag-based migration feature, a system for detecting stability based on a deep neural network, and an electronic device.
Background
The fabricated building is a building which is formed by transferring a large amount of field operation work in the traditional construction mode to a factory, processing and manufacturing building components and accessories (such as floor slabs, wall slabs, stairs, balconies and the like) in the factory, transporting the components and accessories to a building construction site, and assembling and installing the components and the accessories on the site in a reliable connection mode. Most of the components of the prefabricated building are plate-shaped (wall panels and the like) components, and are required to be transported to an installation position by lifting equipment during factory or field construction.
Because the platy member rocks easily when handling, at present, in order to evaluate the rocking of the platy member when handling, some handling equipment are additionally provided with a torsion sensor for sensing the torsion value of a cable for handling the platy member, thereby determining whether the rocking of the platy member exceeds a necessary amplitude to improve the safety when in transportation. However, the detection of the wobble by the torsion value alone still has a problem of insufficient accuracy.
Therefore, an optimized technical scheme for detecting the stability of the lifted plate-shaped member is expected.
At present, deep learning and neural networks have been widely applied in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
The deep learning and the development of the neural network provide a new solution for the stability detection of the lifted plate-shaped member.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a dynamic feature label-based migration feature neural network training method, a deep neural network stability detection method, a dynamic feature label-based migration feature neural network training system, a deep neural network stability detection system and electronic equipment.
According to one aspect of the application, a training method of a neural network based on dynamic feature label-based migration features is provided, and comprises the following steps:
step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video;
step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value;
and step 3: respectively passing the current frame and the previous frame through a depth convolution neural network to obtain a first feature map and a second feature map;
and 4, step 4: matrix multiplying the second feature map and the first feature map to obtain a third feature map;
and 5: calculating a first cross entropy function value of the label value and the first feature map and a second cross entropy function value of the label value and the third feature map respectively and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fusion cross entropy loss function value;
step 6: fusing the first feature map and the second feature map to obtain a classification feature map;
and 7: passing the classification feature map through a classifier to obtain a classification loss function value;
and 8: updating parameters of the deep convolutional neural network based on a weighted sum of the fused cross-entropy loss function values and the classification loss function values and by backpropagation of gradient descent.
In the above training method for a neural network based on a dynamic feature and a label-based migration feature, step 6: fusing the first feature map and the second feature map to obtain a classification feature map, comprising: calculating a weighted sum of the first feature map and the second feature map by pixel position to obtain the classification feature map.
In the above training method for a neural network based on a dynamic feature and a label-based migration feature, step 7: passing the classification feature map through a classifier to obtain a classification loss function value, comprising: passing the classification feature map through one or more fully connected layers to obtain a classification feature vector; inputting the classification feature vector into a Softmax classification function to obtain a classification result; and inputting the classification result and the real value into a loss function to obtain the classification loss function value.
In the above method for training a neural network based on a dynamic feature tag-based migration feature, the current frame and the previous frame have a preset time interval therebetween.
In the above training method for a neural network based on a dynamic feature and a label-based migration feature, the deep convolutional neural network is a deep residual error network.
According to another aspect of the present application, there is provided a stability detection method based on a deep neural network, including:
acquiring an image to be detected, wherein the image to be detected is an image of a plate-shaped member lifted by a lifting device; and
and inputting the image to be detected into the deep convolution neural network and the classifier trained according to the training method of the neural network based on the dynamic feature and the migration feature of the label to obtain a classification result, wherein the classification result is used for indicating whether the stability of the lifted plate-shaped member meets the preset requirement or not.
According to another aspect of the present application, there is provided a training system for a neural network based on dynamic feature tag-based migration features, comprising:
a surveillance video acquisition unit for performing step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video;
a tag value acquisition unit configured to perform step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value;
a static feature map generation unit, configured to execute step 3: respectively enabling the current frame and the previous frame obtained by the surveillance video obtaining unit to pass through a depth convolution neural network to obtain a first feature map and a second feature map;
a dynamic characteristic map generating unit, configured to execute step 4: matrix-multiplying the second feature map obtained by the static feature map generation unit with the first feature map to obtain a third feature map;
a fusion cross entropy loss function value calculation unit for performing step 5: calculating a first cross entropy function value of the label value obtained by the label value obtaining unit and the first feature map obtained by the static feature map generating unit and a second cross entropy function value of the label value obtained by the label value obtaining unit and the third feature map obtained by the dynamic feature map generating unit respectively, and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fused cross entropy loss function value;
a classification feature map generation unit, configured to execute step 6: fusing the first feature map and the second feature map obtained by the static feature map generation unit to obtain a classification feature map;
a classification loss function value calculation unit for executing step 7: the classification feature map obtained by the classification feature map generating unit passes through a classifier to obtain a classification loss function value;
a parameter updating unit, configured to perform step 8: updating parameters of the deep convolutional neural network by backpropagation of gradient descent based on a weighted sum of the fusion cross-entropy loss function value obtained by the fusion cross-entropy loss function value calculation unit and the classification loss function value obtained by the classification loss function value calculation unit.
In the above training system for a neural network based on dynamic feature label-based migration features, the classification feature map generation unit is further configured to calculate a weighted sum of the first feature map and the second feature map by pixel position to obtain the classification feature map.
In the above training system for a neural network based on a dynamic feature label-based migration feature, the classification loss function value calculating unit includes: the coding subunit is used for enabling the classification feature map to pass through one or more full connection layers so as to obtain a classification feature vector; a classification result obtaining subunit, configured to input the classification feature vector into a Softmax classification function to obtain a classification result; and the loss function value calculating operator unit is used for inputting the classification result and the real value into a loss function so as to obtain the classification loss function value.
In the above training system of a neural network based on dynamic feature tag-based migration features, the current frame and the previous frame have a preset time interval therebetween.
In the training system of the neural network based on the dynamic feature and the label-based migration feature, the deep convolutional neural network is a deep residual error network.
According to still another aspect of the present application, there is provided a deep neural network-based stability detection system, including:
the device comprises an image acquisition unit to be detected, a lifting unit and a control unit, wherein the image acquisition unit to be detected is used for acquiring an image to be detected, and the image to be detected is an image of a plate-shaped member lifted by lifting equipment; and
and the classification result generating unit is used for inputting the image to be detected obtained by the image to be detected obtaining unit into the deep convolution neural network and the classifier trained according to the training method of the neural network based on the dynamic feature and the label-based migration feature so as to obtain a classification result, wherein the classification result is used for indicating whether the stability of the lifted plate-shaped member meets the preset requirement or not.
According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform a method of training a neural network based on dynamic feature tag-based migration features as described above, or a method of stability detection based on a deep neural network.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to execute the method for training a neural network based on dynamic feature tag-based migration features or the method for detecting stability based on a deep neural network as described above.
Compared with the prior art, the dynamic feature label-based migration feature-based neural network training method, the deep neural network-based stability detection method, the dynamic feature label-based migration feature-based neural network training system, the deep neural network-based stability detection system and the electronic device are provided by the application.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 illustrates an application scenario diagram of a dynamic feature label-based migration feature-based neural network training method and a deep neural network-based stability detection method according to an embodiment of the present application;
FIG. 2 illustrates a flow diagram of a method of training a neural network based on dynamic feature tag-based migration features according to an embodiment of the present application;
FIG. 3 illustrates a system architecture diagram of a method for training a neural network based on dynamic feature tag-based migration features according to an embodiment of the present application;
fig. 4 illustrates a training method of a neural network based on dynamic feature label-based migration features according to an embodiment of the present application, step 7: a flow chart for passing the classification feature map through a classifier to obtain a classification loss function value;
FIG. 5 illustrates a flow chart of a method of deep neural network based stability detection in accordance with an embodiment of the present application;
FIG. 6 illustrates a block diagram of a training system for a neural network based on dynamic feature tag-based migration features in accordance with an embodiment of the present application.
Fig. 7 illustrates a block diagram of a classification loss function value calculation unit in a training system of a dynamic feature label-based migration feature based neural network according to an embodiment of the present application.
FIG. 8 illustrates a block diagram of a deep neural network-based stability detection system in accordance with an embodiment of the present application.
FIG. 9 illustrates a block diagram of an electronic device in accordance with an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of a scene
As described above, the prefabricated building refers to a building which is fabricated by transferring a large amount of field work in a conventional construction method to a factory, processing and manufacturing building components and accessories (such as floor slabs, wall panels, stairs, balconies, etc.) in the factory, transporting the components and accessories to a building construction site, and assembling and installing the components and accessories on the site through a reliable connection method. Most of the components of the prefabricated building are plate-shaped (wall panels and the like) components, and are required to be transported to an installation position by lifting equipment during factory or field construction.
Because the platy member rocks easily when handling, at present, in order to evaluate the rocking of the platy member when handling, some handling equipment are additionally provided with a torsion sensor for sensing the torsion value of a cable for handling the platy member, thereby determining whether the rocking of the platy member exceeds a necessary amplitude to improve the safety when in transportation. However, the detection of the wobble by the torsion value alone still has a problem of insufficient accuracy.
Therefore, an optimized technical scheme for detecting the stability of the lifted plate-shaped member is expected.
Since the shake of the plate-shaped member directly appears as a visual feature, the applicant of the present application expects to perform shake detection by extracting and classifying image features from a shake image of the plate-shaped member by computer vision technology based on deep learning.
In the practical process, the applicant of the present application finds that, on one hand, image association between consecutive frames (not necessarily adjacent frames, but several frames apart) in a video representing shaking includes relatively rich information for determining shaking, and on the other hand, a torsion value of an existing torsion sensor also includes relatively rich information for determining shaking, and if information of various aspects can be effectively fused, on the one hand, performance of training, including training speed and robustness of model training, and on the other hand, accuracy of a trained model can also be improved.
Therefore, in the technical solution of the present application, the applicant of the present application calculates cross entropy function values between the tag values and the static features and the dynamic features, respectively, on the basis of the cross entropy loss function, and calculates a fusion cross entropy loss function value for characterizing the label-based migration of the dynamic features by a weighted sum of the cross entropy function values of the two, thereby effectively fusing the above information at the time of model training.
Specifically, a video of the lifting equipment when the plate-shaped member is lifted is obtained, a current frame and a previous frame are extracted, and a torsion value of a cable for lifting the plate-shaped member is sensed by a torsion sensor corresponding to the current frame and is used as a tag value. Then, a first feature map and a second feature map are obtained by respectively passing the current frame and the previous frame through a convolutional neural network, and the first feature map is multiplied by the second feature map to obtain a third feature map. Then, a first cross entropy function value of the label value and the first feature map and a second cross entropy function value of the label value and the third feature map are respectively calculated, and a weighted sum of the first cross entropy function value and the second cross entropy function value is calculated to obtain a fused cross entropy loss function value.
And, obtaining a classification feature map by weight fusion of the first feature map and the second feature map, and obtaining a classification loss function value by a classifier, thereby training the convolutional neural network based on a weighted sum of the fusion cross-entropy loss function value and the classification loss function value.
Based on this, the present application proposes a training method for a neural network based on a dynamic feature label-based migration feature, which includes: step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video; step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value; and step 3: respectively passing the current frame and the previous frame through a depth convolution neural network to obtain a first feature map and a second feature map; and 4, step 4: matrix multiplying the second feature map and the first feature map to obtain a third feature map; and 5: calculating a first cross entropy function value of the label value and the first feature map and a second cross entropy function value of the label value and the third feature map respectively and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fusion cross entropy loss function value; step 6: fusing the first feature map and the second feature map to obtain a classification feature map; and 7: passing the classification feature map through a classifier to obtain a classification loss function value; and, step 8: updating parameters of the deep convolutional neural network based on a weighted sum of the fused cross-entropy loss function values and the classification loss function values and by backpropagation of gradient descent.
Based on this, the present application also provides a stability detection method based on a deep neural network, which includes: acquiring an image to be detected, wherein the image to be detected is an image of a plate-shaped member lifted by a lifting device; and inputting the image to be detected into the deep convolution neural network and the classifier trained according to the training method of the neural network based on the dynamic feature and the migration feature of the label to obtain a classification result, wherein the classification result is used for indicating whether the stability of the lifted plate-shaped member meets the preset requirement or not.
Fig. 1 illustrates an application scenario diagram of a dynamic feature label-based migration feature-based neural network training method and a deep neural network-based stability detection method according to an embodiment of the present application.
As shown in fig. 1, in the training phase of the application scenario, a monitoring video when the plate-shaped member is lifted by a lifting device through a camera (e.g., C as illustrated in fig. 1) and a torsion value of a cable lifting the plate-shaped member is sensed as a tag value through a torsion sensor (e.g., D as illustrated in fig. 1); then, the surveillance video is input into a server (e.g., S as illustrated in fig. 1) deployed with a training algorithm of a neural network based on the dynamic feature tag-based migration feature, wherein the server is capable of training the neural network for stability detection of the hoisted plate-shaped member with image frames in the surveillance video based on the training algorithm of the neural network based on the dynamic feature tag-based migration feature.
After the neural network is trained through the training algorithm of the neural network based on the dynamic feature and the label-based migration feature, the stability of the lifted plate-shaped member can be detected based on the deep neural network.
Further, as shown in fig. 1, in the application stage of the application scenario, an image to be detected is obtained through a camera (for example, as indicated by C in fig. 1), where the image to be detected is an image of a plate-shaped member lifted by a lifting device; then, the image to be detected is input into a server (for example, S as illustrated in fig. 1) deployed with a stability detection algorithm based on a deep neural network, wherein the server can process the image to be detected based on the stability detection algorithm of the deep neural network to generate a detection result for indicating whether the stability of the hoisted plate-shaped member meets a preset requirement.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
FIG. 2 illustrates a flow chart of a method of training a neural network based on dynamic feature tag-based migration features. As shown in fig. 2, a method for training a neural network based on dynamic feature label-based migration features according to an embodiment of the present application includes: step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video; step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value; and step 3: respectively passing the current frame and the previous frame through a depth convolution neural network to obtain a first feature map and a second feature map; and 4, step 4: matrix multiplying the second feature map and the first feature map to obtain a third feature map; and 5: calculating a first cross entropy function value of the label value and the first feature map and a second cross entropy function value of the label value and the third feature map respectively and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fusion cross entropy loss function value; step 6: fusing the first feature map and the second feature map to obtain a classification feature map; and 7: passing the classification feature map through a classifier to obtain a classification loss function value; and, step 8: updating parameters of the deep convolutional neural network based on a weighted sum of the fused cross-entropy loss function values and the classification loss function values and by backpropagation of gradient descent.
Fig. 3 illustrates an architecture diagram of a training method of a neural network based on dynamic feature label-based migration features according to an embodiment of the present application. As shown IN fig. 3, IN the network architecture of the training method of the neural network based on the dynamic feature tag-based migration feature, first, a current frame (e.g., IN0 as illustrated IN fig. 3) and a previous frame (e.g., IN1 as illustrated IN fig. 3) extracted from a surveillance video when a lifting apparatus lifts a plate-shaped member are passed through a deep convolutional neural network (e.g., CNN as illustrated IN fig. 3) to obtain a first feature map (e.g., F1 as illustrated IN fig. 3) corresponding to the current frame and a second feature map (e.g., F2 as illustrated IN fig. 3) corresponding to the previous frame; then, matrix-multiplying the second feature map with the first feature map to obtain a third feature map (e.g., F3 as illustrated in fig. 3); then, respectively calculating a first cross entropy function value of a tag value (for example, as illustrated as P in fig. 3) and the first feature map and a second cross entropy function value of the tag value and the third feature map, and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fused cross entropy loss function value, wherein the tag value is a torsion value of a cable for which a torsion sensor corresponding to the current frame induces the plate-shaped member to be lifted; further, fusing the first feature map and the second feature map to obtain a classification feature map (e.g., Fc as illustrated in fig. 3); then, the classification characteristic graph passes through a classifier to obtain a classification loss function value; finally, parameters of the deep convolutional neural network are updated based on a weighted sum of the fused cross-entropy loss function values and the classification loss function values and by back propagation of gradient descent.
In step 1, a monitoring video of a lifting device when lifting a plate-shaped member is obtained, and a current frame and a previous frame are extracted from the monitoring video. As described above, in the technical solution of the present application, since the shake of the plate-shaped member is directly expressed as a visual feature, the applicant of the present application desires to perform shake detection by extracting and classifying image features from a shake image of the plate-shaped member by a computer vision technique based on deep learning.
In the practical process, the applicant of the present application finds that, on one hand, image association between consecutive frames (not necessarily adjacent frames, but several frames apart) in a video representing shaking includes relatively rich information for determining shaking, and on the other hand, a torsion value of an existing torsion sensor also includes relatively rich information for determining shaking, and if information of various aspects can be effectively fused, on the one hand, performance of training, including training speed and robustness of model training, and on the other hand, accuracy of a trained model can also be improved.
Therefore, in the technical scheme of the application, a surveillance video when the lifting equipment lifts the plate-shaped member is obtained as training data, and a current frame and a previous frame are extracted from the surveillance video, wherein the image association between the current frame and the previous frame contains richer information for judging shaking.
It should be noted that, in the embodiment of the present application, there is a preset time interval between the current frame and the previous frame. For example, in a specific example of the present application, the current frame and the previous frame are adjacent frames, or a preset number of image frames are spaced between the current frame and the previous frame.
In step 2, a torsion value of the cable for sensing the hoisting of the plate-shaped member by the torsion sensor corresponding to the current frame is obtained as a tag value. It should be appreciated that the torsion value contains a relatively large amount of information for determining sloshing.
In step 3, the current frame and the previous frame are respectively passed through a deep convolutional neural network to obtain a first feature map and a second feature map. That is, the current frame and the previous frame are processed with a deep convolutional neural network to extract high-dimensional implicit features in the current frame and the previous frame.
In embodiments of the present application, the deep convolutional neural network may employ a deep residual neural network, e.g., ResNet 50. It should be known to those skilled in the art that, compared to the conventional convolutional neural network, the deep residual network is an optimized network structure proposed on the basis of the conventional convolutional neural network, which mainly solves the problem of gradient disappearance during the training process. The depth residual error network introduces a residual error network structure, the network layer can be made deeper through the residual error network structure, and the problem of gradient disappearance can not occur. The residual error network uses the cross-layer link thought of a high-speed network for reference, breaks through the convention that the traditional neural network only can provide N layers as input from the input layer of the N-1 layer, enables the output of a certain layer to directly cross several layers as the input of the later layer, and has the significance of providing a new direction for the difficult problem that the error rate of the whole learning model is not reduced and inversely increased by superposing multiple layers of networks.
In step 4, the second characteristic diagram is subjected to matrix multiplication with the first characteristic diagram to obtain a third characteristic diagram. More specifically, the first feature map and the second feature map are matrix-multiplied to apply the information in the second feature map to the first feature map to obtain the third feature map. It should be understood that the second feature map corresponds to the previous frame, and thus the first feature map is matrix multiplied with the second feature map, meaning that information in the previous frame is acted on in the current frame, so that the third feature map obtained includes the image correlation between the current frame and the previous frame.
In step 5, a first cross entropy function value of the label value and the first feature map and a second cross entropy function value of the label value and the third feature map are calculated respectively, and a weighted sum of the first cross entropy function value and the second cross entropy function value is calculated to obtain a fused cross entropy loss function value. Here, the first cross entropy loss function value represents a probability that the feature distribution in the first feature map meets the label value, and the second cross entropy loss function value represents a probability that the feature distribution in the third feature map meets the label value.
It should be understood that the first characteristic diagram corresponds to the current frame, and the third characteristic diagram is obtained by multiplying the second characteristic diagram by the matrix of the first characteristic diagram, so that the first characteristic diagram represents the static characteristics of the lifted plate-shaped member, and the third characteristic diagram represents the dynamic characteristics of the lifted plate-shaped member. That is, in the technical solution of the present application, on the basis of the cross entropy loss function, the training method calculates cross entropy function values between the tag values and the static feature and the dynamic feature, respectively, and calculates a fusion cross entropy loss function value for characterizing the tag-based migration of the dynamic feature by a weighted sum of the cross entropy function values based on the two, thereby effectively fusing the above information at the time of model training.
In step 6, the first feature map and the second feature map are fused to obtain a classification feature map. That is, the first feature map and the second feature map are fused by weighting to obtain a classification feature map. Here, the first characteristic map and the second characteristic map represent static characteristics of the plate-like member to be lifted at different times, and therefore, the classification characteristic map also represents static characteristics.
In a specific example of the present application, a process of fusing the first feature map and the second feature map to obtain a classification feature map includes: calculating a weighted sum of the first feature map and the second feature map by pixel position to obtain the classification feature map. I.e. a weighted sum between the first feature map and the second feature map at the pixel level to obtain the classification feature map.
In step 7, the classification feature map is passed through a classifier to obtain a classification loss function value. More specifically, in the embodiment of the present application, a process of passing the classification feature map through a classifier to obtain a classification loss function value includes: first, the classification feature map is passed through one or more fully connected layers to obtain a classification feature vector. That is, the classification feature map is encoded using one or more fully-connected layers as an encoder to fully utilize information at various locations in the classification feature map. Then, the classification feature vector is input into a Softmax classification function to obtain a classification result. Then, the classification result and the real value are input into a loss function to obtain the classification loss function value. Here, the classification loss function value represents a probability that the classification result conforms to the true value.
Fig. 4 illustrates a training method of a neural network based on dynamic feature label-based migration features according to an embodiment of the present application, step 7: and passing the classification feature map through a classifier to obtain a classification loss function value. As shown in fig. 4, in the embodiment of the present application, step 7: passing the classification feature map through a classifier to obtain a classification loss function value, comprising: step 71: passing the classification feature map through one or more fully connected layers to obtain a classification feature vector; step 72: inputting the classification feature vector into a Softmax classification function to obtain a classification result; and, step 73: and inputting the classification result and the real value into a loss function to obtain the classification loss function value.
In step 8, parameters of the deep convolutional neural network are updated based on a weighted sum of the fused cross-entropy loss function values and the classification loss function values and by back propagation of gradient descent. This is the conventional procedure of deep neural network update, i.e. the parameters of the deep convolutional neural network are updated by the BP algorithm.
It is worth mentioning that in the embodiment of the present application, while updating the parameters of the deep convolutional neural network based on the weighted sum of the fusion cross entropy loss function value and the classification loss function value, the parameters of the classifier may also be updated synchronously. That is, the deep convolutional neural network and the classifier may be trained jointly.
In summary, the training method of the neural network based on the migration feature of the dynamic feature based on the label according to the embodiment of the present application is illustrated, wherein in the training process, on the basis of training the neural network with the classification loss function values, cross entropy function values between the label values and the static feature and the dynamic feature used for representing the object to be detected are respectively calculated, and a fusion cross entropy loss function value used for representing the migration of the dynamic feature based on the label is calculated by a weighted sum of the cross entropy function values based on the two, so as to effectively fuse the dynamic feature information and the static feature information of the object to be detected during model training.
According to another aspect of the application, a stability detection method based on a deep neural network is also provided.
Fig. 5 illustrates a flowchart of a method for deep neural network-based stability detection according to an embodiment of the present application. As shown in fig. 5, a method for detecting stability based on a deep neural network according to an embodiment of the present application includes: s110, acquiring an image to be detected, wherein the image to be detected is an image of a plate-shaped member lifted by lifting equipment; and S120, inputting the image to be detected into the deep convolution neural network and the classifier trained according to the training method of the neural network based on the dynamic feature and the label migration feature so as to obtain a classification result, wherein the classification result is used for indicating whether the stability of the lifted plate-shaped member meets the preset requirement or not.
Exemplary System
FIG. 6 illustrates a block diagram of a training system for a neural network based on dynamic feature tag-based migration features in accordance with an embodiment of the present application.
As shown in fig. 6, a training system 600 for a neural network based on dynamic feature label-based migration features according to an embodiment of the present application includes: a surveillance video obtaining unit 610, configured to perform step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video; a tag value obtaining unit 620, configured to perform step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value; a static feature map generating unit 630, configured to perform step 3: respectively passing the current frame and the previous frame obtained by the surveillance video obtaining unit 610 through a depth convolutional neural network to obtain a first feature map and a second feature map; a dynamic characteristic map generating unit 640, configured to perform step 4: matrix-multiplying the second feature map obtained by the static feature map generating unit 630 with the first feature map to obtain a third feature map; a fusion cross entropy loss function value calculating unit 650, configured to perform step 5: calculating a first cross entropy function value of the label value obtained by the label value obtaining unit 620 and the first feature map obtained by the static feature map generating unit 630 and a second cross entropy function value of the label value obtained by the label value obtaining unit 620 and the third feature map obtained by the dynamic feature map generating unit 640, respectively, and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fused cross entropy loss function value; a classification feature map generating unit 660, configured to perform step 6: fusing the first feature map and the second feature map obtained by the static feature map generation unit 630 to obtain a classification feature map; a classification loss function value calculation unit 670, configured to perform step 7: passing the classification feature map obtained by the classification feature map generation unit 660 through a classifier to obtain a classification loss function value; and a parameter updating unit 680, configured to perform step 8: updating parameters of the deep convolutional neural network by backpropagation of gradient descent based on a weighted sum of the fusion cross-entropy loss function value obtained by the fusion cross-entropy loss function value calculation unit 650 and the classification loss function value obtained by the classification loss function value calculation unit 670.
In an example, in the training system 600, the classification feature map generating unit 660 is further configured to calculate a weighted sum of the first feature map and the second feature map according to pixel positions to obtain the classification feature map.
In one example, in the training system 600, as shown in fig. 7, the classification loss function value calculation unit 670 includes: an encoding subunit 671, configured to pass the classification feature map through one or more fully-connected layers to obtain a classification feature vector; a classification result obtaining subunit 672, configured to input the classification feature vector into a Softmax classification function to obtain a classification result; and a loss function value calculation operator unit 673, configured to input the classification result and the true value into a loss function to obtain the classification loss function value.
In one example, in the training system 600, the current frame and the previous frame have a preset time interval therebetween.
In one example, in the training system 600 described above, the deep convolutional neural network is a deep residual network.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the training system 600 have been described in detail in the above description of the training method of the neural network based on the dynamic feature tag-based migration feature with reference to fig. 1 to 4, and thus, a repetitive description thereof will be omitted.
As described above, the training system 600 according to the embodiment of the present application may be implemented in various terminal devices, such as a server for stability detection. In one example, the training system 600 according to embodiments of the present application may be integrated into a terminal device as a software module and/or a hardware module. For example, the training system 600 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the training system 600 may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the training system 600 and the terminal device may be separate devices, and the training system 600 may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information according to an agreed data format.
According to another aspect of the application, a stability detection system based on a deep neural network is also provided.
FIG. 8 illustrates a block diagram of a deep neural network-based stability detection system in accordance with an embodiment of the present application. As shown in fig. 8, a stability detection system 800 based on a deep neural network according to an embodiment of the present application includes: the image acquiring unit 810 to be detected is used for acquiring an image to be detected, wherein the image to be detected is an image of a plate-shaped member lifted by lifting equipment; and a classification result generating unit 820, configured to input the image to be detected obtained by the image to be detected obtaining unit 810 into the deep convolutional neural network and the classifier trained according to the above-described training method for the neural network based on the migration features of the dynamic features and the labels, so as to obtain a classification result, where the classification result is used to indicate whether the stability of the lifted plate-shaped member meets a preset requirement.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described detection system 800 have been described in detail in the above description of the deep neural network-based stability detection method with reference to fig. 5, and thus, a repetitive description thereof will be omitted.
As described above, the detection system 800 according to the embodiment of the present application may be implemented in various terminal devices, such as a server for stability detection and the like. In one example, the detection system 800 according to embodiments of the application may be integrated into the terminal device as one software module and/or hardware module. For example, the detection system 800 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the detection system 800 may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the detection system 800 and the terminal device may be separate devices, and the detection system 800 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information according to an agreed data format.
Exemplary electronic device
Next, an electronic apparatus according to an embodiment of the present application is described with reference to fig. 9.
FIG. 9 illustrates a block diagram of an electronic device in accordance with an embodiment of the present application.
As shown in fig. 9, the electronic device 10 includes one or more processors 11 and a memory 12.
The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
In one example, the electronic device 10 may further include: an input system 13 and an output system 14, which are interconnected by a bus system and/or other form of connection mechanism (not shown).
The input system 13 may comprise, for example, a keyboard, a mouse, etc.
The output system 14 can output various information including the detection result to the outside. The output system 14 may include, for example, a display, speakers, a printer, and a communication network and its connected remote output devices, among others.
Of course, for simplicity, only some of the components of the electronic device 10 relevant to the present application are shown in fig. 9, and components such as buses, input/output interfaces, and the like are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.
Exemplary computer program product and computer-readable storage Medium
In addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the method for training a dynamic feature tag-based migration feature-based neural network, or steps in functions in a deep neural network-based stability detection method, according to various embodiments of the present application, as described in the "exemplary methods" section above in this specification.
The computer program product may be written with program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform the method for training a neural network based on dynamic feature tag-based migration features described in the "exemplary methods" section above in this specification, or the steps in the method for detecting stability based on deep neural networks.
The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. A training method of a neural network based on dynamic feature label-based migration features is characterized by comprising the following steps:
step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video;
step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value;
and step 3: respectively passing the current frame and the previous frame through a depth convolution neural network to obtain a first feature map and a second feature map;
and 4, step 4: matrix multiplying the second feature map and the first feature map to obtain a third feature map;
and 5: calculating a first cross entropy function value of the label value and the first feature map and a second cross entropy function value of the label value and the third feature map respectively and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fusion cross entropy loss function value;
step 6: fusing the first feature map and the second feature map to obtain a classification feature map;
and 7: passing the classification feature map through a classifier to obtain a classification loss function value;
and 8: updating parameters of the deep convolutional neural network based on a weighted sum of the fused cross-entropy loss function values and the classification loss function values and by backpropagation of gradient descent.
2. The method for training a neural network based on dynamic feature label-based migration features of claim 1, wherein the step 6: fusing the first feature map and the second feature map to obtain a classification feature map, comprising:
calculating a weighted sum of the first feature map and the second feature map by pixel position to obtain the classification feature map.
3. The method for training a neural network based on dynamic feature label-based migration features of claim 2, wherein the step 7: passing the classification feature map through a classifier to obtain a classification loss function value, comprising:
passing the classification feature map through one or more fully connected layers to obtain a classification feature vector;
inputting the classification feature vector into a Softmax classification function to obtain a classification result; and
and inputting the classification result and the real value into a loss function to obtain the classification loss function value.
4. The method for training a neural network based on dynamic feature tag-based migration features according to claim 1, wherein the current frame and the previous frame have a preset time interval therebetween.
5. The method of training a dynamic feature label-based migration feature based neural network of claim 1, wherein the deep convolutional neural network is a deep residual network.
6. A stability detection method based on a deep neural network is characterized by comprising the following steps:
acquiring an image to be detected, wherein the image to be detected is an image of a plate-shaped member lifted by a lifting device; and
inputting the image to be detected into the deep convolutional neural network and the classifier trained by the training method of the neural network based on the migration characteristics of the dynamic characteristics and the labels according to any one of claims 1 to 6 to obtain a classification result, wherein the classification result is used for indicating whether the stability of the lifted plate-shaped member meets the preset requirement or not.
7. A system for training a neural network based on dynamic feature tag-based migration features, comprising:
a surveillance video acquisition unit for performing step 1: acquiring a monitoring video when a lifting device lifts a plate-shaped member and extracting a current frame and a previous frame from the monitoring video;
a tag value acquisition unit configured to perform step 2: acquiring a torsion value of a cable for sensing the plate-shaped member to be lifted by the torsion sensor corresponding to the current frame as a tag value;
a static feature map generation unit, configured to execute step 3: respectively enabling the current frame and the previous frame obtained by the surveillance video obtaining unit to pass through a depth convolution neural network to obtain a first feature map and a second feature map;
a dynamic characteristic map generating unit, configured to execute step 4: matrix-multiplying the second feature map obtained by the static feature map generation unit with the first feature map to obtain a third feature map;
a fusion cross entropy loss function value calculation unit for performing step 5: calculating a first cross entropy function value of the label value obtained by the label value obtaining unit and the first feature map obtained by the static feature map generating unit and a second cross entropy function value of the label value obtained by the label value obtaining unit and the third feature map obtained by the dynamic feature map generating unit respectively, and calculating a weighted sum of the first cross entropy function value and the second cross entropy function value to obtain a fused cross entropy loss function value;
a classification feature map generation unit, configured to execute step 6: fusing the first feature map and the second feature map obtained by the static feature map generation unit to obtain a classification feature map;
a classification loss function value calculation unit for executing step 7: the classification feature map obtained by the classification feature map generating unit passes through a classifier to obtain a classification loss function value;
a parameter updating unit, configured to perform step 8: updating parameters of the deep convolutional neural network by backpropagation of gradient descent based on a weighted sum of the fusion cross-entropy loss function value obtained by the fusion cross-entropy loss function value calculation unit and the classification loss function value obtained by the classification loss function value calculation unit.
8. The system according to claim 7, wherein the classification feature map generation unit is further configured to calculate a weighted sum of the first feature map and the second feature map by pixel position to obtain the classification feature map.
9. A stability detection system based on a deep neural network is characterized by comprising:
the device comprises an image acquisition unit to be detected, a lifting unit and a control unit, wherein the image acquisition unit to be detected is used for acquiring an image to be detected, and the image to be detected is an image of a plate-shaped member lifted by lifting equipment; and
a classification result generating unit, configured to input the image to be detected obtained by the image to be detected obtaining unit into the deep convolution neural network and the classifier trained according to the training method for the neural network based on the migration features of the dynamic features and the labels as set forth in any one of claims 1 to 6, so as to obtain a classification result, where the classification result is used to indicate whether the stability of the lifted plate-shaped member meets a preset requirement.
10. An electronic device, comprising:
a processor; and
a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform the method of training a dynamic feature tag-based migration feature-based neural network according to any one of claims 1-6 or the method of detecting stability based on a deep neural network according to claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110074093.2A CN112800912A (en) | 2021-01-20 | 2021-01-20 | Dynamic feature based label-based migration feature neural network training method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110074093.2A CN112800912A (en) | 2021-01-20 | 2021-01-20 | Dynamic feature based label-based migration feature neural network training method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112800912A true CN112800912A (en) | 2021-05-14 |
Family
ID=75810650
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110074093.2A Pending CN112800912A (en) | 2021-01-20 | 2021-01-20 | Dynamic feature based label-based migration feature neural network training method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112800912A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115520786A (en) * | 2022-08-25 | 2022-12-27 | 江苏广坤铝业有限公司 | Aluminum material discharging auxiliary transfer equipment and transfer control method |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014229971A (en) * | 2013-05-20 | 2014-12-08 | 株式会社朋栄 | Rolling shutter distortion correction and video image stabilization processing method |
| JP2017200852A (en) * | 2016-04-28 | 2017-11-09 | 日立造船株式会社 | Calculation device, method for controlling calculation device, control program and recording medium |
| CN109409256A (en) * | 2018-10-10 | 2019-03-01 | 东南大学 | A kind of forest rocket detection method based on 3D convolutional neural networks |
| WO2019072845A1 (en) * | 2017-10-10 | 2019-04-18 | Dsm Ip Assets B.V. | Smart hoisting rope |
| KR20190096309A (en) * | 2019-05-29 | 2019-08-19 | 엘지전자 주식회사 | Vehicle control apparatus |
| KR102141163B1 (en) * | 2020-05-06 | 2020-08-04 | 국방과학연구소 | Neural network learning method and apparatus for generating synthetic aperture radar image |
| CN111723617A (en) * | 2019-03-20 | 2020-09-29 | 顺丰科技有限公司 | Method, device and equipment for recognizing actions and storage medium |
| CN112489087A (en) * | 2020-12-13 | 2021-03-12 | 深圳市进致网络科技有限公司 | Method for detecting shaking of suspension type operation platform for high-rise building construction |
| CN112750096A (en) * | 2021-01-08 | 2021-05-04 | 南京炜俊网络科技有限公司 | Neural network training method for monitoring bagged cement capacity |
-
2021
- 2021-01-20 CN CN202110074093.2A patent/CN112800912A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014229971A (en) * | 2013-05-20 | 2014-12-08 | 株式会社朋栄 | Rolling shutter distortion correction and video image stabilization processing method |
| JP2017200852A (en) * | 2016-04-28 | 2017-11-09 | 日立造船株式会社 | Calculation device, method for controlling calculation device, control program and recording medium |
| WO2019072845A1 (en) * | 2017-10-10 | 2019-04-18 | Dsm Ip Assets B.V. | Smart hoisting rope |
| CN109409256A (en) * | 2018-10-10 | 2019-03-01 | 东南大学 | A kind of forest rocket detection method based on 3D convolutional neural networks |
| CN111723617A (en) * | 2019-03-20 | 2020-09-29 | 顺丰科技有限公司 | Method, device and equipment for recognizing actions and storage medium |
| KR20190096309A (en) * | 2019-05-29 | 2019-08-19 | 엘지전자 주식회사 | Vehicle control apparatus |
| KR102141163B1 (en) * | 2020-05-06 | 2020-08-04 | 국방과학연구소 | Neural network learning method and apparatus for generating synthetic aperture radar image |
| CN112489087A (en) * | 2020-12-13 | 2021-03-12 | 深圳市进致网络科技有限公司 | Method for detecting shaking of suspension type operation platform for high-rise building construction |
| CN112750096A (en) * | 2021-01-08 | 2021-05-04 | 南京炜俊网络科技有限公司 | Neural network training method for monitoring bagged cement capacity |
Non-Patent Citations (1)
| Title |
|---|
| CHEN ZHI-MEI ET AL.: "Intelligent anti-swing control for bridge crane", 《J. CENT. SOUTH UNIV.》, pages 2774 - 2781 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115520786A (en) * | 2022-08-25 | 2022-12-27 | 江苏广坤铝业有限公司 | Aluminum material discharging auxiliary transfer equipment and transfer control method |
| CN115520786B (en) * | 2022-08-25 | 2024-02-27 | 江苏广坤铝业有限公司 | Aluminum product discharging auxiliary transfer equipment and transfer control method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10402690B2 (en) | System and method for learning random-walk label propagation for weakly-supervised semantic segmentation | |
| KR102252298B1 (en) | Method and apparatus for recognizing facial expression | |
| JP2023541532A (en) | Text detection model training method and apparatus, text detection method and apparatus, electronic equipment, storage medium, and computer program | |
| CN116415654A (en) | Data processing method and related equipment | |
| JP2018041319A (en) | Movement state learning device, movement state recognition unit, method, and program | |
| CN112861012B (en) | Recommendation method and device based on context and user long-term and short-term preference adaptive learning | |
| CN112183166A (en) | Method and device for determining training sample and electronic equipment | |
| CN112508041A (en) | Training method of neural network for spray control based on classification result label | |
| CN111931862B (en) | Method and system for detecting illegal posted advertisements and electronic equipment | |
| CN112446352A (en) | Behavior recognition method, behavior recognition device, behavior recognition medium, and electronic device | |
| JP6248593B2 (en) | Posture detection apparatus, posture detection method, and posture detection program | |
| CN112800912A (en) | Dynamic feature based label-based migration feature neural network training method | |
| CN112489087A (en) | Method for detecting shaking of suspension type operation platform for high-rise building construction | |
| CN111523351A (en) | Neural network training method and device and electronic equipment | |
| CN116247824B (en) | Control method and system for power equipment | |
| CN112465805A (en) | Neural network training method for quality detection of steel bar stamping and bending | |
| CN112016677A (en) | Deep neural network training method and device and electronic equipment | |
| CN115690438A (en) | Determination method of aerial view characteristics, target detection method and device and electronic equipment | |
| CN111898544A (en) | Character and image matching method, device and equipment and computer storage medium | |
| CN114330634A (en) | Neural network processing method and related equipment | |
| JP7176626B2 (en) | Movement situation learning device, movement situation recognition device, model learning method, movement situation recognition method, and program | |
| CN110516717B (en) | Method and apparatus for generating image recognition model | |
| CN113483813A (en) | Scaffold safety detection method and system | |
| CN112686882A (en) | Method for detecting stability of pipeline erected on pipeline support frame | |
| CN112836752A (en) | Intelligent sampling parameter control method based on feature map fusion of depth values |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20231226 Address after: Room 408-8, 4th Floor, Building 2, Haichuang Technology Center, Cangqian Street, Yuhang District, Hangzhou City, Zhejiang Province, 313000 Applicant after: HANGZHOU ZHUILIE TECHNOLOGY Co.,Ltd. Address before: 210043 room 101-6, building 17, phase I, human resources service Industrial Park, Jiangbei new district, Nanjing City, Jiangsu Province Applicant before: Jiangsu Tianmu UAV Technology Co.,Ltd. |
|
| TA01 | Transfer of patent application right |