CN113344877A - Reinforcing steel bar model training method and device based on convolutional neural network - Google Patents
Reinforcing steel bar model training method and device based on convolutional neural network Download PDFInfo
- Publication number
- CN113344877A CN113344877A CN202110637095.8A CN202110637095A CN113344877A CN 113344877 A CN113344877 A CN 113344877A CN 202110637095 A CN202110637095 A CN 202110637095A CN 113344877 A CN113344877 A CN 113344877A
- Authority
- CN
- China
- Prior art keywords
- neural network
- detection channel
- convolution
- convolutional neural
- output
- 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
- 238000012549 training Methods 0.000 title claims abstract description 91
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 36
- 229910001294 Reinforcing steel Inorganic materials 0.000 title claims abstract description 22
- 238000001514 detection method Methods 0.000 claims abstract description 117
- 238000005070 sampling Methods 0.000 claims abstract description 51
- 229910000831 Steel Inorganic materials 0.000 claims abstract description 40
- 239000010959 steel Substances 0.000 claims abstract description 40
- 230000004927 fusion Effects 0.000 claims abstract description 11
- 238000003062 neural network model Methods 0.000 claims description 44
- 238000012545 processing Methods 0.000 claims description 21
- 238000004590 computer program Methods 0.000 claims description 8
- 230000002787 reinforcement Effects 0.000 claims description 3
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 239000000463 material Substances 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30136—Metal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the invention provides a reinforcing steel bar model training method and a reinforcing steel bar model training device based on a convolutional neural network, wherein the method comprises the following steps: sampling an original image of the end face of the steel bar to obtain a training sample, and training the training sample; downsampling a convolution layer in the convolutional neural network model to obtain a first detection channel for predicting a large-size target; performing up-sampling on output in the first detection channel, and performing image fusion on an up-sampling result and a feature image in the convolutional layer to obtain a second detection channel for predicting a medium-size target and a third detection channel for predicting a small-size target; deleting the first detection channel of the convolutional layer to obtain an updated convolutional neural network model, and detecting the training sample through the updated convolutional neural network model to obtain the quantity information of the reinforcing steel bars. By adopting the method, the training parameters can be reduced when the convolutional neural network model is used for training, namely, the time and the memory required by the training model are reduced.
Description
Technical Field
The invention relates to the technical field of visual recognition, in particular to a reinforcing steel bar model training method and device based on a convolutional neural network.
Background
In various constructions, such as urban infrastructures, roads, bridges, etc., or civil buildings, etc., steel bars are required to be used as foundation materials. Data show that the annual output of the reinforcing steel bars in China currently exceeds 2 hundred million tons, and the reinforcing steel bars are indispensable materials in the building industry.
Generally, when the reinforcing steel bars are used as base materials, human resources are needed to count the number of the reinforcing steel bars entering a construction site, but the time is consumed during the manual counting, and the accuracy cannot be guaranteed, so that some reinforcing steel bar counting methods through a convolutional neural network exist at present.
However, the conventional method for counting the reinforcing steel bars through the convolutional neural network only saves manpower resources, and the time consumption for acquiring the sample and obtaining the counting result of the reinforcing steel bars through model training is long, and a large amount of memory is required for corresponding data processing, so that the progress of other steps of the engineering can be influenced.
Disclosure of Invention
Aiming at the problems in the prior art, the embodiment of the invention provides a reinforcing steel bar model training method and device based on a convolutional neural network.
The embodiment of the invention provides a reinforcing steel bar model training method based on a convolutional neural network, which comprises the following steps:
sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training;
down-sampling a convolutional layer in the convolutional neural network model, and extracting the original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target;
performing up-sampling on the output in the first detection channel, and performing image fusion on the up-sampling result and the feature image in the convolutional layer to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target;
deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the reinforcing steel bars.
In one embodiment, the method further comprises:
and acquiring a preset full convolution network structure, carrying out N times of downsampling on a convolution layer in the convolution neural network model through the full convolution network structure, and carrying out convolution processing on a result of the Nth downsampling to obtain the output of the first detection channel.
In one embodiment, the method further comprises:
carrying out N-1 times of downsampling on the convolutional layer in the convolutional neural network model, and carrying out convolution processing on the result of the N-1 times of downsampling to obtain the output of the second detection channel;
and performing 2 times of upsampling on the output of the second detection channel, splicing the result of the 2 times of upsampling with the result of the N-2 times of downsampling, and performing convolution processing on the spliced result to obtain the output of the third detection channel.
In one embodiment, the method further comprises:
and modifying the convolution number of the detection channels in the updated convolution neural network model.
In one embodiment, the convolutional neural network model includes:
YOLO v3 convolutional neural network model.
The embodiment of the invention provides a reinforcing steel bar model training device based on a convolutional neural network, which comprises:
the sampling module is used for sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training;
the down-sampling module is used for down-sampling the convolutional layer in the convolutional neural network model, extracting the original image and obtaining the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target;
the up-sampling module is used for up-sampling the output in the first detection channel and carrying out image fusion on the up-sampling result and the characteristic image in the convolutional layer to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target;
and the deleting module is used for deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars.
In one embodiment, the apparatus further comprises:
and the acquisition module is used for acquiring a preset full convolution network structure, carrying out N times of downsampling on a convolution layer in the convolution neural network model through the full convolution network structure, and carrying out convolution processing on the result of the Nth downsampling to obtain the output of the first detection channel.
In one embodiment, the apparatus further comprises:
the convolution processing module is used for carrying out N-1 times of downsampling on a convolution layer in the convolution neural network model and carrying out convolution processing on the result of the N-1 times of downsampling to obtain the output of the second detection channel;
and the splicing module is used for performing 2-time upsampling on the output of the second detection channel, splicing the result of the 2-time upsampling with the result of the N-2-time downsampling, and performing convolution processing on the splicing result to obtain the output of the third detection channel.
The embodiment of the invention provides electronic equipment which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor executes the program to realize the steps of the reinforcement model training method based on the convolutional neural network.
Embodiments of the present invention provide a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the above-mentioned convolutional neural network-based steel bar model training method.
According to the reinforcing steel bar model training method and device based on the convolutional neural network, provided by the embodiment of the invention, an original image of the end face of a reinforcing steel bar is sampled to obtain a training sample, and the training sample is input into the convolutional neural network model for training; performing downsampling on a convolutional layer in the convolutional neural network model, and extracting an original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target; the output in the first detection channel is subjected to up-sampling, and the up-sampling result and the characteristic image in the convolutional layer are subjected to image fusion to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target; deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars. This allows for a reduction of training parameters when training through the convolutional neural network model, i.e. a reduction of the time and memory required for training the model.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flowchart of a method for training a steel bar model based on a convolutional neural network according to an embodiment of the present invention;
FIG. 2 is a structural diagram of a steel bar model training device based on a convolutional neural network in the embodiment of the present invention;
fig. 3 is a schematic structural diagram of an electronic device in an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a schematic flow diagram of a method for training a steel bar model based on a convolutional neural network according to an embodiment of the present invention, and as shown in fig. 1, an embodiment of the present invention provides a method for training a steel bar model based on a convolutional neural network, including:
and S101, sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training.
Specifically, an original image obtained by shooting the end face of the steel bar is sampled, that is, the format, size, pixels and the like of the sample are processed to obtain a training sample of the model, and the training sample is input into a convolutional neural network model for training, wherein the convolutional neural network model may be a YOLO v3 convolutional neural network model.
And S102, performing downsampling on the convolutional layer in the convolutional neural network model, and extracting the original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target.
Specifically, downsampling is performed on a convolutional layer in a convolutional neural network model, an original image is extracted, output of a first detection channel is obtained, a preset full convolutional network structure Darknet-53 can be obtained, then downsampling is directly performed on the convolutional layer through step length 2 through the Darknet-53, downsampling is performed for N times, after sampling is performed for 5 times, namely 32 times, the original image is extracted into 13 × 13 feature maps and output, and a first prediction channel 13 × 255 for predicting a large-size target is obtained.
Step S103, performing up-sampling on the output in the first detection channel, and performing image fusion on the up-sampling result and the feature image in the convolutional layer to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting the medium-size target, and the third detection channel is used for predicting the small-size target.
Specifically, the output of the first detection channel is up-sampled, and the up-sampled result is image-fused with the feature image in the convolution layer, so as to obtain the output of the second detection channel and the output of the third detection channel, and specifically, the 2 nd prediction channel 26 × 255 for predicting medium and small sizes and the 3 rd prediction channel 52 × 255 for predicting small-size targets can be obtained through the up-sampling and fusion method of FPN (feature pyramid).
It is also possible to directly perform subsequent convolution processing on 26 × 512 after the 4 th downsampling in the YOLO v3 model to obtain a channel for detecting the medium-sized object, and since the counting of the rebars is only concerned with the rebars in the image, the output of the channel is 26 × 18, 18=3 (4+1+ 1).
The method comprises the steps of performing convolution calculation on 26 x 256 characteristic graphs obtained by detecting medium-sized channels to obtain 26 x 128 characteristic graphs, performing 2 times of upsampling on the characteristic graphs to obtain 52 x 128 characteristic graphs, splicing the characteristic graphs with 52 x 256 characteristic graphs obtained by the third downsampling to obtain 52 x 384 characteristic graphs, performing convolution calculation on the characteristic graphs, modifying the number of convolutions of the detection channels in an updated convolution neural network model, and modifying the number of 1 x 1 convolutions of the detection channels of the YOLOv3 model from 255 to 18 to obtain channels for detecting small-sized targets, namely 52 x 18.
And step S104, deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars.
Specifically, a first detection channel (13 × 13 channel) for predicting a large-size target is deleted, a new deep neural network model is built, the training samples are detected through the updated convolutional neural network model, and the number information of the steel bars is obtained, and compared with a YOLO v3 model, the new convolutional neural network model only reserves 26 × 26 channels for predicting a medium size and 52 × 52 channels with a small size, so that the number of convolution layers of the model is reduced by about 30% compared with the number of layers of YOLO v3, the time average of the training model is reduced by about 30% under different training times, and meanwhile, the memory occupation amount is also reduced.
The embodiment of the invention provides a reinforcing steel bar model training method based on a convolutional neural network, which comprises the steps of sampling an original image of an end face of a reinforcing steel bar to obtain a training sample, and inputting the training sample into the convolutional neural network model for training; performing downsampling on a convolutional layer in the convolutional neural network model, and extracting an original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target; the output in the first detection channel is subjected to up-sampling, and the up-sampling result and the characteristic image in the convolutional layer are subjected to image fusion to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target; deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars. This allows for a reduction of training parameters when training through the convolutional neural network model, i.e. a reduction of the time and memory required for training the model.
In another embodiment, the method for training the steel bar model based on the convolutional neural network specifically comprises the following steps:
and processing the steel bar picture needing target detection, and entering a Darknet-53 deep neural network without a full connection layer for feature extraction according to the size of 416 × 416. In order to avoid the negative effect of the gradient caused by the pooling layer, Darknet-53 directly uses step 2 to perform down-sampling on the convolution layer, and after 5 times of sampling, namely 32 times of sampling, the original image is extracted into a feature map of 13 × 13 and output, and a 1 st prediction path 13 × 255 for predicting a large-size target is obtained. In order to enhance the accuracy of the algorithm for detecting the small targets, a similar FPN (feature pyramid) upsampling and merging method is adopted in YOLOv3, so as to obtain a 2 nd predicted path 26 × 255 for predicting medium-sized and small-sized targets and a 3 rd predicted path 52 × 255 for predicting small-sized targets, and then 13 × 13 channels for predicting large-sized targets are removed, and a new deep neural network model is built.
Fig. 2 is a steel bar model training device based on a convolutional neural network according to an embodiment of the present invention, which includes: a sampling module S201, a down-sampling module S202, an up-sampling module S203, and a deleting module S204, wherein:
and the sampling module S201 is used for sampling the original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into the convolutional neural network model for training.
A down-sampling module S202, configured to perform down-sampling on the convolutional layer in the convolutional neural network model, extract the original image, and obtain an output of a first detection channel, where the first detection channel is used to predict a large-size target.
And the upsampling module S203 is configured to upsample the output in the first detection channel, and perform image fusion on the upsampling result and the feature image in the convolutional layer to obtain an output of a second detection channel and an output of a third detection channel, where the second detection channel is used to predict a medium-size target, and the third detection channel is used to predict a small-size target.
And the deleting module S204 is used for deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars.
In one embodiment, the apparatus may further comprise:
and the acquisition module is used for acquiring a preset full convolution network structure, carrying out N times of downsampling on a convolution layer in the convolution neural network model through the full convolution network structure, and carrying out convolution processing on the result of the Nth downsampling to obtain the output of the first detection channel.
In one embodiment, the apparatus may further comprise:
and the convolution processing module is used for carrying out N-1 times of downsampling on the convolution layer in the convolution neural network model, carrying out convolution processing on the result of the N-1 times of downsampling and obtaining the output of the second detection channel.
And the splicing module is used for performing 2-time upsampling on the output of the second detection channel, splicing the result of the 2-time upsampling with the result of the N-2-time downsampling, and performing convolution processing on the splicing result to obtain the output of the third detection channel.
In one embodiment, the apparatus may further comprise:
and the modifying module is used for modifying the convolution number of the detection channels in the updated convolution neural network model.
For specific limitations of the convolutional neural network-based steel bar model training device, reference may be made to the above limitations of the convolutional neural network-based steel bar model training method, and details are not repeated here. The modules in the convolutional neural network-based steel bar model training device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
Fig. 3 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 3: a processor (processor)301, a memory (memory)302, a communication Interface (Communications Interface)303 and a communication bus 304, wherein the processor 301, the memory 302 and the communication Interface 303 complete communication with each other through the communication bus 304. The processor 301 may call logic instructions in the memory 302 to perform the following method: sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training; performing downsampling on a convolutional layer in the convolutional neural network model, and extracting an original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target; the output in the first detection channel is subjected to up-sampling, and the up-sampling result and the characteristic image in the convolutional layer are subjected to image fusion to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target; deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars.
Furthermore, the logic instructions in the memory 302 may be implemented in software functional units and stored in a computer readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes 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 invention. 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.
In another aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented to perform the transmission method provided in the foregoing embodiments when executed by a processor, and for example, the method includes: sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training; performing downsampling on a convolutional layer in the convolutional neural network model, and extracting an original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target; the output in the first detection channel is subjected to up-sampling, and the up-sampling result and the characteristic image in the convolutional layer are subjected to image fusion to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target; deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A steel bar model training method based on a convolutional neural network is characterized by comprising the following steps:
sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training;
down-sampling a convolutional layer in the convolutional neural network model, and extracting the original image to obtain the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target;
performing up-sampling on the output in the first detection channel, and performing image fusion on the up-sampling result and the feature image in the convolutional layer to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target;
deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the reinforcing steel bars.
2. The convolutional neural network-based steel bar model training method as claimed in claim 1, wherein the downsampling convolutional layers in the convolutional neural network model, extracting the original image, and obtaining the output of the first detection channel comprises:
and acquiring a preset full convolution network structure, carrying out N times of downsampling on a convolution layer in the convolution neural network model through the full convolution network structure, and carrying out convolution processing on a result of the Nth downsampling to obtain the output of the first detection channel.
3. The convolutional neural network-based rebar model training method of claim 2, further comprising:
carrying out N-1 times of downsampling on the convolutional layer in the convolutional neural network model, and carrying out convolution processing on the result of the N-1 times of downsampling to obtain the output of the second detection channel;
and performing 2 times of upsampling on the output of the second detection channel, splicing the result of the 2 times of upsampling with the result of the N-2 times of downsampling, and performing convolution processing on the spliced result to obtain the output of the third detection channel.
4. The convolutional neural network-based steel bar model training method as claimed in claim 1, wherein after obtaining the updated convolutional neural network model, the method comprises:
and modifying the convolution number of the detection channels in the updated convolution neural network model.
5. The convolutional neural network based steel reinforcement model training method of claim 1, wherein the convolutional neural network model comprises:
YOLO v3 convolutional neural network model.
6. A steel bar model training device based on a convolutional neural network is characterized by comprising:
the sampling module is used for sampling an original image of the end face of the steel bar to obtain a training sample, and inputting the training sample into a convolutional neural network model for training;
the down-sampling module is used for down-sampling the convolutional layer in the convolutional neural network model, extracting the original image and obtaining the output of a first detection channel, wherein the first detection channel is used for predicting a large-size target;
the up-sampling module is used for up-sampling the output in the first detection channel and carrying out image fusion on the up-sampling result and the characteristic image in the convolutional layer to obtain the output of a second detection channel and the output of a third detection channel, wherein the second detection channel is used for predicting a medium-size target, and the third detection channel is used for predicting a small-size target;
and the deleting module is used for deleting the first detection channel of the convolution layer in the convolution neural network model to obtain an updated convolution neural network model, and detecting the training sample through the updated convolution neural network model to obtain the quantity information of the steel bars.
7. The convolutional neural network-based steel bar model training device as claimed in claim 6, wherein the device further comprises:
and the acquisition module is used for acquiring a preset full convolution network structure, carrying out N times of downsampling on a convolution layer in the convolution neural network model through the full convolution network structure, and carrying out convolution processing on the result of the Nth downsampling to obtain the output of the first detection channel.
8. The convolutional neural network-based steel bar model training device as claimed in claim 6, wherein the device further comprises:
the convolution processing module is used for carrying out N-1 times of downsampling on a convolution layer in the convolution neural network model and carrying out convolution processing on the result of the N-1 times of downsampling to obtain the output of the second detection channel;
and the splicing module is used for performing 2-time upsampling on the output of the second detection channel, splicing the result of the 2-time upsampling with the result of the N-2-time downsampling, and performing convolution processing on the splicing result to obtain the output of the third detection channel.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the convolutional neural network-based steel model training method as claimed in any one of claims 1 to 5.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the convolutional neural network-based steel reinforcement model training method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110637095.8A CN113344877A (en) | 2021-06-08 | 2021-06-08 | Reinforcing steel bar model training method and device based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110637095.8A CN113344877A (en) | 2021-06-08 | 2021-06-08 | Reinforcing steel bar model training method and device based on convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113344877A true CN113344877A (en) | 2021-09-03 |
Family
ID=77475227
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110637095.8A Pending CN113344877A (en) | 2021-06-08 | 2021-06-08 | Reinforcing steel bar model training method and device based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113344877A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115496000A (en) * | 2022-11-14 | 2022-12-20 | 深圳市地铁集团有限公司 | Method, system and equipment for calculating reinforcement engineering quantity of intercity railway |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109902631A (en) * | 2019-03-01 | 2019-06-18 | 北京视甄智能科技有限公司 | A kind of fast face detecting method based on image pyramid |
| CN111639740A (en) * | 2020-05-09 | 2020-09-08 | 武汉工程大学 | Steel bar counting method based on multi-scale convolution neural network |
| CN111967305A (en) * | 2020-07-01 | 2020-11-20 | 华南理工大学 | Real-time multi-scale target detection method based on lightweight convolutional neural network |
| CN112101322A (en) * | 2020-11-17 | 2020-12-18 | 深圳市优必选科技股份有限公司 | Target object detection model, training method, object detection method and related equipment |
| CN112329658A (en) * | 2020-11-10 | 2021-02-05 | 江苏科技大学 | Method for improving detection algorithm of YOLOV3 network |
| CN112446350A (en) * | 2020-12-09 | 2021-03-05 | 武汉工程大学 | Improved method for detecting cotton in YOLOv3 complex cotton field background |
| CN112464954A (en) * | 2020-12-06 | 2021-03-09 | 上海悠络客电子科技股份有限公司 | Lightweight target detection network applied to embedded equipment and training method |
-
2021
- 2021-06-08 CN CN202110637095.8A patent/CN113344877A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109902631A (en) * | 2019-03-01 | 2019-06-18 | 北京视甄智能科技有限公司 | A kind of fast face detecting method based on image pyramid |
| CN111639740A (en) * | 2020-05-09 | 2020-09-08 | 武汉工程大学 | Steel bar counting method based on multi-scale convolution neural network |
| CN111967305A (en) * | 2020-07-01 | 2020-11-20 | 华南理工大学 | Real-time multi-scale target detection method based on lightweight convolutional neural network |
| CN112329658A (en) * | 2020-11-10 | 2021-02-05 | 江苏科技大学 | Method for improving detection algorithm of YOLOV3 network |
| CN112101322A (en) * | 2020-11-17 | 2020-12-18 | 深圳市优必选科技股份有限公司 | Target object detection model, training method, object detection method and related equipment |
| CN112464954A (en) * | 2020-12-06 | 2021-03-09 | 上海悠络客电子科技股份有限公司 | Lightweight target detection network applied to embedded equipment and training method |
| CN112446350A (en) * | 2020-12-09 | 2021-03-05 | 武汉工程大学 | Improved method for detecting cotton in YOLOv3 complex cotton field background |
Non-Patent Citations (1)
| Title |
|---|
| 刘威: "基于图像处理的钢管自动识别系统", 《中国优秀硕士学位论文全文数据库 工程科技I辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115496000A (en) * | 2022-11-14 | 2022-12-20 | 深圳市地铁集团有限公司 | Method, system and equipment for calculating reinforcement engineering quantity of intercity railway |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108664981B (en) | Salient image extraction method and device | |
| CN109685124A (en) | Road disease recognition methods neural network based and device | |
| CN112800915B (en) | Building change detection method, device, electronic equipment and storage medium | |
| CN112233240B (en) | Three-dimensional vector data slicing method and device of three-dimensional vector map and electronic equipment | |
| CN113361578B (en) | Training method and device for image processing model, electronic equipment and storage medium | |
| CN114821605B (en) | Text processing method, device, equipment and medium | |
| CN111582175A (en) | High-resolution remote sensing image semantic segmentation method sharing multi-scale countermeasure characteristics | |
| CN111597932A (en) | Road crack image identification method, device and system based on convolutional neural network | |
| CN113344877A (en) | Reinforcing steel bar model training method and device based on convolutional neural network | |
| CN114187459A (en) | Training method and device of target detection model, electronic equipment and storage medium | |
| CN114715145B (en) | Trajectory prediction method, device and equipment and automatic driving vehicle | |
| CN112184687A (en) | Road crack detection method based on capsule characteristic pyramid and storage medium | |
| WO2017056804A1 (en) | Inspection results search device and method | |
| CN116823800A (en) | Bridge concrete crack detection method based on deep learning under complex background | |
| CN114897272A (en) | Liquid level prediction method and device for sewage inspection well, electronic equipment and storage medium | |
| CN113158856A (en) | Processing method and device for extracting target area in remote sensing image | |
| CN117094660A (en) | Construction monitoring method and system based on digital twin technology | |
| CN116556718A (en) | Repairing method, repairing device, repairing equipment and repairing storage medium for building structural component | |
| CN115019321A (en) | Text recognition method, text model training method, text recognition device, text model training equipment and storage medium | |
| CN114067221B (en) | Remote sensing image woodland extraction method, system, device and medium | |
| CN112906478B (en) | Target object identification method, device, equipment and storage medium | |
| CN113780297A (en) | Image processing method, device, equipment and storage medium | |
| CN110487787B (en) | Component loss detection method and device, storage medium and terminal equipment | |
| CN112581441A (en) | Pavement crack detection method and device, electronic equipment and storage medium | |
| CN116071770B (en) | Method, device, equipment and medium for general identification of form |
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: 20210903 |