CN111950457A - Oil field safety production image identification method and system - Google Patents

Oil field safety production image identification method and system Download PDF

Info

Publication number
CN111950457A
CN111950457A CN202010808340.2A CN202010808340A CN111950457A CN 111950457 A CN111950457 A CN 111950457A CN 202010808340 A CN202010808340 A CN 202010808340A CN 111950457 A CN111950457 A CN 111950457A
Authority
CN
China
Prior art keywords
image data
design
video
video image
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
Application number
CN202010808340.2A
Other languages
Chinese (zh)
Inventor
周建峰
安军刚
李晓芳
朱运周
刘凯
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Heimer Pandora Data Technology Shenzhen Co ltd
Original Assignee
Heimer Pandora Data Technology Shenzhen Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Heimer Pandora Data Technology Shenzhen Co ltd filed Critical Heimer Pandora Data Technology Shenzhen Co ltd
Priority to CN202010808340.2A priority Critical patent/CN111950457A/en
Publication of CN111950457A publication Critical patent/CN111950457A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/40Scenes; Scene-specific elements in video content
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/048Activation functions
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/44Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an image identification method and system for oil field safety production, which comprises the following steps: the method comprises the steps that a camera device shoots an oilfield field video in real time, and preprocessing and coding are carried out on pictures in the oilfield field video in batches to obtain video image data; sequentially carrying out dark channel filtering data enhancement design, receptive field design, activation function design, channel attention mechanism design, pyramid pooling design and unbalance training design on the video image data to obtain final recognition target same-size video image data; and decoding and frame sequencing the video image data of the final recognition target with the same size to obtain a recognition video, and outputting and marking the recognition target. Has the advantages that: the influence of weather on the blur of the taken picture is overcome; the receptive field is enlarged, more useful characteristic information is extracted, and the problem that the target is small and cannot be identified is solved. Finally, the recognition target is clearer, and the recognition rate is improved.

Description

Oil field safety production image identification method and system
Technical Field
The invention relates to the technical field of petroleum safety, in particular to an image identification method and system for oil field safety production.
Background
The problems of high operation mobility and uncertainty of service operation intersection exist in the oil field safety operation, and corresponding protective equipment needs to be equipped for field operation. The environmental insecurity factors, the human insecurity, the machine insecurity and the administrative defects involved in the entire process may cause immeasurable damage and loss.
For example, in the oil field safety operation, the wearing of a safety helmet of a worker, the wearing of a worker's clothes, flame, smoke, the intrusion of a stranger, the face of a worker and the like need to be strictly controlled. Therefore, in the oilfield operation area, some identification technical measures have been taken to solve the above problems.
The existing method for identifying the wearing of the safety helmet is to directly train and learn by using YOLOv3 and then carry out wearing detection, obtain a result by using video stream detection of a previous frame image and predict position and category information of a next frame, and then cross-link and cross-link a detection frame and a prediction frame to finally realize tracking. However, the method of down-sampling per se causes difficulty in identifying small targets and blocking problems, and in the method, the targets are difficult to analyze in an oilfield production operation field from the perspective of an algorithm per se.
The work service identification is that data based on a business scene is trained, tested and verified by using the same method of safety helmet identification, then the scene image data is subjected to prediction type and marked area detection identification, and then the result is obtained and stored and evidence is obtained. The prior art method is generally utilized in the same business scene due to the aspect of service detection and identification. Although the method can solve the basic service problem through pedestrian detection under the normal weather condition, the resolution distortion of image data and video caused by the diffuse reflection of the sensor under different weather conditions causes a great amount of error recognition problems.
The flame identification is to collect and obtain evidence through the same algorithm video, image marking pretreatment, flame identification, smoke positioning smoke and flame generating area and timely alarm, and simultaneously store and obtain evidence of relevant results and store the evidence into a database for analysis and other services. The traditional flame and smoke detection also adopts the method in the prior art to identify the error phenomenon that the alarm of the foggy weather occurs.
In conclusion, the above problems cannot be solved perfectly in the safety operation of the oil field, and a great potential safety hazard still exists.
Disclosure of Invention
Aiming at the problems, the invention provides an image recognition method for oil field safety production, which is used for acquiring a video image of an oil production field, eliminating weather interference through dark channel filtering data enhancement design, enabling the image to be more accurate and clear, combining receptive field design, activation function design, channel attention mechanism design, pyramid pooling design and unbalance training design to obtain a recognition object, and rapidly and accurately extracting features. The production site condition is convenient to supervise.
In order to achieve the purpose, the invention adopts the following specific technical scheme:
the key technology of the image recognition method for oil field safety production is as follows: the method comprises the following specific steps:
s1: the method comprises the steps that a camera device shoots an oilfield field video in real time, and preprocessing and coding are carried out on pictures in the oilfield field video in batches to obtain video image data;
s2: according to the weather time interval condition, dark channel filtering data enhancement design is carried out on video image data to obtain enhanced video image data;
s3: carrying out receptive field design on the enhanced video image data to obtain video image data with a visual field determining range;
s4: determining identification characteristics, correspondingly setting identification thresholds of the identification characteristics, performing activation function design and channel attention mechanism design on the video image data in the visual field determination range, and extracting a final identification target in the video image data in the visual field determination range;
s5: carrying out pyramid pooling design and unbalanced training design on each final recognition target in the video image data of the visual field determining range to obtain the video image data of the final recognition target in the same size;
s6: and decoding and frame sequencing the video image data of the final recognition target with the same size to obtain a recognition video, and outputting and marking the recognition target.
Further, for image recognition design processing, the video image data at least includes an arrangement order of all pictures in the video, an original pixel value of each frame of picture in the video, a length value and a width value of each picture, a total number of pixels | w |, of each input image in the video image data, and a pixel variance of the input image
Figure BDA0002629981600000031
Depth value, width value, height value of each picture. Wherein the total number of pixels of each input image is | w |, the pixel variance of the input image
Figure BDA0002629981600000032
Can be obtained by intermediate calculation, and the data calculation belongs to the prior art and is not described in detail.
The further technical scheme is as follows: the enhanced video image data in step S2 is image data obtained by solving the enhanced design to a pixel value, and the image pixel value q output after the enhanced designiThe calculation function of (a) is:
qi=akIi+bk
qipixel values for an image in the output enhanced video image data; i isiIs the original pixel value of the picture;
k and i are pixel index values, ak,bkThe coefficient of the function at the moment k is the central coordinate of the current picture; the value of i is 0 to 255;
wherein the coefficient bkIs a linear cost function
Figure BDA0002629981600000033
The value corresponding to the minimum value; piThe vector of the input image is used for adjusting the image blurring degree parameter, and the value range is 0-2; sigma iewkIs an input loss function; coefficient akThe calculation formula of (2) is as follows:
Figure BDA0002629981600000041
the | w | is the total number of pixels of each input image;
Figure BDA0002629981600000042
is the pixel variance of the input image.
The constant coefficient obtained by solving by adopting the calculation method is adjusted to solve the problem that the system identification rate effect is low due to weather reasons of the image. Then, image data is marked mainly by using RGB three channels of x-y pixels for image normalization acquired by an image sensor, then, a video is subjected to framing and is marked by adopting a marking tool, semi-automatic training identification is adopted in the early stage of a training process, then, a pseudo label of a model is analyzed for residual errors of a loss function, and then, data with large fluctuation is subjected to enhanced marking training.
And a further technical scheme is that the receptive field is designed in order to extract more characteristic information. In step S3, the picture size O of the video image data with the determined range of the visual field obtained by the visual field design is:
Figure BDA0002629981600000043
wherein i' is the length value and the width value of the picture in the video, and the length value and the width value are equal; s is the convolution kernel step length; p is the filling pixel value of the corresponding visual field area when the visual field is increased; k' is the size of the convolution kernel; is an odd number between 3 and 9; d is an introduced hyper-parameter; (d-1) the number of empty cells inserted for increasing the perceived visual field.
Further describing, the identification features are a helmet wearing identification feature, a work service identification feature, a flame identification feature, a stranger intrusion identification feature and a face identification feature.
Further, the activation function is designed by adopting a Mish activation function, specifically;
Mish=x*tanh(ln(1+ex))
Figure BDA0002629981600000044
ln is logarithm, e is natural constant, x is visual field determining range video image data; tanh is high-quality characteristic image data after convolution filtering;
the activation function improves the utilization of useful information of the model, thereby better generalizing and accurately outputting high-quality characteristic information.
Further still, the channel attention mechanism design function is: :
input:c×w×h
globalpooling:c×1×1
FC:
Figure BDA0002629981600000051
FC:c×1×1
sigmoid:c×1×1
input is tan h, and c w h is the depth, width and height of the input image before the channel attention mechanism is designed; globalpopooling is global pooling; c, outputting the image depth after the channel attention mechanism is designed; sigmoid is a subfunction of the Mish activation function;
FC is the output prediction value of the full connection layer, and the output prediction value comprises the prediction target category and the probability of each prediction target category; and extracting a final recognition target in the visual field determination range video image data according to a set recognition threshold.
The method is characterized in that the calculation amount is properly increased, meanwhile, the characteristic information module design is utilized more efficiently and accurately, the attention mechanism in the field of image recognition and target detection generally focuses on soft attention, the design generally focuses on region characteristics to perform characteristic expression, and the soft attention focuses more on regions or channels to achieve the certainty of attention so as to express the attention through a network.
The SEblock mainly comprises two components of Squeeze and Excitation mapping. Because convolution is only a local space calculation problem, the importance information acquired by global information through a receptive field in a global channel information analysis network is difficult to acquire, the importance information is encoded into global characteristics by a Squeeze characteristic encoding mode, and global average pooling is used
Sequeeze is a global feature description method for extracting the relation existing between channels to express the non-linear feature which is not mutually exclusive between the channels, and meanwhile, in order to reduce the complexity of the model and improve the generalization capability of the model, an FC layer of a full-connection layer bottleck structure is adopted to reduce the dimension.
The Excitation is that the coefficient of the dimension is activated by using the hash to recover the original dimension, and finally, the signing multiplication is realized by using a sigmoid activation mode for each channel.
The channel attention of the SEblock module is mainly used for solving the problem that the target of the project is blocked and detecting, and the accuracy of the model is better improved by combining the information context of the channel information peripheral area.
Further, the pyramid pooling design adopts an SPP Block pyramid pooling module to perform scale training on each batch of pictures, and pictures with consistent sizes are output.
By the aid of the design, model updating which is only required to be finely adjusted is achieved under the condition that the oil field digital images are changed, large-scale initialization model training is not required, multi-scale training and sensitivity of the recognition solution model to different sizes are achieved, and unified prediction tensor sizes output by the model are guaranteed.
The utility model provides an oil field safety production image identification system which the key technology lies in: the cognitive computing system comprises an acquisition module, an algorithm module and a cognitive computing module, wherein the acquisition module is connected with the algorithm module, and the algorithm module is connected with the cognitive computing module; the acquisition module is internally provided with a camera device, a data screening unit and an enhancement design unit; a receptive field design unit, an activation function design unit, a channel attention mechanism design unit, a pyramid pooling design unit and an unbalance training design unit are arranged in the algorithm module; the cognitive computation module comprises a micro-service unit.
The acquisition module is used for acquiring video data, screening, enhancing, marking, sequencing, batching and the like.
The algorithm module is internally provided with a shop algorithm system, such as a face recognition algorithm, a CAFENet algorithm, an encryption container and the like, and is used for performing a receptive field design unit, an activation function design unit, a channel attention mechanism design unit, a pyramid pooling design unit and an imbalance training design unit.
The cognitive computation module comprises a micro-service unit and mainly realizes dynamic resource scheduling by using distributed micro-service computation through HTTP request density, and realizes the functions of face recognition and the like.
Compared with the prior art, the invention has the beneficial effects that:
1. the design of dark channel filtering data enhancement is adopted, and when video data acquisition is carried out under a large number of different weather conditions, guiding filtering is needed to carry out data enhancement on dark channel video images, so that image distortion caused by weather changes and different weather conditions is effectively overcome. And a dark channel filtering data enhancement design is adopted, and a constant coefficient is obtained by solving to adjust the image, so that the problem of low system identification rate effect of the image due to weather is solved.
2. By means of the receptive field design, the receptive field is expanded, and more useful characteristic information is extracted.
3. By activating function design and channel attention mechanism design, a higher-quality and more accurate recognition target is obtained. The problem of among the prior art safety helmet target little unable discernment is solved.
4. Through pyramid pooling design and unbalanced training design, the output data of the same batch are consistent, and through unbalanced weighting design, the final recognition target is clearer, and the recognition rate is improved.
Drawings
FIG. 1 is a flow chart of the identification method of the present invention;
FIG. 2 is a diagram of an arbitrary frame in a petroleum safety production site video;
FIG. 3 is a graph of the recognition effect of FIG. 2;
FIG. 4 is a block diagram of a recognition system.
Description of the drawings: figures 2 and 3 are presented using color pictures as it relates to picture recognition.
Detailed Description
The following provides a more detailed description of the embodiments and the operation of the present invention with reference to the accompanying drawings.
As can be seen from fig. 1, the method for identifying the image of the oil field safety production comprises the following specific steps:
s1: the method comprises the steps that a camera device shoots an oilfield field video in real time, and preprocessing and coding are carried out on pictures in the oilfield field video in batches to obtain video image data;
in the embodiment, the video image data includes the arrangement order of all pictures in the video, the original pixel value of each frame of picture in the video, the length value and the width value of each picture, the total number of pixels | w | of each input image in the video image data, and the pixel variance σ of the input imagek 2Depth value, width value and height value of each picture.
S2: according to the weather time interval condition, dark channel filtering data enhancement design is carried out on video image data to obtain enhanced video image data;
the enhanced video image data in step S2 is image data obtained by solving the enhanced design to a pixel value, and the image pixel value q output after the enhanced designiThe calculation function of (a) is:
qi=akIi+bk
qipixel values for an image in the output enhanced video image data; i isiIs the original pixel value of the picture;
k and i are pixel index values, ak,bkThe coefficient of the function at the moment k is the central coordinate of the current picture; the value of i is 0 to 255;
wherein the coefficient bkIs a linear cost function
Figure BDA0002629981600000081
The value corresponding to the minimum value; piIs a vector of the input image;
in order to adjust the image blur degree parameter, in the present embodiment, the value range is 2.
∑iewkIs an input loss function;
coefficient akThe calculation formula of (2) is as follows:
Figure BDA0002629981600000082
the | w | is the total number of pixels of each input image;
Figure BDA0002629981600000083
is the pixel variance of the input image.
In the present embodiment, the pixel value q is based on the output imageiEnhanced video image data can be obtained.
The constant coefficient obtained by solving by adopting the calculation method is adjusted to solve the problem that the system identification rate effect is low due to weather reasons of the image. Then, image data is marked mainly by using 906-720 pixel RGB three-channel normalization on an image collected by an image sensor, then a video is subjected to framing and is marked by adopting a marking tool, semi-automatic training identification is adopted in the early stage of a training process, then a pseudo label of a model is analyzed for residual errors of a loss function, and then data with large fluctuation is subjected to enhanced marking training.
S3: carrying out receptive field design on the enhanced video image data to obtain video image data with a visual field determining range;
in this embodiment, the formula for calculating the picture size O of the video image data with the view determining range obtained by the view design in S3 is as follows:
Figure BDA0002629981600000091
wherein i' is the length value and the width value of the picture in the video, and the length value and the width value are equal; s is the convolution kernel step length; p is the filling pixel value of the corresponding visual field area when the visual field is increased;
k' is the size of the convolution kernel; is an odd number between 3 and 9;
in this embodiment, in the convolution calculation, the convolution kernels are set to 9 and 7 in the initial stage, and the convolution kernel size is set to 3 finally as the convolution progresses.
d is an introduced hyper-parameter; (d-1) the number of empty cells inserted for increasing the perceived visual field.
S4: determining identification characteristics, correspondingly setting identification thresholds of the identification characteristics, performing activation function design and channel attention mechanism design on the video image data in the visual field determination range, and extracting a final identification target in the video image data in the visual field determination range;
in this embodiment, the identification features include a helmet wearing identification feature, a work service identification feature, a flame identification feature, a stranger intrusion identification feature, and a face identification feature.
In this embodiment, the activation function is designed by using a Mish activation function, specifically;
Mish=x*tanh(ln(1+ex))
Figure BDA0002629981600000101
ln is logarithm, e is natural constant, x is visual field determining range video image data; tanh is high-quality characteristic image data after convolution filtering; the channel attention mechanism design function is as follows:
input:c×w×h
globalpooling:c×1×1
FC:
Figure BDA0002629981600000102
FC:c×1×1
sigmoid:c×1×1
input is tan h, and c w h is the depth, width and height of the input image before the channel attention mechanism is designed; globalpopooling is global pooling; c, outputting the image depth after the channel attention mechanism is designed; sigmoid is a subfunction of the Mish activation function;
FC is the output prediction value of the full connection layer, and the output prediction value comprises the prediction target category and the probability of each prediction target category; and extracting a final recognition target in the visual field determination range video image data according to a set recognition threshold.
S5: carrying out pyramid pooling design and unbalanced training design on each final recognition target in the video image data of the visual field determining range to obtain the video image data of the final recognition target in the same size;
and the pyramid pooling design adopts an SPP Block pyramid pooling module to perform scale training on each batch of pictures and outputs the pictures with consistent sizes.
S6: and decoding and frame sequencing the video image data of the final recognition target with the same size to obtain a recognition video, and outputting and marking the recognition target.
An image recognition system for oil field safety production can be seen by combining with a figure 4, and comprises an acquisition module 1, an algorithm module 2 and a cognitive calculation module 3, wherein the acquisition module 1 is connected with the algorithm module 2, and the algorithm module 2 is connected with the cognitive calculation module 3; the acquisition module 1 is internally provided with a camera device, a data screening unit and an enhancement design unit; a receptive field design unit, an activation function design unit, a channel attention mechanism design unit, a pyramid pooling design unit and an unbalance training design unit are arranged in the algorithm module 2; the cognitive computation module 3 comprises a microservice unit.
In this embodiment, as can be seen from a comparison between fig. 2 and fig. 3, the mark recognition target is marked by frame selection using a wire frame of green, red, purple, blue, or the like.
After outputting the identification target, the identification tag is output, and in this embodiment, the identification tag includes: normal, nohat (no safety helmet worn), nocoverall (no workwear), flame (fire), smoke (smoke).
In the present embodiment, as can be seen in fig. 3, a green wire frame is used to select the identification target without wearing the safety helmet; selecting an identification target for wearing the safety helmet by adopting a blue line frame;
in the present embodiment, as can be seen from fig. 3, the purple wire frame is used to select the identification target of the work clothes; and selecting the recognition target without wearing the frock clothes by using a red wire frame.
It should be noted that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make variations, modifications, additions or substitutions within the spirit and scope of the present invention.

Claims (8)

1. An image identification method for oil field safety production is characterized by comprising the following specific steps:
s1: the method comprises the steps that a camera device shoots an oilfield field video in real time, and preprocessing and coding are carried out on pictures in the oilfield field video in batches to obtain video image data;
s2: according to the weather time interval condition, dark channel filtering data enhancement design is carried out on video image data to obtain enhanced video image data;
s3: carrying out receptive field design on the enhanced video image data to obtain video image data with a visual field determining range;
s4: determining identification characteristics, correspondingly setting identification thresholds of the identification characteristics, performing activation function design and channel attention mechanism design on the video image data in the visual field determination range, and extracting a final identification target in the video image data in the visual field determination range;
s5: carrying out pyramid pooling design and unbalanced training design on each final recognition target in the video image data of the visual field determining range to obtain the video image data of the final recognition target in the same size;
s6: and decoding and frame sequencing the video image data of the final recognition target with the same size to obtain a recognition video, and outputting and marking the recognition target.
2. The image recognition method for oilfield safety production according to claim 1, characterized in that: the video image data at least comprises the arrangement sequence of all pictures in the video, the original pixel value of each frame of picture in the video, the length value and the width value of each picture, the total pixel number | w | of each input image in the video image data, and the pixel variance of the input image
Figure FDA0002629981590000011
Depth value, width value, height value of each picture.
3. The image recognition method for oilfield safety production according to claim 2, characterized in that: the enhanced video image data in step S2 is image data obtained by solving the enhanced design to a pixel value, and the image pixel value q output after the enhanced designiThe calculation function of (a) is:
qi=akIi+bk
qipixel values for an image in the output enhanced video image data; i isiIs the original pixel value of the picture;
k and i are pixel index values, ak,bkThe coefficient of the function at the moment k is the central coordinate of the current picture; the value of i is 0 to 255;
wherein the coefficient bkIs a linear cost function
Figure FDA0002629981590000021
The value corresponding to the minimum value; piThe vector of the input image is used for adjusting the image blurring degree parameter, and the value range is 0-2; sigma iewkIs an input loss function; coefficient akThe calculation formula of (2) is as follows:
Figure FDA0002629981590000022
the | w | is the total number of pixels of each input image;
Figure FDA0002629981590000023
is the pixel variance of the input image.
4. The image recognition method for oilfield safety production according to claim 2, characterized in that: the picture size O of the video image data of the view determination range obtained after the view design in step S3 is:
Figure FDA0002629981590000024
wherein i' is the length value and the width value of the picture in the video, and the length value and the width value are equal; s is the convolution kernel step length; p is the filling pixel value of the corresponding visual field area when the visual field is increased; k' is the size of the convolution kernel; is an odd number between 3 and 9; d is an introduced hyper-parameter; (d-1) the number of empty cells inserted for increasing the perceived visual field.
5. The image recognition method for oilfield safety production according to claim 2, characterized in that: the identification features are safety helmet wearing identification features, worker clothes identification features, flame identification features, stranger intrusion identification features and face identification features.
6. The image recognition method for oilfield safety production according to claim 5, wherein: the activation function is designed by adopting a Mish activation function, and specifically, the Mish activation function is designed;
Mish=x*tanh(ln(1+ex))
Figure FDA0002629981590000031
ln is logarithm, e is natural constant, x is visual field determining range video image data; tanh is high-quality characteristic image data after convolution filtering;
the channel attention mechanism design function is as follows: :
input:c×w×h
globalpooling:c×1×1
FC:
Figure FDA0002629981590000032
FC:c×1×1
sigmoid:c×1×1
input is tan h, and c w h is the depth, width and height of the input image before the channel attention mechanism is designed; globalpopooling is global pooling; c, outputting the image depth after the channel attention mechanism is designed; sigmoid is a subfunction of the Mish activation function;
FC is the output prediction value of the full connection layer, and the output prediction value comprises the prediction target category and the probability of each prediction target category; and extracting a final recognition target in the visual field determination range video image data according to a set recognition threshold.
7. The image recognition method for oilfield safety production according to claim 6, wherein: and the pyramid pooling design adopts an SPP Block pyramid pooling module to perform scale training on each batch of pictures and outputs the pictures with consistent sizes.
8. The oil field safety production image recognition system is characterized in that: the cognitive computing system comprises an acquisition module (1), an algorithm module (2) and a cognitive computing module (3), wherein the acquisition module (1) is connected with the algorithm module (2), and the algorithm module (2) is connected with the cognitive computing module (3);
the acquisition module (1) is internally provided with a camera device, a data screening unit and an enhancement design unit;
a receptive field design unit, an activation function design unit, a channel attention mechanism design unit, a pyramid pooling design unit and an unbalance training design unit are arranged in the algorithm module (2);
the cognitive computation module (3) comprises a micro-service unit.
CN202010808340.2A 2020-08-12 2020-08-12 Oil field safety production image identification method and system Pending CN111950457A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010808340.2A CN111950457A (en) 2020-08-12 2020-08-12 Oil field safety production image identification method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010808340.2A CN111950457A (en) 2020-08-12 2020-08-12 Oil field safety production image identification method and system

Publications (1)

Publication Number Publication Date
CN111950457A true CN111950457A (en) 2020-11-17

Family

ID=73332792

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010808340.2A Pending CN111950457A (en) 2020-08-12 2020-08-12 Oil field safety production image identification method and system

Country Status (1)

Country Link
CN (1) CN111950457A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113191983A (en) * 2021-05-18 2021-07-30 陕西师范大学 Image denoising method and device based on deep learning attention mechanism
CN113610043A (en) * 2021-08-19 2021-11-05 海默潘多拉数据科技(深圳)有限公司 Industrial drawing table structured recognition method and system
CN114169385A (en) * 2021-09-28 2022-03-11 北京工业大学 MSWI process combustion state identification method based on mixed data enhancement
CN118172714A (en) * 2024-05-15 2024-06-11 中国石油大学(华东) Multi-scale construction equipment robustness identification method in oilfield underground operation environment

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103236038A (en) * 2013-04-10 2013-08-07 成都国腾电子技术股份有限公司 Quick defogging processing assembly for fog images
CN110728223A (en) * 2019-10-08 2020-01-24 济南东朔微电子有限公司 Helmet wearing identification method based on deep learning
CN111199523A (en) * 2019-12-24 2020-05-26 深圳供电局有限公司 Power equipment identification method and device, computer equipment and storage medium
CN111507271A (en) * 2020-04-20 2020-08-07 北京理工大学 Airborne photoelectric video target intelligent detection and identification method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103236038A (en) * 2013-04-10 2013-08-07 成都国腾电子技术股份有限公司 Quick defogging processing assembly for fog images
CN110728223A (en) * 2019-10-08 2020-01-24 济南东朔微电子有限公司 Helmet wearing identification method based on deep learning
CN111199523A (en) * 2019-12-24 2020-05-26 深圳供电局有限公司 Power equipment identification method and device, computer equipment and storage medium
CN111507271A (en) * 2020-04-20 2020-08-07 北京理工大学 Airborne photoelectric video target intelligent detection and identification method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HE K 等: "Guided image filtering", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》, pages 1397 - 1409 *
核桃分心木: "暗通道先验之Guided filter导向滤波", pages 1 - 2, Retrieved from the Internet <URL:https://zhuanlan.zhihu.com/p/36813673> *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113191983A (en) * 2021-05-18 2021-07-30 陕西师范大学 Image denoising method and device based on deep learning attention mechanism
CN113610043A (en) * 2021-08-19 2021-11-05 海默潘多拉数据科技(深圳)有限公司 Industrial drawing table structured recognition method and system
CN114169385A (en) * 2021-09-28 2022-03-11 北京工业大学 MSWI process combustion state identification method based on mixed data enhancement
CN114169385B (en) * 2021-09-28 2024-04-09 北京工业大学 MSWI process combustion state identification method based on mixed data enhancement
CN118172714A (en) * 2024-05-15 2024-06-11 中国石油大学(华东) Multi-scale construction equipment robustness identification method in oilfield underground operation environment

Similar Documents

Publication Publication Date Title
CN111178183B (en) Face detection method and related device
CN111950457A (en) Oil field safety production image identification method and system
CN109684925B (en) Depth image-based human face living body detection method and device
Wang et al. Investigation into recognition algorithm of helmet violation based on YOLOv5-CBAM-DCN
Lin et al. MSAFF-Net: Multiscale attention feature fusion networks for single image dehazing and beyond
CN110570435B (en) Method and device for carrying out damage segmentation on vehicle damage image
CN108229418B (en) Human body key point detection method and apparatus, electronic device, storage medium, and program
CN109978854B (en) Screen content image quality evaluation method based on edge and structural features
CN112396635B (en) Multi-target detection method based on multiple devices in complex environment
CN109635634A (en) A kind of pedestrian based on stochastic linear interpolation identifies data enhancement methods again
CN112861690A (en) Multi-method fused remote sensing image change detection method and system
CN111325051A (en) Face recognition method and device based on face image ROI selection
CN111339902A (en) Liquid crystal display number identification method and device of digital display instrument
CN117636134B (en) Panoramic image quality evaluation method and system based on hierarchical moving window attention
CN114581432A (en) Tongue appearance tongue image segmentation method based on deep learning
CN104361357A (en) Photo set classification system and method based on picture content analysis
CN116311214B (en) License plate recognition method and device
CN115375991A (en) Strong/weak illumination and fog environment self-adaptive target detection method
CN112989958A (en) Helmet wearing identification method based on YOLOv4 and significance detection
Gao et al. Let you see in haze and sandstorm: Two-in-one low-visibility enhancement network
CN116704324A (en) Target detection method, system, equipment and storage medium based on underwater image
CN105469413B (en) It is a kind of based on normalization ring weighting without refer to smear restoration image synthesis method for evaluating quality
CN114299590A (en) Training method of face completion model, face completion method and system
CN113221812A (en) Training method of face key point detection model and face key point detection method
CN115240123B (en) Intelligent monitoring system-oriented method for detecting violent behaviors in dark place

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