CN111639546A - Small-scale target cloud computing identification method and device based on neural network - Google Patents
Small-scale target cloud computing identification method and device based on neural network Download PDFInfo
- Publication number
- CN111639546A CN111639546A CN202010384329.8A CN202010384329A CN111639546A CN 111639546 A CN111639546 A CN 111639546A CN 202010384329 A CN202010384329 A CN 202010384329A CN 111639546 A CN111639546 A CN 111639546A
- Authority
- CN
- China
- Prior art keywords
- data
- cloud computing
- neural network
- video stream
- target
- 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
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 25
- 238000001514 detection method Methods 0.000 claims abstract description 45
- 238000003062 neural network model Methods 0.000 claims abstract description 20
- 238000012544 monitoring process Methods 0.000 claims abstract description 17
- 235000019504 cigarettes Nutrition 0.000 claims description 19
- 239000004816 latex Substances 0.000 claims description 5
- 229920000126 latex Polymers 0.000 claims description 5
- 238000003384 imaging method Methods 0.000 abstract description 6
- 238000012545 processing Methods 0.000 abstract description 2
- 230000003321 amplification Effects 0.000 abstract 1
- 238000003199 nucleic acid amplification method Methods 0.000 abstract 1
- 238000010276 construction Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000000391 smoking effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 239000000779 smoke Substances 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002265 prevention Effects 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
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Computer Interaction (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the field of video processing, in particular to a small-scale target cloud computing identification method and device based on a neural network. The small-scale target cloud computing identification method based on the neural network comprises the following steps: acquiring video stream data of a monitoring camera, detecting whether a person exists in the video stream data, and if the person exists in the video stream data, intercepting a current frame picture and storing the current frame picture as first data; acquiring a first parameter in the first data, if the first parameter value is smaller than a preset parameter value, carrying out amplification operation on the current frame picture, intercepting the amplified frame picture and storing the frame picture as second data; and carrying out target recognition on the second data through the pre-trained neural network model to obtain a target recognition result. According to the method, the image with the portrait is processed, the image is amplified, the amplified image can be used for better target identification, and therefore the target size is small even if the target is far away from an imaging system, and accurate detection can be achieved through the method.
Description
Technical Field
The invention relates to the field of video processing, in particular to a small-scale target cloud computing identification method and device based on a neural network.
Background
In a construction site, safety is always the first place, and safety issues are among the prevention. Although construction enterprises have requirements for employees to comply with safety rules, such as: when in construction, the user needs to wear safety helmets, do not smoke, do not use mobile phones, wear work clothes and the like. However, many workers deal with the inspection, wear safety helmets, do not smoke, do not use mobile phones, wear work clothes when leadership inspection exists, and randomly do not lead inspection, which brings great hidden danger to safety.
In order to guarantee and supervise unsafe behaviors of workers who do not wear safety helmets, smoke, play mobile phones, do not wear work clothes and the like, a construction party generally implements deployment of the camera intelligent detection system. The camera intelligent detection method is adopted to realize the identification and detection of unsafe behaviors, and has the advantages of low cost, rapid deployment, convenient maintenance and the like.
However, the traditional camera target detection has the following defects: when the target is far away from the imaging system, the target size is small, the image of the target only occupies a single or a plurality of pixels in the scene image, the characteristics of the target are not obvious enough, and the target is easy to be interfered by image noise to finally cause no detection and detection omission; in addition, in order to improve detection accuracy and guarantee real-time performance, a camera adopting a deep learning detection algorithm needs to be equipped with GPU hardware which is stronger in computational power resources than a CPU, so that deployment cost is increased.
Disclosure of Invention
Therefore, a small-scale target cloud computing identification method based on a neural network is needed to be provided for solving the problem that the identification rate is low when the distance from an imaging system is far away at the construction site. The specific technical scheme is as follows:
a small-scale target cloud computing identification method based on a neural network comprises the following steps:
acquiring video stream data of a monitoring camera, detecting whether a person exists in the video stream data, and if the person exists in the video stream data, capturing a current frame picture and storing the current frame picture as first data;
acquiring a first parameter in the first data, judging whether the first parameter value is smaller than a preset parameter value, if so, amplifying the current frame picture, and capturing the amplified frame picture to store as second data, wherein the first parameter comprises: human body pixel area;
performing target recognition on the second data through a pre-trained neural network model to obtain a target recognition result, wherein the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes.
Further, the "performing target recognition on the second data through the pre-trained neural network model" further includes the steps of:
and sending the second data to a cloud computing platform, wherein the cloud computing platform allocates GPU computing power resources as required, and the cloud computing platform performs target recognition on the second data through a pre-trained neural network model.
Further, the "detecting whether there is a portrait in the video stream data" further includes the steps of:
detecting whether a portrait exists in the video stream data through a human body detection algorithm, wherein the human body detection algorithm comprises one or more of the following steps: HOG + AdaBoost, ICF + AdaBoost, DPM + latex SVM.
Further, the method also comprises the following steps:
and judging whether the employee complies with safety rules according to the target identification result.
Further, the preset parameter value is a minimum detection pixel area value.
In order to solve the technical problem, the small-scale target cloud computing and identifying device based on the neural network is further provided, and the specific technical scheme is as follows:
a small-scale target cloud computing identification device based on a neural network comprises: the device comprises a video stream acquisition unit, a video stream detection unit, a picture adjustment unit and a cloud computing unit;
the video stream acquisition unit is configured to: acquiring video stream data of a monitoring camera;
the video stream detection unit is configured to: detecting whether a portrait exists in the video stream data, if so, capturing a current frame picture and storing the current frame picture as first data;
the picture adjusting unit is used for: acquiring a first parameter in the first data, judging whether the first parameter value is smaller than a preset parameter value, if so, amplifying the current frame picture, and capturing the amplified frame picture to store as second data, wherein the first parameter comprises: human body pixel area;
the cloud computing unit is to: performing target recognition on the second data through a pre-trained neural network model to obtain a target recognition result, wherein the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes.
Further, the cloud computing unit is further configured to: GPU computational power resources are allocated as needed.
Further, the video stream detection unit is further configured to: detecting whether a portrait exists in the video stream data through a human body detection algorithm, wherein the human body detection algorithm comprises one or more of the following steps: HOG + AdaBoost, ICF + AdaBoost, DPM + latex SVM.
Further, the cloud computing unit is further configured to: and judging whether the employee complies with safety rules according to the target identification result.
Further, the preset parameter value is a minimum detection pixel area value.
The invention has the beneficial effects that: the method comprises the steps of acquiring video stream data of a monitoring camera, detecting whether a person exists in the video stream data, and if the person exists in the video stream data, intercepting a current frame picture and storing the current frame picture as first data; the first parameter comprises a human body pixel area, whether the first parameter value is smaller than a preset parameter value or not is judged, if the first parameter value is smaller than the preset parameter value, a current frame picture is amplified, the amplified frame picture is intercepted and stored as second data, the amplified picture can be better subjected to data recognition, the data recognition accuracy is improved, target recognition is carried out on the second data according to a pre-trained neural network model to obtain a target recognition result, and the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes. And judging whether the employee complies with the safety rules or not according to the target identification result. According to the method, the picture with the portrait is processed, when the area value of the human body pixel in the picture is lower than the preset parameter value, the picture is amplified, the amplified picture can be used for better target identification, so that the target size is small even if the target is far away from an imaging system, accurate detection can be achieved through the method, meanwhile, a camera does not need to be additionally provided with GPU hardware with stronger computing resources, and the deployment cost is greatly reduced.
Drawings
Fig. 1 is a flowchart of a small-scale target cloud computing identification method based on a neural network according to an embodiment;
FIG. 2 is a diagram illustrating a first parameter value being less than a predetermined parameter value according to an embodiment;
FIG. 3 is a diagram illustrating an adjusted frame according to an embodiment;
fig. 4 is a schematic block diagram of a small-scale target cloud computing identification apparatus based on a neural network according to an embodiment.
Description of reference numerals:
400. a small-scale target cloud computing identification device based on a neural network,
401. a video stream acquisition unit for acquiring a video stream,
402. a video stream detection unit for detecting a video stream,
403. a picture adjusting unit for adjusting the picture of the image,
404. and a cloud computing unit.
Detailed Description
To explain technical contents, structural features, and objects and effects of the technical solutions in detail, the following detailed description is given with reference to the accompanying drawings in conjunction with the embodiments.
Referring to fig. 1 to 3, in the present embodiment, a small-scale target cloud computing identification method based on a neural network is implemented as follows:
step S101: and acquiring video stream data of the monitoring camera. The video stream data of the monitoring camera is preferably acquired in real time, and real-time monitoring can be realized without detection omission. In other embodiments, different time intervals can be set according to actual conditions to obtain video stream data of the monitoring camera, so that the calculation amount can be reduced, and the calculation resources can be greatly saved.
After the video stream data of the monitoring camera is acquired, executing the step S102: is there a portrait in the video stream data? If there is a person in the video stream data, step S103 is executed: and intercepting the current frame picture and storing the current frame picture as first data.
Whether a portrait exists in the video stream data is detected through a human body detection algorithm, wherein the human body detection algorithm comprises one or more of the following steps: HOG + AdaBoost, ICF + AdaBoost, DPM + latex SVM.
Step S104: and acquiring a first parameter in the first data.
After the first parameter in the first data is obtained, step S105 is executed: is the first parameter value less than a preset parameter value? If the first parameter value is smaller than the preset parameter value, executing step S106: amplifying the current frame, intercepting the amplified frame and storing the frame as second data, wherein the first parameter comprises: area of human body pixels.
Specifically, when the area of the human body pixel is more than or equal to 500 pixels, the picture does not need to be adjusted; when the area of the human body pixel is smaller than 500 pixels, as shown in fig. 2, the image is amplified, and the method of increasing the optical zoom factor of the camera step by step and detecting the area of the human body pixel is adopted until the area of the human body pixel is larger than or equal to 500 pixels, as shown in fig. 3. And intercepting the amplified frame picture and storing the frame picture as second data.
And if the first parameter value is larger than or equal to a preset parameter value, the current frame picture is not adjusted.
Wherein the preset parameter value is a minimum detection pixel area value.
In other embodiments, the video picture can also be adjusted in time according to the proportional relationship between the human body and the small-scale objects such as the safety helmet, the cigarette, the mobile phone, the work clothes and the like in the picture, and when the human body is enlarged to be above the preset pixel area, the small-scale objects such as the safety helmet, the cigarette, the mobile phone, the work clothes and the like in the original picture are changed in scale, so that the occupation ratio in the picture reaches the degree of easy detection.
Specifically, the proportion of different small-scale targets to the human body is different according to design models and practical experience. The area of the human body pixel is required to be more than or equal to 500 pixels by safety helmets and work clothes, and the area of the human body pixel is required to be more than or equal to 5000 pixels by cigarettes and mobile phones.
Step S107: performing target recognition on the second data through a pre-trained neural network model to obtain a target recognition result, wherein the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes.
For the pre-trained neural network model, models such as VGG and Resnet are used. The model is formed by marking a large amount of image data acquired on site in a construction site by using labelme and other marking tools according to labels of a helmet, such as worn, not worn, smoking, not smoking, mobile phone, not mobile phone, wearing work clothes, not wearing work clothes and the like, training the marked data by using a CNN convolutional neural network to obtain a target recognition neural network model of the helmet, the cigarette, the mobile phone, the work clothes and the like, and the model can be used for recognizing and detecting targets of the helmet, the cigarette, the mobile phone, the work clothes and the like on the newly acquired image data.
Furthermore, whether the staff obeys the safety rules can be judged according to the target identification result. When the safety helmet is identified to be not worn or smoked or a mobile phone is played or the worker clothes are not worn, the worker is judged not to comply with the safety rules.
Further, the "performing target recognition on the second data through the pre-trained neural network model" further includes the steps of:
and sending the second data to a cloud computing platform, wherein the cloud computing platform allocates GPU computing power resources as required, and the cloud computing platform performs target recognition on the second data through a pre-trained neural network model. And target identification is carried out on the second data through the cloud computing platform, so that the camera does not need GPU hardware which is stronger than CPU computational resources, and the deployment cost is greatly reduced.
The method comprises the steps of acquiring video stream data of a monitoring camera, detecting whether a person exists in the video stream data, and if the person exists in the video stream data, intercepting a current frame picture and storing the current frame picture as first data; the first parameter comprises a human body pixel area, whether the first parameter value is smaller than a preset parameter value or not is judged, if the first parameter value is smaller than the preset parameter value, a current frame picture is amplified, the amplified frame picture is intercepted and stored as second data, the amplified picture can be better subjected to data recognition, the data recognition accuracy is improved, target recognition is carried out on the second data according to a pre-trained neural network model to obtain a target recognition result, and the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes. And judging whether the employee complies with the safety rules or not according to the target identification result. According to the method, the picture with the portrait is processed, when the area value of the human body pixel in the picture is lower than the preset parameter value, the picture is amplified, the amplified picture can be used for better target identification, so that the target size is small even if the target is far away from an imaging system, accurate detection can be achieved through the method, meanwhile, a camera does not need to be additionally provided with GPU hardware with stronger computing resources, and the deployment cost is greatly reduced.
Referring to fig. 2 to 4, in the present embodiment, an embodiment of a small-scale target cloud computing recognition apparatus 400 based on a neural network is as follows:
a neural network-based small-scale target cloud computing recognition device 400, comprising: a video stream acquisition unit 401, a video stream detection unit 402, a picture adjustment unit 403, and a cloud computing unit 404;
the video stream acquiring unit 401 is configured to: acquiring video stream data of a monitoring camera; the video stream data of the monitoring camera is preferably acquired in real time, and real-time monitoring can be realized without detection omission. In other embodiments, different time intervals can be set according to actual conditions to obtain video stream data of the monitoring camera, so that the calculation amount can be reduced, and the calculation resources can be greatly saved.
The video stream detection unit 402 is configured to: detecting whether a portrait exists in the video stream data, if so, capturing a current frame picture and storing the current frame picture as first data;
wherein the video stream detection unit 402 is further configured to: detecting whether a portrait exists in the video stream data through a human body detection algorithm, wherein the human body detection algorithm comprises one or more of the following steps: HOG + AdaBoost, ICF + AdaBoost, DPM + latex SVM.
The picture adjusting unit 403 is configured to: acquiring a first parameter in the first data, judging whether the first parameter value is smaller than a preset parameter value, if so, amplifying the current frame picture, and capturing the amplified frame picture to store as second data, wherein the first parameter comprises: human body pixel area;
specifically, when the area of the human body pixel is more than or equal to 500 pixels, the picture does not need to be adjusted; when the area of the human body pixel is smaller than 500 pixels, as shown in fig. 2, the image is amplified, and the method of increasing the optical zoom factor of the camera step by step and detecting the area of the human body pixel is adopted until the area of the human body pixel is larger than or equal to 500 pixels, as shown in fig. 3. And intercepting the amplified frame picture and storing the frame picture as second data.
And if the first parameter value is larger than or equal to a preset parameter value, the current frame picture is not adjusted.
Wherein the preset parameter value is a minimum detection pixel area value.
In other embodiments, the video picture can also be adjusted in time according to the proportional relationship between the human body and the small-scale objects such as the safety helmet, the cigarette, the mobile phone, the work clothes and the like in the picture, and when the human body is enlarged to be above the preset pixel area, the small-scale objects such as the safety helmet, the cigarette, the mobile phone, the work clothes and the like in the original picture are changed in scale, so that the occupation ratio in the picture reaches the degree of easy detection.
Specifically, the proportion of different small-scale targets to the human body is different according to design models and practical experience. The area of the human body pixel is required to be more than or equal to 500 pixels by safety helmets and work clothes, and the area of the human body pixel is required to be more than or equal to 5000 pixels by cigarettes and mobile phones.
The cloud computing unit 404 is configured to: performing target recognition on the second data through a pre-trained neural network model to obtain a target recognition result, wherein the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes. A
For the pre-trained neural network model, models such as VGG and Resnet are used. The model is formed by marking a large amount of image data acquired on site in a construction site by using labelme and other marking tools according to labels of a helmet, such as worn, not worn, smoking, not smoking, mobile phone, not mobile phone, wearing work clothes, not wearing work clothes and the like, training the marked data by using a CNN convolutional neural network to obtain a target recognition neural network model of the helmet, the cigarette, the mobile phone, the work clothes and the like, and the model can be used for recognizing and detecting targets of the helmet, the cigarette, the mobile phone, the work clothes and the like on the newly acquired image data.
Further, the cloud computing unit 404 is further configured to: and judging whether the employee complies with safety rules according to the target identification result. When the safety helmet is identified to be not worn or smoked or a mobile phone is played or the worker clothes are not worn, the worker is judged not to comply with the safety rules.
Further, the cloud computing unit 404 is further configured to: GPU computational power resources are allocated as needed. And target identification is carried out on the second data through the cloud computing platform, so that the camera does not need GPU hardware which is stronger than CPU computational resources, and the deployment cost is greatly reduced.
Acquiring video stream data of a monitoring camera through a video stream acquisition unit 401, detecting whether a person exists in the video stream data through a video stream detection unit 402, and if the person exists in the video stream data, intercepting a current frame picture and storing the current frame picture as first data; the first parameter includes a human body pixel area, whether the first parameter value is smaller than a preset parameter value is judged through the picture adjusting unit 403, if the first parameter value is smaller than the preset parameter value, a current frame picture is amplified, the amplified frame picture is captured and stored as second data, the amplified picture can better perform data recognition, the data recognition accuracy is improved, the cloud computing unit 404 performs target recognition on the second data according to a pre-trained neural network model to obtain a target recognition result, and the target includes one or more of the following: safety helmets, cigarettes, mobile phones and work clothes. And judging whether the employee complies with the safety rules or not according to the target identification result. According to the method, the picture with the portrait is processed, when the area value of the human body pixel in the picture is lower than the preset parameter value, the picture is amplified, the amplified picture can be used for better target identification, so that the target size is small even if the target is far away from an imaging system, accurate detection can be achieved through the method, meanwhile, a video stream acquisition unit 401 in the device, namely a camera does not need to additionally install GPU hardware with stronger computing resources, and deployment cost is greatly reduced.
It should be noted that, although the above embodiments have been described herein, the invention is not limited thereto. Therefore, based on the innovative concepts of the present invention, the technical solutions of the present invention can be directly or indirectly applied to other related technical fields by making changes and modifications to the embodiments described herein, or by using equivalent structures or equivalent processes performed in the content of the present specification and the attached drawings, which are included in the scope of the present invention.
Claims (10)
1. A small-scale target cloud computing identification method based on a neural network is characterized by comprising the following steps:
acquiring video stream data of a monitoring camera, detecting whether a person exists in the video stream data, and if the person exists in the video stream data, capturing a current frame picture and storing the current frame picture as first data;
acquiring a first parameter in the first data, judging whether the first parameter value is smaller than a preset parameter value, if so, amplifying the current frame picture, and capturing the amplified frame picture to store as second data, wherein the first parameter comprises: human body pixel area;
performing target recognition on the second data through a pre-trained neural network model to obtain a target recognition result, wherein the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes.
2. The small-scale target cloud computing identification method based on the neural network as claimed in claim 1,
the method for performing target recognition on the second data through the pre-trained neural network model further comprises the following steps of:
and sending the second data to a cloud computing platform, wherein the cloud computing platform allocates GPU computing power resources as required, and the cloud computing platform performs target recognition on the second data through a pre-trained neural network model.
3. The small-scale target cloud computing identification method based on the neural network as claimed in claim 1,
the method for detecting whether the portrait exists in the video stream data further comprises the following steps:
detecting whether a portrait exists in the video stream data through a human body detection algorithm, wherein the human body detection algorithm comprises one or more of the following steps: HOG + AdaBoost, ICF + AdaBoost, DPM + latex SVM.
4. The small-scale target cloud computing identification method based on the neural network as claimed in claim 1, further comprising the steps of:
and judging whether the employee complies with safety rules according to the target identification result.
5. The small-scale target cloud computing identification method based on the neural network as claimed in claim 1,
the preset parameter value is the minimum detection pixel area value.
6. A small-scale target cloud computing identification device based on a neural network is characterized by comprising the following components: the device comprises a video stream acquisition unit, a video stream detection unit, a picture adjustment unit and a cloud computing unit;
the video stream acquisition unit is configured to: acquiring video stream data of a monitoring camera;
the video stream detection unit is configured to: detecting whether a portrait exists in the video stream data, if so, capturing a current frame picture and storing the current frame picture as first data;
the picture adjusting unit is used for: acquiring a first parameter in the first data, judging whether the first parameter value is smaller than a preset parameter value, if so, amplifying the current frame picture, and capturing the amplified frame picture to store as second data, wherein the first parameter comprises: human body pixel area;
the cloud computing unit is to: performing target recognition on the second data through a pre-trained neural network model to obtain a target recognition result, wherein the target comprises one or more of the following: safety helmets, cigarettes, mobile phones and work clothes.
7. The small-scale target cloud computing identification device based on the neural network as claimed in claim 6,
the cloud computing unit is further to: GPU computational power resources are allocated as needed.
8. The small-scale target cloud computing identification device based on the neural network as claimed in claim 6,
the video stream detection unit is further configured to: detecting whether a portrait exists in the video stream data through a human body detection algorithm, wherein the human body detection algorithm comprises one or more of the following steps: HOG + AdaBoost, ICF + AdaBoost, and DPM + latentSVM.
9. The small-scale target cloud computing identification device based on the neural network as claimed in claim 6,
the cloud computing unit is further to: and judging whether the employee complies with safety rules according to the target identification result.
10. The small-scale target cloud computing identification device based on the neural network as claimed in claim 6,
the preset parameter value is the minimum detection pixel area value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010384329.8A CN111639546A (en) | 2020-05-07 | 2020-05-07 | Small-scale target cloud computing identification method and device based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010384329.8A CN111639546A (en) | 2020-05-07 | 2020-05-07 | Small-scale target cloud computing identification method and device based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111639546A true CN111639546A (en) | 2020-09-08 |
Family
ID=72329135
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010384329.8A Pending CN111639546A (en) | 2020-05-07 | 2020-05-07 | Small-scale target cloud computing identification method and device based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111639546A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112200092A (en) * | 2020-10-13 | 2021-01-08 | 深圳龙岗智能视听研究院 | Intelligent smoking detection method based on variable-focus movement of dome camera |
| CN114155367A (en) * | 2022-02-09 | 2022-03-08 | 北京阿丘科技有限公司 | Method, device and equipment for detecting defects of printed circuit board and storage medium |
| CN118314502A (en) * | 2024-04-25 | 2024-07-09 | 联合智慧技术有限公司 | Video and image content identification grabbing method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108304787A (en) * | 2018-01-17 | 2018-07-20 | 河南工业大学 | Road target detection method based on convolutional neural networks |
| CN108319926A (en) * | 2018-02-12 | 2018-07-24 | 安徽金禾软件股份有限公司 | A kind of the safety cap wearing detecting system and detection method of building-site |
| WO2018133666A1 (en) * | 2017-01-17 | 2018-07-26 | 腾讯科技（深圳）有限公司 | Method and apparatus for tracking video target |
| CN108460358A (en) * | 2018-03-20 | 2018-08-28 | 武汉倍特威视系统有限公司 | Safety cap recognition methods based on video stream data |
| CN109117827A (en) * | 2018-09-05 | 2019-01-01 | 武汉市蓝领英才科技有限公司 | Work clothes work hat wearing state automatic identifying method and alarm system based on video |
| WO2019128304A1 (en) * | 2017-12-29 | 2019-07-04 | 南京阿凡达机器人科技有限公司 | Human body fall-down detection method and device |
| CN110287804A (en) * | 2019-05-30 | 2019-09-27 | 广东电网有限责任公司 | A kind of electric operating personnel's dressing recognition methods based on mobile video monitor |
| 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 |
-
2020
- 2020-05-07 CN CN202010384329.8A patent/CN111639546A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018133666A1 (en) * | 2017-01-17 | 2018-07-26 | 腾讯科技（深圳）有限公司 | Method and apparatus for tracking video target |
| WO2019128304A1 (en) * | 2017-12-29 | 2019-07-04 | 南京阿凡达机器人科技有限公司 | Human body fall-down detection method and device |
| CN108304787A (en) * | 2018-01-17 | 2018-07-20 | 河南工业大学 | Road target detection method based on convolutional neural networks |
| CN108319926A (en) * | 2018-02-12 | 2018-07-24 | 安徽金禾软件股份有限公司 | A kind of the safety cap wearing detecting system and detection method of building-site |
| CN108460358A (en) * | 2018-03-20 | 2018-08-28 | 武汉倍特威视系统有限公司 | Safety cap recognition methods based on video stream data |
| CN109117827A (en) * | 2018-09-05 | 2019-01-01 | 武汉市蓝领英才科技有限公司 | Work clothes work hat wearing state automatic identifying method and alarm system based on video |
| CN110287804A (en) * | 2019-05-30 | 2019-09-27 | 广东电网有限责任公司 | A kind of electric operating personnel's dressing recognition methods based on mobile video monitor |
| 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 |
Non-Patent Citations (1)
| Title |
|---|
| 刘黎明 等: "《云计算技术应用研究》", 电子科技大学出版社, pages: 51 - 70 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112200092A (en) * | 2020-10-13 | 2021-01-08 | 深圳龙岗智能视听研究院 | Intelligent smoking detection method based on variable-focus movement of dome camera |
| CN112200092B (en) * | 2020-10-13 | 2023-06-23 | 深圳龙岗智能视听研究院 | Intelligent smoking detection method based on zoom movement of dome camera |
| CN114155367A (en) * | 2022-02-09 | 2022-03-08 | 北京阿丘科技有限公司 | Method, device and equipment for detecting defects of printed circuit board and storage medium |
| CN118314502A (en) * | 2024-04-25 | 2024-07-09 | 联合智慧技术有限公司 | Video and image content identification grabbing method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111639546A (en) | Small-scale target cloud computing identification method and device based on neural network | |
| CN108921159B (en) | Method and device for detecting wearing condition of safety helmet | |
| CN103726879B (en) | Utilize camera automatic capturing mine ore deposit to shake and cave in and the method for record warning in time | |
| CN110633612B (en) | Monitoring method and system for inspection robot | |
| CN106682620A (en) | Human face image acquisition method and device | |
| US20200057886A1 (en) | Gesture recognition method and apparatus, electronic device, and computer-readable storage medium | |
| SG11201902856QA (en) | Eyeball movement analysis method and device, and storage medium | |
| CN112487921B (en) | Face image preprocessing method and system for living body detection | |
| CN111401310B (en) | Kitchen sanitation safety supervision and management method based on artificial intelligence | |
| CN112396658A (en) | Indoor personnel positioning method and positioning system based on video | |
| CN111444555B (en) | Temperature measurement information display method and device and terminal equipment | |
| CN106881716A (en) | Human body follower method and system based on 3D cameras robot | |
| US20190064311A1 (en) | Active person positioning device and activity data acquisition device | |
| CN110557628A (en) | Method and device for detecting shielding of camera and electronic equipment | |
| CN111382592A (en) | Living body detection method and apparatus | |
| CN111243229A (en) | Old people falling risk assessment method and system | |
| CN111104910A (en) | Method for monitoring garbage delivery behavior and related product | |
| CN109389040B (en) | Inspection method and device for safety dressing of personnel in operation field | |
| CN112507772A (en) | Face recognition security system and suspicious person detection and early warning method | |
| CN111597879A (en) | Gesture detection method, device and system based on monitoring video | |
| CN111325133A (en) | Image processing system based on artificial intelligence recognition | |
| CN106529500A (en) | Information processing method and system | |
| CN110738607A (en) | Method, device and equipment for shooting driving license based on artificial intelligence and storage medium | |
| WO2022057806A1 (en) | Background image self-updating method, apparatus and device, and storage medium | |
| CN112446254A (en) | Face tracking method and related device |
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 |
Application publication date: 20200908 |
|
| RJ01 | Rejection of invention patent application after publication |