CN112307916A - Alarm monitoring method based on visible light camera - Google Patents
Alarm monitoring method based on visible light camera Download PDFInfo
- Publication number
- CN112307916A CN112307916A CN202011131393.1A CN202011131393A CN112307916A CN 112307916 A CN112307916 A CN 112307916A CN 202011131393 A CN202011131393 A CN 202011131393A CN 112307916 A CN112307916 A CN 112307916A
- Authority
- CN
- China
- Prior art keywords
- network
- target
- cnn
- fast
- detection
- 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 33
- 238000012544 monitoring process Methods 0.000 title claims abstract description 24
- 238000001514 detection method Methods 0.000 claims abstract description 69
- 238000013527 convolutional neural network Methods 0.000 claims description 55
- 238000012549 training Methods 0.000 claims description 31
- 238000010586 diagram Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 230000036544 posture Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000008034 disappearance Effects 0.000 abstract description 3
- 238000004880 explosion Methods 0.000 abstract description 3
- 238000000605 extraction Methods 0.000 abstract description 3
- 230000015556 catabolic process Effects 0.000 abstract description 2
- 238000006731 degradation reaction Methods 0.000 abstract description 2
- 238000012360 testing method Methods 0.000 description 3
- 238000011410 subtraction method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- 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
Abstract
The invention discloses an alarm monitoring method based on a visible light camera, which adopts a target detection network based on fast R-CNN for detection, selects a ResNet-101 network to replace VGG as a feature extraction network, and adopts the ResNet-101 network consisting of a residual error model to solve the problems of gradient disappearance or explosion and network degradation caused by the increase of network depth. The characteristics extracted by the ResNet-101 network are more detailed, and the accuracy of target detection is improved.
Description
Technical Field
The invention relates to an alarm monitoring method, in particular to an alarm monitoring method based on a visible light camera, and belongs to the technical field of alarm monitoring.
Background
With the development of science and technology, monitoring systems are widely applied to various industries of society, but most of the existing monitoring systems need to manually check whether abnormal conditions occur in a monitored area, and the accuracy of manual checking is greatly reduced when a plurality of monitoring videos are watched for a long time. At present, some monitoring systems are used to check whether there is a suspicious person in the monitored area, and a target detection method can be adopted to solve the problem, and the current target detection methods include a background difference method, a frame difference method and the like. However, since a camera in a monitoring system may be in a non-stationary state, it may cause the background and color texture of the detected image to be very complex, and if a general identification method such as a background subtraction method, a frame subtraction method, etc. is adopted, the detection result will have problems of false detection, missed detection, etc. due to the disadvantage of poor effectiveness of background modeling in the algorithm.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide an alarm monitoring method based on a visible light camera, which can improve the accuracy of alarm monitoring.
In order to solve the technical problem, the technical scheme adopted by the invention is as follows: an alarm monitoring method based on a visible light camera comprises the following steps:
s01), after the detection instruction of the user is obtained, or within the preset detection time, the system starts to detect;
s02), carrying out screenshot on the video of the visible light camera to obtain an original image;
s03), preprocessing and size transformation are carried out on the image acquired in the step S02;
s04), carrying out target detection on the image obtained in the step S03 by using a target detection network based on Faster R-CNN, and detecting whether a suspicious target meeting the characteristics exists in the image;
s05), judging the detection result of the step S04, and returning to the step S01 if no suspicious target appears; if the suspicious target appears, reminding the client, giving an alarm, displaying the detection image generated in the step S04 to the client, and waiting for the feedback result of the client;
s06), performing different processing according to the feedback result obtained in the step S5, if the client confirms that the target is a suspicious target, storing the detection picture generated in the step S04, and if the user feeds back that the target is not the suspicious target, deleting the images in the steps S02, S03 and S04, so as to reduce the occupation of the memory space;
s07), judging whether the detection ending time is reached or whether a detection stopping command of the user is received, if so, stopping the detection, otherwise, returning to the step S02.
Further, the target detection network based on Fast R-CNN in step S04 includes a Fast R-CNN network and an RPN network, the RPN network samples random region information of the image as a suggested region and trains regions that may contain targets, the Fast R-CNN network further processes the region information collected by the RPN network, determines a target category in the region, adjusts the size of the region, and locates a specific position of the target in the image.
Further, the process of target detection by the target detection network based on the Faster R-CNN is as follows:
s41), inputting pictures with fixed size into the Fast R-CNN network, and entering a shared convolution layer to extract a characteristic diagram;
s42), transmitting the characteristic diagram into an RPN network, predicting the position of a window containing a target through a convolution layer based on sliding operation, performing convolution operation twice by the convolution layer, judging whether the area where the sliding window is located belongs to the target or not, if so, reserving a suggested area and inputting the suggested area into a Fast R-CNN network; discarding the suggested region if the information in the widget is identified as background; the other convolution operation is used for calculating the offset position of the suggestion region and obtaining the position information of the suggestion region in the actual picture;
s43), in the Fast R-CNN network, inputting the feature map and the suggested area output by the RPN network into a pooling layer to generate an interested area, pooling the interested area into a vector with a fixed length, and performing target classification prediction and bounding box regression prediction as the input of a full-connection layer.
Further, target detection is carried out by using a target detection network based on the Faster R-CNN, a picture database containing various human body postures is established, human body targets which only face a camera and are clear are marked by using a marking boundary frame in the picture database, an image set in the picture database is divided into a sample set and a training set, the ratio of the sample set to the training set is 4:1, and the established picture database is adopted to train the target detection network.
Further, the process of training the target detection network is as follows:
A1) training the ResNet-101 network model by using data in the picture database to obtain an ImageNet model;
A2) initializing the RPN by using the ImageNet model generated in the step A1, training the RPN by using the ImageNet model in the convolution operation in the RPN network, and collecting a suggested region;
A3) initializing Fast R-CNN by using an ImageNet model, using the ImageNet model in the convolution operation in the Fast R-CNN network, and training the Fast R-CNN network by using the suggested region generated in the step A2;
A4) fixing the convolutional layer after the Fast R-CNN network training, and carrying out secondary training on the RPN network;
AA 5), fixing the shared convolution layer of RPN and Faster R-CNN, fine-tuning Fast R-CNN using the proposed region generated in the fourth step.
Further, the ImageNet model consists of a residual model.
Further, in the training of the target detection network, training ResNet, taking the first four convolutional neural networks in the ResNet-101 network as initialization parameters of a Fast R-CNN and RPN shared convolutional layer, and extracting image characteristics; in Fast R-CNN, the last stage of the ResNet-101 network is used as an initialization parameter to detect the network.
Further, in step S02, a capture time interval is set, and the visible light camera is captured according to the capture time interval.
The invention has the beneficial effects that: the alarm monitoring method based on the visible light camera adopts a target detection network based on fast R-CNN for detection, selects ResNet-101 network to replace VGG as a feature extraction network, and adopts the ResNet-101 network consisting of a residual error model to solve the problems of gradient disappearance or explosion and network degradation caused by the increase of network depth. The characteristics extracted by the ResNet-101 network are more detailed, and the accuracy of target detection is improved.
Drawings
FIG. 1 is a flow chart of the present method;
fig. 2 is a flow chart of target detection.
Detailed Description
The invention is further described with reference to the following figures and specific embodiments.
Example 1
The embodiment discloses an alarm monitoring method based on a visible light camera, as shown in fig. 1, comprising the following steps:
s01), after the detection instruction of the user is obtained, or within the preset detection time, the system starts to detect;
according to step S01, the method may set a detection time in advance, and when the detection start time is reached, the system may detect whether there is a suspicious person in the monitored area, and when the detection end time is reached, the system stops detecting. In addition, the user can manually turn on the test, and when the test is turned on, the system will turn on the test immediately.
S02), carrying out screenshot on the video of the visible light camera to obtain an original image;
in step S02, a capture time interval is set, and a screen shot is taken for the visible light camera according to the capture time interval. In this embodiment, the algorithm time of Faster r-cnn is set to T, and the acquisition time interval is set to 3/2T.
S03), preprocessing and size transformation are carried out on the image acquired in the step S02;
in this embodiment, the preprocessing is to remove apparent noise in the picture;
the size transformation is to convert the preprocessed pictures into a uniform size.
S04), carrying out target detection on the image obtained in the step S03 by using a target detection network based on Faster R-CNN, and detecting whether a suspicious target meeting the characteristics exists in the image;
s05), judging the detection result of the step S04, and returning to the step S01 if no suspicious target appears; if the suspicious target appears, reminding the client, giving an alarm, displaying the detection image generated in the step S04 to the client, and waiting for the feedback result of the client;
s06), performing different processing according to the feedback result obtained in the step S05, if the client confirms that the target is a suspicious target, storing the detection picture generated in the step S04, and if the user feeds back that the target is not the suspicious target, deleting the images in the steps S02, S03 and S04, so as to reduce the occupation of the memory space;
s07), judging whether the detection ending time is reached or whether a detection stopping command of the user is received, if so, stopping the detection, otherwise, returning to the step S02.
In this embodiment, the target detection network based on Fast R-CNN in step S04 includes Fast R-CNN network and RPN network, where the RPN network samples random region information of the image as a suggested region and trains regions where they may contain targets, and the Fast R-CNN network further processes the region information acquired by the RPN network to determine the type of the target in the region, adjust the size of the region, and locate the specific position of the target in the image.
As shown in FIG. 2, the process of target detection by the target detection network based on the Faster R-CNN is as follows:
s41), inputting pictures with fixed size into the Fast R-CNN network, and entering a shared convolution layer to extract a characteristic diagram;
s42), transmitting the characteristic diagram into an RPN network, predicting the position of a window containing a target through a convolution layer based on sliding operation, performing convolution operation twice by the convolution layer, judging whether the area where the sliding window is located belongs to the target or not, if so, reserving a suggested area and inputting the suggested area into a Fast R-CNN network; discarding the suggested region if the information in the widget is identified as background; the other convolution operation is used for calculating the offset position of the suggestion region and obtaining the position information of the suggestion region in the actual picture;
in this embodiment, the sliding operation is to select 3 × 3 regions on the feature map as input of the convolution layer in sequence by sliding a window (the size of the window is 3 × 3), so as to perform feature extraction on the 3 × 3 regions on the feature map with each pixel point as the center.
S43), in the Fast R-CNN network, inputting the feature map and the suggested area output by the RPN network into a pooling layer to generate an interested area, pooling the interested area into a vector with a fixed length, and performing target classification prediction and bounding box regression prediction as the input of a full-connection layer.
In the embodiment, target detection is performed by using a target detection network based on fast R-CNN, a picture database containing various human body postures is established, human body targets which only face a camera and are clear are marked in the picture database by using a marking boundary frame, an image set in the picture database is divided into a sample set and a training set, the ratio of the sample set to the training set is 4:1, and the established picture database is adopted to train the target detection network.
The process of training the target detection network comprises the following steps:
A1) training the ResNet-101 network model by using data in the picture database to obtain an ImageNet model;
A2) initializing the RPN by using the ImageNet model generated in the step A1, training the RPN by using the ImageNet model in the convolution operation in the RPN network, and collecting a suggested region;
A3) initializing Fast R-CNN by using an ImageNet model, using the ImageNet model in the convolution operation in the Fast R-CNN network, and training the Fast R-CNN network by using the suggested region generated in the step A2;
A4) fixing the convolutional layer after the Fast R-CNN network training, and carrying out secondary training on the RPN network;
AA 5), fixing the shared convolution layer of RPN and Faster R-CNN, fine-tuning Fast R-CNN using the proposed region generated in the fourth step.
In this embodiment, the ImageNet model is composed of a residual error model. Because the depth of the network is an important factor for achieving good effects, the ResNet-101 with a 101-layer network has been selected as a feature selection network model, and as the depth of the network increases, the feature layer also increases correspondingly. However, the problem of gradient disappearance or gradient explosion caused by too large depth value also occurs, that is, the performance of the network becomes worse and worse, the number of layers increases, and the error rate significantly increases, and the ResNet model composed of the residual error model can better solve the above problem.
In the training of the target detection network, training ResNet, taking the first four convolutional neural networks in the ResNet-101 network as initialization parameters of a Fast R-CNN and RPN shared convolutional layer, and extracting image characteristics; in Fast R-CNN, the last stage of the ResNet-101 network is used as an initialization parameter to detect the network.
The foregoing description is only for the basic principle and the preferred embodiments of the present invention, and modifications and substitutions by those skilled in the art according to the present invention belong to the protection scope of the present invention.
Claims (8)
1. An alarm monitoring method based on a visible light camera is characterized in that: the method comprises the following steps:
s01), after the detection instruction of the user is obtained, or within the preset detection time, the system starts to detect;
s02), carrying out screenshot on the video of the visible light camera to obtain an original image;
s03), preprocessing and size transformation are carried out on the image acquired in the step S02;
s04), carrying out target detection on the image obtained in the step S03 by using a target detection network based on Faster R-CNN, and detecting whether a suspicious target meeting the characteristics exists in the image;
s05), judging the detection result of the step S04, and returning to the step S01 if no suspicious target appears; if the suspicious target appears, reminding the client, giving an alarm, displaying the detection image generated in the step S04 to the client, and waiting for the feedback result of the client;
s06), performing different processing according to the feedback result obtained in the step S5, if the client confirms that the target is a suspicious target, storing the detection picture generated in the step S04, and if the user feeds back that the target is not the suspicious target, deleting the images in the steps S02, S03 and S04, so as to reduce the occupation of the memory space;
s07), judging whether the detection ending time is reached or whether a detection stopping command of the user is received, if so, stopping the detection, otherwise, returning to the step S02.
2. The alarm monitoring method based on the visible light camera according to claim 1, characterized in that: the target detection network based on the Faster R-CNN in the step S04 comprises a Fast R-CNN network and an RPN network, wherein the RPN network samples random area information of the image as a suggested area and trains areas possibly containing the target, and the Fast R-CNN network further processes the area information acquired by the RPN network, determines the type of the target in the area, adjusts the size of the area and positions the specific position of the target in the image.
3. The alarm monitoring method based on the visible light camera according to claim 2, characterized in that: the process of target detection by the target detection network based on the Faster R-CNN is as follows:
s41), inputting pictures with fixed size into the Fast R-CNN network, and entering a shared convolution layer to extract a characteristic diagram;
s42), transmitting the characteristic diagram into an RPN network, predicting the position of a window containing a target through a convolution layer based on sliding operation, performing convolution operation twice by the convolution layer, judging whether the area where the sliding window is located belongs to the target or not, if so, reserving a suggested area and inputting the suggested area into a Fast R-CNN network; discarding the suggested region if the information in the widget is identified as background; the other convolution operation is used for calculating the offset position of the suggestion region and obtaining the position information of the suggestion region in the actual picture;
s43), in the Fast R-CNN network, inputting the feature map and the suggested area output by the RPN network into a pooling layer to generate an interested area, pooling the interested area into a vector with a fixed length, and performing target classification prediction and bounding box regression prediction as the input of a full-connection layer.
4. The alarm monitoring method based on the visible light camera according to claim 2, characterized in that: the method comprises the steps of carrying out target detection by using a target detection network based on fast R-CNN, establishing a picture database containing various human body postures, marking a human body target which only faces a camera and is clear by using a marking boundary box in the picture database, dividing an image set in the picture database into a sample set and a training set, wherein the ratio of the sample set to the training set is 4:1, and training the target detection network by using the established picture database.
5. The visible-light-camera-based alarm monitoring method according to claim 4, wherein: the process of training the target detection network comprises the following steps:
A1) training the ResNet-101 network model by using data in the picture database to obtain an ImageNet model;
A2) initializing the RPN by using the ImageNet model generated in the step A1, training the RPN by using the ImageNet model in the convolution operation in the RPN network, and collecting a suggested region;
A3) initializing Fast R-CNN by using an ImageNet model, using the ImageNet model in the convolution operation in the Fast R-CNN network, and training the Fast R-CNN network by using the suggested region generated in the step A2;
A4) fixing the convolutional layer after the Fast R-CNN network training, and carrying out secondary training on the RPN network;
AA 5), fixing the shared convolution layer of RPN and Faster R-CNN, fine-tuning Fast R-CNN using the proposed region generated in the fourth step.
6. The alarm monitoring method based on the visible light camera according to claim 5, wherein: the ImageNet model consists of a residual model.
7. The alarm monitoring method based on the visible light camera according to claim 5, wherein: in the training of the target detection network, training ResNet, taking the first four convolutional neural networks in the ResNet-101 network as initialization parameters of a Fast R-CNN and RPN shared convolutional layer, and extracting image characteristics; in Fast R-CNN, the last stage of the ResNet-101 network is used as an initialization parameter to detect the network.
8. The alarm monitoring method based on the visible light camera according to claim 1, characterized in that: in step S02, a capture time interval is set, and the visible light camera is captured according to the capture time interval.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011131393.1A CN112307916A (en) | 2020-10-21 | 2020-10-21 | Alarm monitoring method based on visible light camera |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011131393.1A CN112307916A (en) | 2020-10-21 | 2020-10-21 | Alarm monitoring method based on visible light camera |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112307916A true CN112307916A (en) | 2021-02-02 |
Family
ID=74328682
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011131393.1A Pending CN112307916A (en) | 2020-10-21 | 2020-10-21 | Alarm monitoring method based on visible light camera |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112307916A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113516040A (en) * | 2021-05-12 | 2021-10-19 | 山东浪潮科学研究院有限公司 | Method for improving two-stage target detection |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102164270A (en) * | 2011-01-24 | 2011-08-24 | 浙江工业大学 | Intelligent video monitoring method and system capable of exploring abnormal events |
CN106691389A (en) * | 2017-01-10 | 2017-05-24 | 胡佳 | Medical information monitoring method and system |
CN108275114A (en) * | 2018-02-27 | 2018-07-13 | 苏州清研微视电子科技有限公司 | A kind of Security for fuel tank monitoring system |
CN109285139A (en) * | 2018-07-23 | 2019-01-29 | 同济大学 | A kind of x-ray imaging weld inspection method based on deep learning |
-
2020
- 2020-10-21 CN CN202011131393.1A patent/CN112307916A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102164270A (en) * | 2011-01-24 | 2011-08-24 | 浙江工业大学 | Intelligent video monitoring method and system capable of exploring abnormal events |
CN106691389A (en) * | 2017-01-10 | 2017-05-24 | 胡佳 | Medical information monitoring method and system |
CN108275114A (en) * | 2018-02-27 | 2018-07-13 | 苏州清研微视电子科技有限公司 | A kind of Security for fuel tank monitoring system |
CN109285139A (en) * | 2018-07-23 | 2019-01-29 | 同济大学 | A kind of x-ray imaging weld inspection method based on deep learning |
Non-Patent Citations (1)
Title |
---|
李明洁: "基于图像分析的输电线路防外力破坏技术研究", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅱ辑》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113516040A (en) * | 2021-05-12 | 2021-10-19 | 山东浪潮科学研究院有限公司 | Method for improving two-stage target detection |
CN113516040B (en) * | 2021-05-12 | 2023-06-20 | 山东浪潮科学研究院有限公司 | Method for improving two-stage target detection |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP3123587B2 (en) | Moving object region extraction method using background subtraction | |
CN110044486B (en) | Method, device and equipment for avoiding repeated alarm of human body inspection and quarantine system | |
CN105930822A (en) | Human face snapshot method and system | |
CN112734731B (en) | Livestock temperature detection method, device, equipment and storage medium | |
US11468683B2 (en) | Population density determination from multi-camera sourced imagery | |
CN111723773B (en) | Method and device for detecting carryover, electronic equipment and readable storage medium | |
CN111753794B (en) | Fruit quality classification method, device, electronic equipment and readable storage medium | |
CN110569770A (en) | Human body intrusion behavior recognition method and device, storage medium and electronic equipment | |
CN111415339A (en) | Image defect detection method for complex texture industrial product | |
CN114241370A (en) | Intrusion identification method and device based on digital twin transformer substation and computer equipment | |
CN116402852A (en) | Dynamic high-speed target tracking method and device based on event camera | |
CN111263955A (en) | Method and device for determining movement track of target object | |
CN113065454B (en) | High-altitude parabolic target identification and comparison method and device | |
CN111178405A (en) | Similar object identification method fusing multiple neural networks | |
CN112307916A (en) | Alarm monitoring method based on visible light camera | |
CN113869110A (en) | Article detection method, device, terminal and computer readable storage medium | |
CN115083008A (en) | Moving object detection method, device, equipment and storage medium | |
CN116485779B (en) | Adaptive wafer defect detection method and device, electronic equipment and storage medium | |
CN109871456B (en) | Method and device for analyzing relationship between watchmen and electronic equipment | |
CN113689585B (en) | Non-inductive attendance card punching method, system and related equipment | |
CN116229336A (en) | Video moving target identification method, system, storage medium and computer | |
US20190102888A1 (en) | Image processing apparatus and method and monitoring system | |
CN107403192B (en) | Multi-classifier-based rapid target detection method and system | |
CN109859200B (en) | Low-altitude slow-speed unmanned aerial vehicle rapid detection method based on background analysis | |
KR102589150B1 (en) | Long-distance object detection system using cumulative difference image |
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: 20210202 |
|
RJ01 | Rejection of invention patent application after publication |