CN116310985A - Video stream data-based abnormal data intelligent identification method, device and equipment - Google Patents
Video stream data-based abnormal data intelligent identification method, device and equipment Download PDFInfo
- Publication number
- CN116310985A CN116310985A CN202310267084.4A CN202310267084A CN116310985A CN 116310985 A CN116310985 A CN 116310985A CN 202310267084 A CN202310267084 A CN 202310267084A CN 116310985 A CN116310985 A CN 116310985A
- Authority
- CN
- China
- Prior art keywords
- video
- feature
- abnormal
- features
- video stream
- 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
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 203
- 238000000034 method Methods 0.000 title claims abstract description 58
- 230000011218 segmentation Effects 0.000 claims abstract description 62
- 230000004927 fusion Effects 0.000 claims description 59
- 238000000605 extraction Methods 0.000 claims description 23
- 238000010606 normalization Methods 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 11
- 238000004088 simulation Methods 0.000 claims description 5
- 230000002547 anomalous effect Effects 0.000 claims 2
- 230000008569 process Effects 0.000 description 17
- 230000008859 change Effects 0.000 description 12
- 238000004891 communication Methods 0.000 description 8
- 238000007726 management method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005065 mining Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 206010000117 Abnormal behaviour Diseases 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 241000283070 Equus zebra Species 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000000007 visual effect 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/44—Event detection
-
- 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
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Computational Linguistics (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the field of image classification, and discloses an abnormal data intelligent identification method, device and equipment based on video stream data, wherein the method comprises the following steps: acquiring an interval video stream and a continuous video stream in video stream data, and extracting features of the interval video stream and the continuous video stream to obtain interval video features and continuous video features; splicing the interval video features and the continuous video features, and carrying out feature updating on the spliced video features; merging the continuous video features and the updated video features, and analyzing analog video frames of the video stream data; inquiring a real video frame corresponding to the analog video frame, calculating an abnormal data index between the analog video frame and the real video frame, and identifying an abnormal video frame of the video stream data; and carrying out grid segmentation on the abnormal video frames, identifying abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid splicing on the abnormal segmentation grids, and identifying abnormal data of the video stream data. The invention can improve the identification accuracy of the abnormal data of the video stream data.
Description
Technical Field
The present invention relates to the field of image classification, and in particular, to a method, an apparatus, and a device for intelligently identifying abnormal data based on video stream data.
Background
The goal of the anomaly data identification of video stream data is to find unusual or unexpected video frames in the video stream data.
At present, abnormal behavior characteristics in video stream data are usually identified by using a neural network model, based on the abnormal behavior characteristics, the predicted abnormal class probability is output, and the abnormal behavior corresponding to the abnormal class probability is identified; the definition standard of the abnormal data and the definition standard of the normal data are not clear, so that the distinction degree between the abnormal data and the normal data is not high; even if an abnormal frame of the video stream data can be identified, the position in the abnormal frame where the specific abnormal data appears cannot be identified. Therefore, since it is difficult to identify an abnormal frame containing all abnormal data, the degree of distinction between the abnormal data and the normal data is not high, and the occurrence of the abnormal data in the abnormal frame is not clear, the abnormal data identification accuracy of the video stream data is low.
Disclosure of Invention
In order to solve the problems, the invention provides an abnormal data intelligent identification method, device and equipment based on video stream data, which can improve the abnormal data identification accuracy of the video stream data by identifying normal frames containing all normal data, increasing the distinction between the abnormal data and the normal data and defining the occurrence condition of the abnormal data in the abnormal frames.
In a first aspect, the present invention provides a method for intelligently identifying abnormal data based on video stream data, including:
acquiring video stream data, acquiring interval video streams and continuous video streams in the video stream data, and extracting features of the interval video streams and the continuous video streams to obtain interval video features and continuous video features;
splicing the interval video features and the continuous video features to obtain spliced video features, and performing feature update on the spliced video features to obtain updated video features;
fusing the continuous video features and the updated video features to obtain fused video features, and analyzing analog video frames of the video stream data based on the interval video features and the fused video features;
Inquiring a real video frame corresponding to the analog video frame from the video stream data, calculating an abnormal data index between the analog video frame and the real video frame, and identifying the abnormal video frame in the video stream data according to the abnormal data index;
and carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid stitching on the abnormal segmentation grids to obtain stitched abnormal grids, and identifying abnormal data of the video stream data based on the stitched abnormal grids.
In a possible implementation manner of the first aspect, the collecting the interval video stream and the continuous video stream in the video stream data includes:
setting the acquisition interval and the acquisition length of the video stream data;
based on the acquisition interval, acquiring video stream data with the number of video frames conforming to the acquisition length from the video stream data at intervals to obtain the interval video stream;
and continuously acquiring video frames of the video stream data to obtain the continuous video stream.
In a possible implementation manner of the first aspect, the performing feature extraction on the interval video stream and the continuous video stream to obtain interval video features and continuous video features includes:
Extracting initial features of the interval video stream by using the following formula to obtain initial interval features:
wherein ,yBN Representing the initial spacing feature, x i Represents the ith pixel point in the interval video frame, b represents the deviation of the convolution layer, w i The weight of the convolution layer is represented, i represents the serial number of the pixel points in the interval video frame, N represents the total number of the pixel points in the interval video frame, mu BN Representing the mean, σ, of a normalized network layer, i.e., layer Batch Normalization BN The standard deviation of the normalized network layer, namely Batch Normalization layer, the epsilon represents constant parameters of the normalized network layer, namely Batch Normalization layer, gamma is taken as the weight of the normalized network layer, namely Batch Normalization layer, and beta is taken as the bias of the normalized network layer, namely Batch Normalization layer;
performing truncation operation on the initial interval feature to obtain the interval video feature;
extracting initial characteristics of the continuous video stream to obtain initial continuous characteristics;
and carrying out convolution operation on the initial continuous feature by using the following formula to obtain the continuous video feature:
wherein Y represents the continuous video feature, X j Represents the j-th pixel point in the continuous video stream frame, B represents the deviation of the convolution layer, W j The weight of the convolution layer is represented, j represents the sequence number of the pixel points in the continuous video stream frame, and M represents the total number of the pixel points in the continuous video stream frame.
In a possible implementation manner of the first aspect, the performing feature update on the stitched video feature to obtain an updated video feature includes:
performing feature expansion on the spliced video features to obtain expansion features;
calculating the feature similarity between the unfolding feature and a feature sample in a preset feature sample library by using the following formula:
wherein ,
representing the feature similarity->
Representing the stitched videoFeature X l Corresponding p-th expansion feature, < >>
Representing the Q-th characteristic sample in the preset characteristic sample library, wherein Q represents the total number of the characteristic samples in the preset characteristic sample library, and exp represents an exponential function;
based on the feature similarity and the preset feature sample library, the feature update is performed on the unfolding feature by using the following formula to obtain the updated video feature:
wherein ,
representing the updated video features->
Representation->
Corresponding feature similarity, < >>
Representing the q-th feature sample in the preset feature sample library,/th feature sample in the preset feature sample library>
Representing the stitched video feature X l Corresponding p-th expansion feature.
In a possible implementation manner of the first aspect, the fusing the continuous video feature and the updated video feature to obtain a fused video feature includes:
performing first feature extraction and truncation on the continuous video features and the updated video features to obtain a continuous truncation result and an updated truncation result;
performing first feature fusion on the continuous cutoff result and the updated cutoff result to obtain a first fusion result;
performing second feature extraction and truncation on the first fusion result and the continuous truncation result to obtain a fusion truncation result and a second truncation result;
fusing the fusion cut-off result and the second cut-off result to obtain a second fusion result;
performing convolution operation on the second fusion result to obtain a convolution fusion result;
fusing the convolution fusion result and the second cut-off result to obtain a third fusion result;
splicing the third fusion result and the convolution fusion result to obtain a splicing result;
and upsampling the spliced result to obtain the fusion video feature.
In a possible implementation manner of the first aspect, the analyzing the analog video frame of the video stream data based on the interval video feature and the fusion video feature includes:
Extracting an updating cut-off result and a fusion cut-off result corresponding to the fusion video features;
feature combination is carried out on the fusion cut-off result and the fusion video feature to obtain a first combination result;
performing feature combination on the first combination result and the updating cut-off result to obtain a second combination result;
and performing feature combination on the second combination result and the interval video features to obtain the analog video frame.
In a possible implementation manner of the first aspect, the querying, from the video stream data, a real video frame corresponding to the analog video frame includes:
identifying an input video frame in the video stream data corresponding to the analog video frame;
and inquiring the next video frame of the input video frame in the video stream data to obtain the real video frame.
In a possible implementation manner of the first aspect, the calculating an abnormal data index between the analog video frame and the real video frame includes:
calculating the similarity of the video frames between the analog video frame and the real video frame by using the following formula:
wherein S (D' t ,D t ) Representing the similarity of the video frames, D' t Analog video frame representing time t, D t The real video frame at the t moment is represented, and N represents the total number of pixel points in the video frame;
based on the video frame similarity, calculating an anomaly data index between the simulated video frame and the real video frame using the following formula:
wherein P (t represents the abnormal data index, S (D' t ,D t ) Representing the similarity of the video frames, D' t Analog video frame representing time t, D t Representing the real video frame at time t.
In a possible implementation manner of the first aspect, the identifying, according to the anomaly data index, an anomaly video frame in the video stream data includes:
and when the abnormal data index is larger than a preset abnormal threshold, taking the real video frame corresponding to the abnormal data index as the abnormal video frame.
In one possible implementation manner of the first aspect, the performing an anomaly mesh stitching on the anomaly segmented mesh to obtain a stitched anomaly mesh includes:
inquiring a frame segmentation grid and an abnormal video frame corresponding to the abnormal segmentation grid to obtain a target frame segmentation grid;
extracting pixel coordinates of the target frame segmentation grid in the abnormal video frame;
And based on the pixel coordinates, performing abnormal grid stitching on the abnormal segmentation grid to obtain the stitched abnormal grid.
In a second aspect, the present invention provides an abnormal data intelligent recognition device based on video stream data, the device comprising:
the feature extraction module is used for acquiring video stream data, acquiring interval video streams and continuous video streams in the video stream data, and carrying out feature extraction on the interval video streams and the continuous video streams to obtain interval video features and continuous video features;
the feature updating module is used for splicing the interval video features and the continuous video features to obtain spliced video features, and carrying out feature updating on the spliced video features to obtain updated video features;
the simulation analysis module is used for fusing the continuous video features and the updated video features to obtain fused video features, and analyzing simulation video frames of the video stream data based on the interval video features and the fused video features;
the abnormal identification module is used for inquiring the real video frames corresponding to the analog video frames from the video stream data, calculating abnormal data indexes between the analog video frames and the real video frames, and identifying the abnormal video frames in the video stream data according to the abnormal data indexes;
The data identification module is used for carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid splicing on the abnormal segmentation grids to obtain spliced abnormal grids, and identifying abnormal data of the video stream data based on the spliced abnormal grids.
In a third aspect, the present invention provides an electronic device comprising:
at least one processor; and a memory communicatively coupled to the at least one processor;
wherein the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the video stream data-based anomaly data intelligent recognition method of any one of the first aspects above.
In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the video stream data-based anomaly data intelligent recognition method according to any one of the first aspects.
Compared with the prior art, the technical principle and beneficial effect of this scheme lie in:
the embodiment of the invention firstly acquires the interval video stream and the continuous video stream in the video stream data for carrying out the subsequent separate feature extraction of the interval video stream and the continuous video stream, acquires the detail features of different angles from the interval video stream and enriches the information extracted from the video stream data, and further, the embodiment of the invention extracts more motion change information from the interval video stream and the continuous video stream by carrying out the feature extraction of the interval video stream and the continuous video stream, because the space transformation of the normal data in the interval video stream and the continuous video stream is not obvious, is usually static, the change of the abnormal data is obvious and is in a motion state, the motion change information in the interval video stream and the continuous video stream needs to be extracted, the embodiment of the invention extracts multi-angle rich features containing the interval video features and the continuous video features from the spliced features by splicing the interval video features and the continuous video features, further obtains updated video features by carrying out feature updating on the spliced video features, is used for giving different weights to the extracted features, carries out feature updating on the spliced video features by selecting features from a feature sample library attached to a model based on weight magnitude, aims at converting the spliced video features into features with high feature similarity with the feature sample library attached to the model, converts features which are not concentrated widely originally into some partial features in a concentrated way, highlights the spliced video feature category, in order to mine from the characteristic of splicing to the multi-angle and abundant characteristic comprising the interval video characteristic and the continuous video characteristic to fuse, further, the embodiment of the invention analyzes the analog video frame of the video stream data based on the interval video characteristic and the fused video characteristic to be used for predicting the next frame data of the video stream data, the predicted next frame data is generally normal data, the normal data is convenient to compare with the abnormal data which may appear at the moment, so as to inquire the real abnormal data in the video stream data. Therefore, the method, the device and the equipment for intelligently identifying the abnormal data based on the video stream data can improve the accuracy of identifying the abnormal data of the video stream data by identifying the normal frames containing all the normal data, increasing the distinction between the abnormal data and the normal data and determining the occurrence condition of the abnormal data in the abnormal frames.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, and it will be obvious to a person skilled in the art that other drawings can be obtained from these drawings without inventive effort.
Fig. 1 is a flow chart of an intelligent recognition method for abnormal data based on video stream data according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating one of the steps of the method for intelligently identifying abnormal data based on video stream data according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating another step of the method for intelligently identifying abnormal data based on video stream data according to the embodiment of the present invention;
fig. 4 is a schematic block diagram of an abnormal data intelligent recognition device based on video stream data according to an embodiment of the present invention;
Fig. 5 is a schematic diagram of an internal structure of an electronic device for implementing an abnormal data intelligent recognition method based on video stream data according to an embodiment of the present invention.
Detailed Description
It should be understood that the detailed description is presented by way of example only and is not intended to limit the invention.
The embodiment of the invention provides an abnormal data intelligent identification method based on video stream data, wherein an execution subject of the abnormal data intelligent identification method based on video stream data comprises, but is not limited to, at least one of a server, a terminal and the like which can be configured to execute the method provided by the embodiment of the invention. In other words, the method for intelligently identifying abnormal data based on video stream data can be executed by software or hardware installed in a terminal device or a server device, wherein the software can be a blockchain platform. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like. The server may be an independent server, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.
Referring to fig. 1, a flow chart of an intelligent recognition method for abnormal data based on video stream data according to an embodiment of the invention is shown. The method for intelligently identifying abnormal data based on video stream data, which is described in fig. 1, comprises the following steps:
s1, acquiring video stream data, acquiring an interval video stream and a continuous video stream in the video stream data, and extracting features of the interval video stream and the continuous video stream to obtain interval video features and continuous video features.
In the embodiment of the invention, the video stream data consists of continuous video frames, and is applied to different scenes and contains different video contents, for example, in a road monitoring video scene, the video stream data comprises walking of people, vehicles, riding of people, the number of flowers and plants, animals and the like; in a campus monitoring video scene, the video stream data comprise data such as student study, teacher teaching, sanitation cleaning by sanitation workers, playground running, dining room dining and the like; if in a live video scene, the video stream data comprise data such as a main broadcasting dialogue exchange, a main broadcasting dancing, a vermicelli and a main broadcasting interaction, etc., it is to be noted that the video stream data in each scene comprise normal data and abnormal data, if in a road monitoring video scene, the abnormal data in the video stream data comprise that a vehicle stops on a zebra crossing, a bicycle walks on a motor vehicle lane, etc.; in a campus monitoring video scene, abnormal data in the video stream data comprise students running in a corridor and the like; such as in live video scenes, the video stream data is abnormal.
Further, the embodiment of the invention acquires the detail features of different angles from the interval video stream and the continuous video stream in the video stream data by acquiring the interval video stream and the continuous video stream, which are used for carrying out subsequent separation feature extraction on the interval video stream and the continuous video stream, so as to enrich the information extracted from the video stream data.
In an embodiment of the present invention, referring to fig. 2, the capturing the interval video stream and the continuous video stream in the video stream data includes:
s201, setting the acquisition interval and the acquisition length of the video stream data;
s202, based on the acquisition interval, acquiring video stream data with the number of video frames conforming to the acquisition length from the video stream data at intervals to obtain the interval video stream;
s203, continuously acquiring video frames of the video stream data to obtain the continuous video stream.
Further, in the embodiment of the present invention, by performing feature extraction on the interval video stream and the continuous video stream, more motion change information is extracted from the interval video stream and the continuous video stream, because spatial transformation of normal data in the interval video stream and the continuous video stream is not obvious, usually static, but change of abnormal data is obvious, and is in a motion state, the motion change information in the interval video stream and the continuous video stream needs to be extracted. The static state corresponding to the normal data, such as a face, does not change obviously, and the dynamic state corresponding to the abnormal data, such as deformation generated when a vehicle goes out of a car accident, and limb change of a pedestrian, compared with normal walking, has huge change, such as running and the like.
In an embodiment of the present invention, the extracting features of the interval video stream and the continuous video stream to obtain interval video features and continuous video features includes: extracting initial features of the interval video stream by using the following formula to obtain initial interval features:
wherein ,yBN Representing the initial spacing feature, x i Represents the ith pixel point in the interval video frame, b represents the deviation of the convolution layer, w i The weight of the convolution layer is represented, i represents the serial number of the pixel points in the interval video frame, N represents the total number of the pixel points in the interval video frame, mu BN Representing the mean, σ, of a normalized network layer, i.e., layer Batch Normalization BN The standard deviation of the normalized network layer, namely Batch Normalization layer, the epsilon represents constant parameters of the normalized network layer, namely Batch Normalization layer, gamma is taken as the weight of the normalized network layer, namely Batch Normalization layer, and beta is taken as the bias of the normalized network layer, namely Batch Normalization layer;
performing truncation operation on the initial interval feature to obtain the interval video feature; extracting initial characteristics of the continuous video stream to obtain initial continuous characteristics; and carrying out convolution operation on the initial continuous feature by using the following formula to obtain the continuous video feature:
Wherein Y represents the continuous video feature, X j Represents the j-th pixel point in the continuous video stream frame, B represents the deviation of the convolution layer, W j The weight of the convolution layer is represented, j represents the sequence number of the pixel points in the continuous video stream frame, and M represents the total number of the pixel points in the continuous video stream frame.
Optionally, the process of performing a truncation operation on the initial interval feature to obtain the interval video feature may be implemented by a RELU truncation function, so as to map a pixel value range in the initial interval feature into a [0,1] range; the process of extracting the initial features of the continuous video stream to obtain initial continuous features is to extract the features of the continuous video stream by using a process similar to the above-mentioned process of extracting the initial features of the interval video stream to obtain initial interval features, and then map the extracted features in the range of [0,1] by using a process similar to the above-mentioned process of cutting the initial interval features to obtain the interval video features to obtain the initial continuous features.
S2, splicing the interval video features and the continuous video features to obtain spliced video features, and carrying out feature update on the spliced video features to obtain updated video features.
The embodiment of the invention is used for mining multi-angle rich features comprising the interval video features and the continuous video features from the spliced features by splicing the interval video features and the continuous video features.
Optionally, the process of splicing the interval video feature and the continuous video feature to obtain the spliced video feature is to splice feature vectors of the interval video feature and the continuous video feature to form a vector with a longer dimension.
Further, in the embodiment of the invention, the feature update is performed on the spliced video features to obtain updated video features, so that the extracted features are given different weights, and the features are selected from a feature sample library attached to a model based on the weight values to perform the feature update on the spliced video features, so that the spliced video features are converted into features with high feature similarity with the feature sample library attached to the model, the features which are originally widely and not concentrated are converted into some concentrated partial features, and the spliced video feature categories are highlighted.
In an embodiment of the present invention, the feature updating the spliced video feature to obtain an updated video feature includes: performing feature expansion on the spliced video features to obtain expansion features; calculating the feature similarity between the unfolding feature and a feature sample in a preset feature sample library by using the following formula:
wherein ,
representing the feature similarity->
Representing the stitched video feature X l Corresponding p-th expansion feature, < >>
Representing the Q-th characteristic sample in the preset characteristic sample library, wherein Q represents the total number of the characteristic samples in the preset characteristic sample library, and exp represents an exponential function;
based on the feature similarity and the preset feature sample library, the feature update is performed on the unfolding feature by using the following formula to obtain the updated video feature:
wherein ,
representing the updated video features->
Representation->
Corresponding feature similarity, < >>
Representing the q-th feature sample in the preset feature sample library,/th feature sample in the preset feature sample library>
Representing the stitched video feature X l Corresponding p-th expansion feature.
The preset feature sample library records normal features of all video stream data, and the normal features are opposite to the abnormal features.
S3, fusing the continuous video features and the updated video features to obtain fused video features, and analyzing analog video frames of the video stream data based on the interval video features and the fused video features.
The embodiment of the invention fuses the continuous video features and the updated video features by fusing the continuous video features and is used for mining multi-angle rich features containing the interval video features and the continuous video features from spliced features.
In an embodiment of the present invention, the fusing the continuous video feature and the updated video feature to obtain a fused video feature includes: performing first feature extraction and truncation on the continuous video features and the updated video features to obtain a continuous truncation result and an updated truncation result; performing first feature fusion on the continuous cutoff result and the updated cutoff result to obtain a first fusion result; performing second feature extraction and truncation on the first fusion result and the continuous truncation result to obtain a fusion truncation result and a second truncation result; fusing the fusion cut-off result and the second cut-off result to obtain a second fusion result; performing convolution operation on the second fusion result to obtain a convolution fusion result; fusing the convolution fusion result and the second cut-off result to obtain a third fusion result; splicing the third fusion result and the convolution fusion result to obtain a splicing result; and upsampling the spliced result to obtain the fusion video feature.
Optionally, the first feature extraction and cutoff are performed on the continuous video feature and the updated video feature to obtain a continuous cutoff result and an updated cutoff result, and the second feature extraction and cutoff are performed on the first fusion result and the continuous cutoff result, and the process of obtaining the fusion cutoff result and the second cutoff result is a process of passing the feature through a conv-BN-RELU layer; and the process of carrying out the feature updating on the spliced video features in the step S2 to obtain updated video features is similar to the principle of obtaining the updated video features, and is not further repeated here.
Further, according to the embodiment of the invention, the analog video frames of the video stream data are analyzed based on the interval video features and the fusion video features to be used for predicting the next frame data of the video stream data, and the predicted next frame data are generally normal data, so that the normal data can be conveniently compared with abnormal data possibly occurring later, and real abnormal data in the video stream data can be queried.
In an embodiment of the present invention, the analyzing the analog video frame of the video stream data based on the interval video feature and the fusion video feature includes: extracting an updating cut-off result and a fusion cut-off result corresponding to the fusion video features; feature combination is carried out on the fusion cut-off result and the fusion video feature to obtain a first combination result; performing feature combination on the first combination result and the updating cut-off result to obtain a second combination result; and performing feature combination on the second combination result and the interval video features to obtain the analog video frame.
The process of combining the fusion truncated result with the fusion video feature to obtain a first combined result, combining the first combined result with the updated truncated result to obtain a second combined result, and combining the second combined result with the interval video feature to obtain the analog video frame is as follows: and splicing the two input features, and passing the splicing result through a conv-BN-RELU layer process.
S4, inquiring a real video frame corresponding to the analog video frame from the video stream data, calculating an abnormal data index between the analog video frame and the real video frame, and identifying the abnormal video frame in the video stream data according to the abnormal data index.
In an embodiment of the present invention, referring to fig. 3, the querying, from the video stream data, the real video frame corresponding to the analog video frame includes:
s301, identifying an input video frame corresponding to the analog video frame in the video stream data;
s302, inquiring the next video frame of the input video frame in the video stream data to obtain the real video frame.
In an embodiment of the present invention, the calculating the abnormal data index between the analog video frame and the real video frame includes: calculating the similarity of the video frames between the analog video frame and the real video frame by using the following formula:
wherein S (D' t ,D t ) Representing the similarity of the video frames, D' t Analog view showing time tFrequency frame, D t The real video frame at the t moment is represented, and N represents the total number of pixel points in the video frame;
based on the video frame similarity, calculating an anomaly data index between the simulated video frame and the real video frame using the following formula:
Wherein P (t represents the abnormal data index, S (D' t ,D t ) Representing the similarity of the video frames, D' t Analog video frame representing time t, D t Representing the real video frame at time t.
In an embodiment of the present invention, the identifying the abnormal video frame in the video stream data according to the abnormal data index includes: and when the abnormal data index is larger than a preset abnormal threshold, taking the real video frame corresponding to the abnormal data index as the abnormal video frame.
S5, carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid stitching on the abnormal segmentation grids to obtain stitched abnormal grids, and identifying abnormal data of the video stream data based on the stitched abnormal grids.
In an embodiment of the present invention, the process of identifying the abnormal segmentation grid in the frame segmentation grid is similar to the principle of the above-mentioned S1-S3 of obtaining the analog video frame of the video stream data finally, and will not be further described herein.
According to the embodiment of the invention, the abnormal grid splicing is carried out on the abnormal segmentation grids, so that the smaller and incomplete grid contents are spliced into the images capable of completely displaying the abnormal data, and the clarity of the occurrence condition of the abnormal data in the abnormal frames is improved.
In an embodiment of the present invention, the performing an abnormal mesh stitching on the abnormal partition mesh to obtain a stitched abnormal mesh includes: inquiring a frame segmentation grid and an abnormal video frame corresponding to the abnormal segmentation grid to obtain a target frame segmentation grid; extracting pixel coordinates of the target frame segmentation grid in the abnormal video frame; and based on the pixel coordinates, performing abnormal grid stitching on the abnormal segmentation grid to obtain the stitched abnormal grid.
In the embodiment of the invention, the content in the spliced abnormal grid is used as the abnormal data.
It can be seen that, in the embodiment of the present invention, firstly, by collecting an interval video stream and a continuous video stream in the video stream data, for performing subsequent separate feature extraction on the interval video stream and the continuous video stream, and obtaining detailed features from different angles from the interval video stream and the continuous video stream, so as to enrich information extracted from the video stream data, further, by performing feature extraction on the interval video stream and the continuous video stream, for extracting more motion change information from the interval video stream and the continuous video stream, since spatial transformation of normal data in the interval video stream and the continuous video stream is not obvious, usually is static, but change of abnormal data is obvious, thus, motion change information in the interval video stream and the continuous video stream needs to be extracted, in the embodiment of the present invention, by stitching the interval video feature and the continuous video feature, for mining from the stitched feature to multi-angle and rich features containing the interval video feature and the continuous video feature, further, the embodiment of the present invention, by converting the updated feature from the interval video stream and the continuous video stream into a new feature, and by performing update on the video feature, the embodiment of the video feature is not significant, and by taking a feature update feature from a feature of the video feature library, the feature is obtained by stitching feature library, and a feature is more important, and a feature is obtained by stitching feature is more than a feature is obtained, and a feature is obtained by a feature is more than has a feature, and a feature is obtained, according to the embodiment of the invention, the continuous video features and the updated video features are fused to be used for mining from the spliced features to multi-angle rich features comprising the interval video features and the continuous video features, and further, the embodiment of the invention analyzes the analog video frame of the video stream data based on the interval video features and the fused video features to be used for predicting the next frame data of the video stream data, wherein the predicted next frame data is generally normal data, so that the follow-up normal data can be conveniently compared with the abnormal data which possibly occur, and real abnormal data in the video stream data can be queried. Therefore, the abnormal data intelligent identification method based on the video stream data can improve the abnormal data identification accuracy of the video stream data by identifying the normal frame containing all the normal data, increasing the distinction between the abnormal data and the normal data and determining the occurrence condition of the abnormal data in the abnormal frame.
Fig. 4 is a functional block diagram of the abnormal data intelligent recognition device based on video stream data according to the present invention.
The abnormal data intelligent recognition device 400 based on video stream data can be installed in electronic equipment. The video stream data based abnormal data intelligent recognition apparatus may include a feature extraction module 401, a feature update module 402, a simulation analysis module 403, an abnormal recognition module 404, and a data recognition module 405 according to the implemented functions. The module of the invention, which may also be referred to as a unit, refers to a series of computer program segments, which are stored in the memory of the electronic device, capable of being executed by the processor of the electronic device and of performing a fixed function.
In the embodiment of the present invention, the functions of each module/unit are as follows:
the feature extraction module 401 is configured to obtain video stream data, collect an interval video stream and a continuous video stream in the video stream data, and perform feature extraction on the interval video stream and the continuous video stream to obtain an interval video feature and a continuous video feature;
the feature updating module 402 is configured to splice the interval video feature and the continuous video feature to obtain a spliced video feature, and perform feature updating on the spliced video feature to obtain an updated video feature;
The analog analysis module 403 is configured to fuse the continuous video feature and the updated video feature to obtain a fused video feature, and analyze an analog video frame of the video stream data based on the interval video feature and the fused video feature;
the anomaly identification module 404 is configured to query the real video frame corresponding to the analog video frame from the video stream data, calculate an anomaly data index between the analog video frame and the real video frame, and identify an anomaly video frame in the video stream data according to the anomaly data index;
the data identifying module 405 is configured to perform grid segmentation on the abnormal video frame to obtain a frame segmentation grid, identify an abnormal segmentation grid in the frame segmentation grid, perform abnormal grid stitching on the abnormal segmentation grid to obtain a stitched abnormal grid, and identify abnormal data of the video stream data based on the stitched abnormal grid.
In detail, the modules in the video stream data based abnormal data intelligent recognition device 400 in the embodiment of the present invention use the same technical means as the video stream data based abnormal data intelligent recognition method described in fig. 1 to 3, and can generate the same technical effects, which are not described herein.
Fig. 5 is a schematic structural diagram of an electronic device for implementing the method for intelligently identifying abnormal data based on video stream data according to the present invention.
The electronic device may comprise a processor 50, a memory 51, a communication bus 52 and a communication interface 53, and may further comprise a computer program stored in the memory 51 and executable on the processor 50, such as an anomaly data intelligent recognition program based on video stream data.
The processor 50 may be formed by an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or may be formed by a plurality of integrated circuits packaged with the same function or different functions, including one or more central processing units (Central Processing unit, CPU), a microprocessor, a digital processing chip, a graphics processor, a combination of various control chips, and so on. The processor 50 is a Control Unit (Control Unit) of the electronic device, connects respective components of the entire electronic device using various interfaces and lines, executes or executes programs or modules stored in the memory 51 (for example, executes an abnormal data intelligent recognition program based on video stream data, etc.), and invokes data stored in the memory 51 to perform various functions of the electronic device and process data.
The memory 51 includes at least one type of readable storage medium including flash memory, a removable hard disk, a multimedia card, a card type memory (e.g., SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, etc. The memory 51 may in some embodiments be an internal storage unit of the electronic device, such as a mobile hard disk of the electronic device. The memory 51 may also be an external storage device of the electronic device in other embodiments, for example, a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like. Further, the memory 51 may also include both an internal storage unit and an external storage device of the electronic device. The memory 51 may be used not only for storing application software installed in an electronic device and various types of data, such as codes of a database-configured connection program, but also for temporarily storing data that has been output or is to be output.
The communication bus 52 may be a peripheral component interconnect standard (peripheral component interconnect, PCI) bus, or an extended industry standard architecture (extended industry standard architecture, EISA) bus, among others. The bus may be classified as an address bus, a data bus, a control bus, etc. The bus is arranged to enable a connection communication between the memory 51 and at least one processor 50 etc.
The communication interface 53 is used for communication between the electronic device 5 and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and/or a wireless interface (e.g., WI-FI interface, bluetooth interface, etc.), typically used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a Display (Display), an input unit such as a Keyboard (Keyboard), or alternatively a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch, or the like. The display may also be referred to as a display screen or display unit, as appropriate, for displaying information processed in the electronic device and for displaying a visual user interface.
Fig. 5 shows only an electronic device with components, and it will be understood by those skilled in the art that the structure shown in fig. 5 is not limiting of the electronic device and may include fewer or more components than shown, or may combine certain components, or a different arrangement of components.
For example, although not shown, the electronic device may further include a power source (such as a battery) for supplying power to the respective components, and the power source may be logically connected to the at least one processor 50 through a power management device, so that functions of charge management, discharge management, and power consumption management are implemented through the power management device. The power supply may also include one or more of any of a direct current or alternating current power supply, recharging device, power failure detection circuit, power converter or inverter, power status indicator, etc. The electronic device may further include various sensors, bluetooth modules, wi-Fi modules, etc., which are not described herein.
It should be understood that the embodiments described are for illustrative purposes only and are not limited in scope by this configuration.
The database-configured connection program stored in the memory 51 in the electronic device is a combination of a plurality of computer programs, which, when run in the processor 50, can implement:
acquiring video stream data, acquiring interval video streams and continuous video streams in the video stream data, and extracting features of the interval video streams and the continuous video streams to obtain interval video features and continuous video features;
Splicing the interval video features and the continuous video features to obtain spliced video features, and performing feature update on the spliced video features to obtain updated video features;
fusing the continuous video features and the updated video features to obtain fused video features, and analyzing analog video frames of the video stream data based on the interval video features and the fused video features;
inquiring a real video frame corresponding to the analog video frame from the video stream data, calculating an abnormal data index between the analog video frame and the real video frame, and identifying the abnormal video frame in the video stream data according to the abnormal data index;
and carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid stitching on the abnormal segmentation grids to obtain stitched abnormal grids, and identifying abnormal data of the video stream data based on the stitched abnormal grids.
In particular, the specific implementation method of the processor 50 on the computer program may refer to the description of the relevant steps in the corresponding embodiment of fig. 1, which is not repeated herein.
Further, the electronic device integrated modules/units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a non-volatile computer readable storage medium. The storage medium may be volatile or nonvolatile. For example, the computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM).
The present invention also provides a storage medium storing a computer program which, when executed by a processor of an electronic device, can implement:
acquiring video stream data, acquiring interval video streams and continuous video streams in the video stream data, and extracting features of the interval video streams and the continuous video streams to obtain interval video features and continuous video features;
splicing the interval video features and the continuous video features to obtain spliced video features, and performing feature update on the spliced video features to obtain updated video features;
Fusing the continuous video features and the updated video features to obtain fused video features, and analyzing analog video frames of the video stream data based on the interval video features and the fused video features;
inquiring a real video frame corresponding to the analog video frame from the video stream data, calculating an abnormal data index between the analog video frame and the real video frame, and identifying the abnormal video frame in the video stream data according to the abnormal data index;
and carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid stitching on the abnormal segmentation grids to obtain stitched abnormal grids, and identifying abnormal data of the video stream data based on the stitched abnormal grids.
In the several embodiments provided in the present invention, it should be understood that the disclosed apparatus, device and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical function division, and there may be other manners of division when actually implemented.
The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units can be realized in a form of hardware or a form of hardware and a form of software functional modules.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.
The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.
It should be noted that in this document, relational terms such as "first" and "second" and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (13)
1. An abnormal data intelligent identification method based on video stream data is characterized by comprising the following steps:
acquiring video stream data, acquiring interval video streams and continuous video streams in the video stream data, and extracting features of the interval video streams and the continuous video streams to obtain interval video features and continuous video features;
splicing the interval video features and the continuous video features to obtain spliced video features, and performing feature update on the spliced video features to obtain updated video features;
fusing the continuous video features and the updated video features to obtain fused video features, and analyzing analog video frames of the video stream data based on the interval video features and the fused video features;
inquiring a real video frame corresponding to the analog video frame from the video stream data, calculating an abnormal data index between the analog video frame and the real video frame, and identifying the abnormal video frame in the video stream data according to the abnormal data index;
and carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid stitching on the abnormal segmentation grids to obtain stitched abnormal grids, and identifying abnormal data of the video stream data based on the stitched abnormal grids.
2. The method of claim 1, wherein the capturing of the alternate video stream and the continuous video stream in the video stream data comprises:
setting the acquisition interval and the acquisition length of the video stream data;
based on the acquisition interval, acquiring video stream data with the number of video frames conforming to the acquisition length from the video stream data at intervals to obtain the interval video stream;
and continuously acquiring video frames of the video stream data to obtain the continuous video stream.
3. The method of claim 1, wherein the performing feature extraction on the interval video stream and the continuous video stream to obtain interval video features and continuous video features comprises:
extracting initial features of the interval video stream by using the following formula to obtain initial interval features:
wherein ,yBN Representing the initial spacing feature, x i Represents the ith pixel point in the interval video frame, b represents the deviation of the convolution layer, w i The weight of the convolution layer is represented, i represents the serial number of the pixel points in the interval video frame, N represents the total number of the pixel points in the interval video frame, mu BN Representing the mean, σ, of a normalized network layer, i.e., layer Batch Normalization BN The standard deviation of the normalized network layer, namely Batch Normalization layer, the epsilon represents constant parameters of the normalized network layer, namely Batch Normalization layer, gamma is taken as the weight of the normalized network layer, namely Batch Normalization layer, and beta is taken as the bias of the normalized network layer, namely Batch Normalization layer;
performing truncation operation on the initial interval feature to obtain the interval video feature;
extracting initial characteristics of the continuous video stream to obtain initial continuous characteristics;
and carrying out convolution operation on the initial continuous feature by using the following formula to obtain the continuous video feature:
wherein Y represents the continuous video feature, X j Represents the j-th pixel point in the continuous video stream frame, B represents the deviation of the convolution layer, W j The weight of the convolution layer is represented, j represents the sequence number of the pixel points in the continuous video stream frame, and M represents the total number of the pixel points in the continuous video stream frame.
4. The method of claim 1, wherein the feature updating the stitched video features to obtain updated video features comprises:
performing feature expansion on the spliced video features to obtain expansion features;
calculating the feature similarity between the unfolding feature and a feature sample in a preset feature sample library by using the following formula:
wherein ,
representing the feature similarity->
Representing the stitched video feature X l A corresponding p-th expansion feature,
representing the Q-th characteristic sample in the preset characteristic sample library, wherein Q represents the total number of the characteristic samples in the preset characteristic sample library, and exp represents an exponential function;
based on the feature similarity and the preset feature sample library, the feature update is performed on the unfolding feature by using the following formula to obtain the updated video feature:
wherein ,
representing the updated video features->
Representation->
Corresponding feature similarity, < >>
Representing the q-th feature sample in the preset feature sample library,/th feature sample in the preset feature sample library>
Representing the stitched video feature X l Corresponding p-th expansion feature.
5. The method of claim 1, wherein the fusing the continuous video feature with the updated video feature results in a fused video feature, comprising:
performing first feature extraction and truncation on the continuous video features and the updated video features to obtain a continuous truncation result and an updated truncation result;
performing first feature fusion on the continuous cutoff result and the updated cutoff result to obtain a first fusion result;
Performing second feature extraction and truncation on the first fusion result and the continuous truncation result to obtain a fusion truncation result and a second truncation result;
fusing the fusion cut-off result and the second cut-off result to obtain a second fusion result;
performing convolution operation on the second fusion result to obtain a convolution fusion result;
fusing the convolution fusion result and the second cut-off result to obtain a third fusion result;
splicing the third fusion result and the convolution fusion result to obtain a splicing result;
and upsampling the spliced result to obtain the fusion video feature.
6. The method of claim 1, wherein the analyzing the analog video frames of the video stream data based on the interval video features and the fusion video features comprises:
extracting an updating cut-off result and a fusion cut-off result corresponding to the fusion video features;
feature combination is carried out on the fusion cut-off result and the fusion video feature to obtain a first combination result;
performing feature combination on the first combination result and the updating cut-off result to obtain a second combination result;
and performing feature combination on the second combination result and the interval video features to obtain the analog video frame.
7. The method according to claim 1, wherein said querying real video frames corresponding to said analog video frames from said video stream data comprises:
identifying an input video frame in the video stream data corresponding to the analog video frame;
and inquiring the next video frame of the input video frame in the video stream data to obtain the real video frame.
8. The method of claim 1, wherein said calculating an anomaly data index between the analog video frame and the real video frame comprises:
calculating the similarity of the video frames between the analog video frame and the real video frame by using the following formula:
wherein S (D' t ,D t ) Representing the similarity of the video frames, D' t Analog video frame representing time t, D t The real video frame at the t moment is represented, and N represents the total number of pixel points in the video frame;
based on the video frame similarity, calculating an anomaly data index between the simulated video frame and the real video frame using the following formula:
wherein Pt represents the anomaly data index, S (D' t ,D t ) Representing the similarity of the video frames, D' t Analog video frame representing time t, D t Representing the real video frame at time t.
9. The method of claim 1, wherein said identifying an anomalous video frame in said video stream data based on said anomalous data index comprises:
and when the abnormal data index is larger than a preset abnormal threshold, taking the real video frame corresponding to the abnormal data index as the abnormal video frame.
10. The method of claim 1, wherein performing anomaly mesh stitching on the anomaly segmented mesh to obtain a stitched anomaly mesh comprises:
inquiring a frame segmentation grid and an abnormal video frame corresponding to the abnormal segmentation grid to obtain a target frame segmentation grid;
extracting pixel coordinates of the target frame segmentation grid in the abnormal video frame;
and based on the pixel coordinates, performing abnormal grid stitching on the abnormal segmentation grid to obtain the stitched abnormal grid.
11. An abnormal data intelligent recognition device based on video stream data, which is characterized by comprising:
the feature extraction module is used for acquiring video stream data, acquiring interval video streams and continuous video streams in the video stream data, and carrying out feature extraction on the interval video streams and the continuous video streams to obtain interval video features and continuous video features;
The feature updating module is used for splicing the interval video features and the continuous video features to obtain spliced video features, and carrying out feature updating on the spliced video features to obtain updated video features;
the simulation analysis module is used for fusing the continuous video features and the updated video features to obtain fused video features, and analyzing simulation video frames of the video stream data based on the interval video features and the fused video features;
the abnormal identification module is used for inquiring the real video frames corresponding to the analog video frames from the video stream data, calculating abnormal data indexes between the analog video frames and the real video frames, and identifying the abnormal video frames in the video stream data according to the abnormal data indexes;
the data identification module is used for carrying out grid segmentation on the abnormal video frames to obtain frame segmentation grids, identifying the abnormal segmentation grids in the frame segmentation grids, carrying out abnormal grid splicing on the abnormal segmentation grids to obtain spliced abnormal grids, and identifying abnormal data of the video stream data based on the spliced abnormal grids.
12. An electronic device, the electronic device comprising:
At least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the video stream data-based anomaly data intelligent recognition method according to any one of claims 1 to 10.
13. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the video stream data-based anomaly data intelligent recognition method according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310267084.4A CN116310985A (en) | 2023-03-14 | 2023-03-14 | Video stream data-based abnormal data intelligent identification method, device and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310267084.4A CN116310985A (en) | 2023-03-14 | 2023-03-14 | Video stream data-based abnormal data intelligent identification method, device and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116310985A true CN116310985A (en) | 2023-06-23 |
Family
ID=86793846
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310267084.4A Pending CN116310985A (en) | 2023-03-14 | 2023-03-14 | Video stream data-based abnormal data intelligent identification method, device and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116310985A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118196978A (en) * | 2024-05-16 | 2024-06-14 | 杭州高达软件***股份有限公司 | Intelligent monitoring system for spherical tank safety construction |
-
2023
- 2023-03-14 CN CN202310267084.4A patent/CN116310985A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118196978A (en) * | 2024-05-16 | 2024-06-14 | 杭州高达软件***股份有限公司 | Intelligent monitoring system for spherical tank safety construction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Arietta et al. | City forensics: Using visual elements to predict non-visual city attributes | |
| CN113095434B (en) | Target detection method and device, electronic equipment and storage medium | |
| CN110088778B (en) | System, method and computer readable medium for scenario memory of automation platform | |
| CN110428091A (en) | Risk Identification Method and relevant device based on data analysis | |
| CN112200081A (en) | Abnormal behavior identification method and device, electronic equipment and storage medium | |
| CN112132032B (en) | Traffic sign board detection method and device, electronic equipment and storage medium | |
| CN110929806B (en) | Picture processing method and device based on artificial intelligence and electronic equipment | |
| CN110782120A (en) | Method, system, equipment and medium for evaluating traffic flow model | |
| CN113382279A (en) | Live broadcast recommendation method, device, equipment, storage medium and computer program product | |
| CN111783712A (en) | Video processing method, device, equipment and medium | |
| CN115830399B (en) | Classification model training method, device, equipment, storage medium and program product | |
| CN115690615B (en) | Video stream-oriented deep learning target recognition method and system | |
| CN116310985A (en) | Video stream data-based abnormal data intelligent identification method, device and equipment | |
| CN113111838A (en) | Behavior recognition method and device, equipment and storage medium | |
| CN114943937A (en) | Pedestrian re-identification method and device, storage medium and electronic equipment | |
| CN114003672B (en) | Method, device, equipment and medium for processing road dynamic event | |
| Ding et al. | Mit-avt clustered driving scene dataset: Evaluating perception systems in real-world naturalistic driving scenarios | |
| CN113822128A (en) | Traffic element identification method, device, equipment and computer readable storage medium | |
| CN113190703A (en) | Intelligent retrieval method and device for video image, electronic equipment and storage medium | |
| CN115083229B (en) | Intelligent recognition and warning system of flight training equipment based on AI visual recognition | |
| CN113609956B (en) | Training method, recognition device, electronic equipment and storage medium | |
| CN113887544B (en) | Image processing method, image processing device, electronic equipment and computer readable storage medium | |
| CN112529116B (en) | Scene element fusion processing method, device and equipment and computer storage medium | |
| CN114511825A (en) | Method, device and equipment for detecting area occupation and storage medium | |
| CN113792801A (en) | Method, device and equipment for detecting dazzling degree of human face and storage medium |
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 |