CN116310601B - Ship behavior classification method based on AIS track diagram and camera diagram group - Google Patents
Ship behavior classification method based on AIS track diagram and camera diagram group Download PDFInfo
- Publication number
- CN116310601B CN116310601B CN202310564867.9A CN202310564867A CN116310601B CN 116310601 B CN116310601 B CN 116310601B CN 202310564867 A CN202310564867 A CN 202310564867A CN 116310601 B CN116310601 B CN 116310601B
- Authority
- CN
- China
- Prior art keywords
- ship
- scoring
- classification
- target
- preset
- 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.)
- Active
Links
- 238000010586 diagram Methods 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000012544 monitoring process Methods 0.000 claims abstract description 50
- 238000011156 evaluation Methods 0.000 claims abstract description 38
- 238000013210 evaluation model Methods 0.000 claims abstract description 24
- 238000001514 detection method Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims description 44
- 238000009826 distribution Methods 0.000 claims description 25
- 238000000605 extraction Methods 0.000 claims description 21
- 238000013527 convolutional neural network Methods 0.000 claims description 20
- 230000004927 fusion Effects 0.000 claims description 19
- 238000012502 risk assessment Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 16
- 230000007613 environmental effect Effects 0.000 claims description 13
- 238000013507 mapping Methods 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 8
- 230000001502 supplementing effect Effects 0.000 claims description 5
- 230000003213 activating effect Effects 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 4
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 230000006399 behavior Effects 0.000 description 33
- 239000013598 vector Substances 0.000 description 5
- 238000010606 normalization Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Classifications
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
- G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/532—Query formulation, e.g. graphical querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- 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
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Abstract
The application relates to the technical field of navigation safety, and provides a ship behavior classification method based on an AIS track diagram and a camera diagram set, wherein the method comprises the following steps: when the AIS equipment detects that the target ship enters the detection range, the camera is controlled to monitor the target ship; constructing a ship monitoring database; drawing a ship plane map based on the ship monitoring database to obtain the ship plane map; constructing an AIS track map of the ship based on AIS information of all ships; inputting the target ship picture and the ship AIS track picture into a state evaluation model to obtain ship state evaluation information; and classifying based on the ship state evaluation information, and determining the ship behavior classification. The problem that the collision early warning judgment accuracy rate of the ship is low by solely depending on AIS information of surrounding ships in a complex weather environment can be solved, and the accuracy rate of the collision early warning judgment of the ship in the complex environment can be improved.
Description
Technical Field
The application relates to the technical field of navigation safety, in particular to a ship behavior classification method based on an AIS track diagram and a camera diagram set.
Background
The automatic ship identification system is a navigation aid system applied to marine safety and communication between ships and shore, and can quickly coordinate mutual communication by acquiring static information of all ships on the nearby sea surface and take necessary avoidance actions, so that the navigation safety of the ships is effectively ensured. The existing ship collision pre-judging method is based on AIS information of surrounding ships to process and judge, influences of external factors such as weather, ship size and visibility are ignored, and particularly the problem of low collision pre-judging accuracy is easily caused in complex environments with low visibility such as foggy days, rainy days or nights, so that navigation risks of ships are increased.
In summary, in the prior art, the problem of low collision early warning judgment accuracy rate exists by solely depending on AIS information of surrounding ships in a complex weather environment.
Disclosure of Invention
Based on this, it is necessary to provide a ship behavior classification method based on the AIS trace map and the camera map set in order to solve the above-mentioned technical problems.
A ship behavior classification method based on AIS trajectory graph and camera graph group, the method comprising: installing AIS equipment and a camera together, wherein the longitude and latitude of the AIS equipment are kept consistent with those of the camera, and when the AIS equipment detects that a target ship enters a detection range, activating signals are synchronized to the camera to control the camera to monitor the target ship; recording monitoring information of all ships in a detection area according to preset interval time, and shooting a target ship picture through a camera, wherein the ship monitoring information comprises longitude and latitude coordinates of the ship, ship heading and ship speed information, and constructing a ship monitoring database; extracting parameters according to preset drawing rules based on the ship monitoring database, and drawing a ship plane graph of the extracted parameters according to the preset drawing rules to obtain a ship plane graph, wherein the ship plane graph comprises a target ship, and the drawing pixels of the target ship are different from those of other ships; when the requirements of the preset shooting quantity are met, summarizing all ship plane graphs according to a time sequence based on AIS information of all ships, and constructing ship AIS track graphs; inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model to obtain ship state evaluation information; classifying according to preset classification requirements based on the ship state evaluation information, and determining ship behavior classification.
A ship behavior classification system based on AIS trajectory and camera map sets, comprising:
the target ship monitoring module is used for installing AIS equipment and the camera together, wherein the longitude and latitude of the AIS equipment are kept consistent with those of the camera, and when the AIS equipment detects that a target ship enters a detection range, a signal is synchronously activated to the camera to control the camera to monitor the target ship;
the ship monitoring information recording module is used for recording monitoring information of all ships in the detection area according to preset interval time, and shooting a target ship picture through the camera, wherein the ship monitoring information comprises longitude and latitude coordinates of the ship, ship course and ship speed information, and a ship monitoring database is constructed;
the ship plane map obtaining module is used for extracting parameters according to preset drawing rules based on the ship monitoring database, and drawing the ship plane map of the extracted parameters according to the preset drawing rules to obtain a ship plane map, wherein the ship plane map comprises a target ship, and the drawing pixels of the target ship are different from those of other ships;
The ship AIS track map construction module is used for summarizing all ship plane maps according to a time sequence based on AIS information of all ships when the ship AIS track map construction module meets the requirement of the preset shooting quantity to construct a ship AIS track map;
the ship state evaluation information acquisition module is used for inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model to acquire ship state evaluation information;
and the ship behavior classification determining module is used for classifying according to preset classification requirements based on the ship state evaluation information and determining ship behavior classification.
According to the ship behavior classification method based on the AIS track map and the camera map group, the problem that collision early warning judgment accuracy is low by singly depending on AIS information of surrounding ships in a complex weather environment can be solved, monitoring information and target ship image information of all ships in a detection area are obtained according to preset interval time, and a ship monitoring database is built based on the ship monitoring information; according to the ship monitoring database, a ship two-dimensional plan is obtained by taking the longitude and latitude of the ship as a coordinate point, the course of the ship as a direction vector and the speed as a vector length; acquiring target ship pictures of a preset shooting number and AIS information of all ships, summarizing two-dimensional plane views of all ships according to a time sequence, and constructing a ship AIS track map; building a state evaluation model based on a convolutional neural network, performing supervision training on the state evaluation model through a plurality of groups of historical data to obtain a state evaluation model meeting the requirements, and then inputting target ship pictures and ship AIS track pictures with preset shooting numbers into the state evaluation model to obtain ship state evaluation information; and finally classifying according to the state evaluation information and preset classification requirements to obtain the safety level of the ship navigation state. The accuracy of the ship collision early warning judgment in the complex environment can be improved.
The foregoing description is only an overview of the technical solutions of the present application, and may be implemented according to the content of the specification in order to make the technical means of the present application more clearly understood, and in order to make the above-mentioned and other objects, features and advantages of the present application more clearly understood, the following detailed description of the present application will be given.
Drawings
Fig. 1 is a schematic flow chart of a ship behavior classification method based on an AIS trajectory graph and a camera graph set;
FIG. 2 is a schematic flow chart of a method for classifying ship behaviors based on AIS trace diagrams and camera diagram sets to obtain a state evaluation model;
FIG. 3 is a schematic flow chart of a component training dataset in a ship behavior classification method based on AIS trace maps and camera map sets;
fig. 4 is a schematic structural diagram of a ship behavior classification system based on AIS trace patterns and camera pattern sets.
Reference numerals illustrate: the ship behavior classification system comprises a target ship monitoring module 1, a ship monitoring information recording module 2, a ship plane diagram obtaining module 3, a ship AIS track diagram constructing module 4, a ship state evaluation information obtaining module 5 and a ship behavior classification determining module 6.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
As shown in fig. 1, the present application provides a ship behavior classification method based on AIS trajectory graph and camera graph group, including:
step S100: installing AIS equipment and a camera together, wherein the longitude and latitude of the AIS equipment are kept consistent with those of the camera, and when the AIS equipment detects that a target ship enters a detection range, activating signals are synchronized to the camera to control the camera to monitor the target ship;
specifically, firstly, AIS equipment of a target ship and a camera are installed together, wherein the longitude and latitude of the AIS equipment are kept consistent with that of the camera, the target ship is a ship to be subjected to ship collision early warning judgment, the AIS equipment is a device for installing a ship automatic identification system on the target ship, and the camera is used for photographing the target ship and has photographing functions in complex weather environments such as rainy days, foggy days, nights and the like. When the AIS equipment detects that the target ship enters the detection range, an equipment activation signal is generated, and the equipment activation signal is sent to a control module of the camera, so that the camera is in a working state and is controlled to monitor the target ship. Wherein the detection range can be customized by a person skilled in the art based on the actual environment, for example: the complex weather environment or the complex sea area can be set as the detection range, for example, when the target ship is in a complex weather environment with low visibility such as rainy days, foggy days, nighttime, etc., the target ship can be regarded as entering the detection range, or when the target ship is in a sea area with large stormy waves, the target ship can also be regarded as entering the detection range. And monitoring the target ship by activating the camera, so as to provide support for obtaining the image information of the target ship in the detection range in the next step.
Step S200: recording monitoring information of all ships in a detection area according to preset interval time, and shooting a target ship picture through a camera, wherein the ship monitoring information comprises longitude and latitude coordinates of the ship, ship heading and ship speed information, and constructing a ship monitoring database;
specifically, the interval time of the monitoring data record is preset, and the monitoring interval time can be set by a person skilled in the art in a user-defined manner, for example: 10 seconds. And acquiring operation data of all ships in the detection area through AIS equipment according to the preset interval time, and recording the operation data as the ship monitoring information, wherein the ship monitoring information comprises ship longitude and latitude coordinates, ship heading and ship speed information. And controlling the camera to shoot a target ship picture according to the preset interval time. And constructing a ship monitoring database based on the ship monitoring information, wherein the ship monitoring database is used for storing a plurality of groups of ship monitoring information and a plurality of target ship pictures and is arranged according to a time sequence.
Step S300: extracting parameters according to preset drawing rules based on the ship monitoring database, and drawing a ship plane graph of the extracted parameters according to the preset drawing rules to obtain a ship plane graph, wherein the ship plane graph comprises a target ship, and the drawing pixels of the target ship are different from those of other ships;
Specifically, parameter extraction is performed on the ship monitoring database according to a preset drawing rule, wherein the parameters comprise ship longitude and latitude coordinates, ship heading and ship speed, the preset drawing rule refers to taking the ship longitude and latitude coordinates as coordinate points, taking the ship heading direction as vector direction, taking the ship speed as vector length, and setting different drawing pixels for a target ship and other ships. And according to the preset drawing rules, drawing vectors of the ships in the two-dimensional plan according to the longitude and latitude coordinates of the ships, the heading of the ships and the speed of the ships, drawing target ships by using red pixels, drawing other ships by using blue pixels, and normalizing the obtained two-dimensional plan to a range of 128 x 128 to obtain a ship plan. By obtaining the ship plan, the movement track of the ship in the preset interval time can be clearly and intuitively displayed.
Step S400: when the requirements of the preset shooting quantity are met, summarizing all ship plane graphs according to a time sequence based on AIS information of all ships, and constructing ship AIS track graphs;
specifically, the number of ship shots is preset, which can be set by those skilled in the art in a custom manner based on actual conditions, for example: 10 times. And when the ship shooting times are equal to the preset shooting quantity, acquiring AIS information of a plurality of groups of all ships, which are the same as the preset ship shooting quantity, summarizing the AIS information of a plurality of groups of all ships on the ship plan according to a time sequence, and drawing running tracks of all ships by using directional arrows, wherein the directional arrows of target ships are drawn by using red pixels, and the directional arrows of other ships are drawn by using blue pixels, so as to obtain a ship AIS track graph. By obtaining the ship AIS track map, the running tracks of all ships in a period of time can be intuitively displayed, and data support is provided for the next step of ship state evaluation.
Step S500: inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model to obtain ship state evaluation information;
as shown in fig. 2, in one embodiment, step S500 of the present application further includes:
step S510: constructing a training data set, wherein the training data set comprises a target ship track graph, a camera shooting graph and risk assessment scoring thereof;
as shown in fig. 3, in one embodiment, step S510 of the present application further includes:
step S511: collecting a target ship track diagram and a camera shooting diagram of a target ship according to the preset data quantity requirement, wherein the camera shooting diagram comprises various environmental features, and the environmental features represent shooting environmental conditions of the target ship;
step S512: transmitting the target ship track map and the camera shooting map of the target ship to a scoring channel for scoring the ship behavior risk degree, and obtaining scoring results of the target ship track map and the camera shooting map of the target ship;
step S513: performing score distribution uniformity analysis based on the scoring result, supplementing missing data when the distribution uniformity does not meet the preset distribution requirement, and repeatedly scoring until the distribution uniformity meets the preset distribution requirement;
Step S514: normalizing all the camera shooting pictures;
step S515: grouping all target ship track diagrams and normalized camera shooting diagrams according to the grouping number to obtain the training data set.
Specifically, a training dataset is constructed, which includes a target ship trajectory graph, a target ship camera shot graph, and a target ship risk degree scoring result. Firstly, collecting a target ship track map and a camera shooting map of a target ship according to a preset data quantity, wherein the preset data quantity can be set in a customized mode based on actual conditions by a person skilled in the art, for example: 10000 sets of data, wherein the camera shooting map comprises a plurality of environmental features, wherein the environmental features characterize shooting environmental conditions in which a target ship is located, such as: various environmental conditions with low visibility such as fog days, rainy days, night, etc.
Constructing a scoring channel, wherein the scoring channel consists of a plurality of professionals with rich ship sailing experience, and then sending the target ship track map and the camera shooting map of the target ship to the scoring channel to comprehensively score the ship behavior risk degree through the plurality of professionals, wherein the scoring process can be scored by the plurality of professionals, for example: a scoring range between 0-1 may be provided, where the higher the score the greater the risk. And taking the average value of scoring of a plurality of professionals as a final risk scoring result, obtaining a risk scoring result of the track map of each target ship and the camera shooting map of the target ship, namely the scoring result, and scoring the track map of the target ship and the camera shooting map of the target ship by the plurality of professionals, so that the accuracy of obtaining the risk scoring result can be improved. And then carrying out distribution uniformity analysis on the scoring result, wherein the distribution uniformity analysis refers to judging whether the scoring result is discrete uniform distribution or not, and presetting distribution requirements, wherein the preset distribution requirements can be set by a person skilled in the art in a self-defined way, for example: the scoring results may be set to 5 levels, with 5 levels containing at least one datum among the 30 consecutive scoring results. And when the distribution uniformity does not meet the preset distribution requirement, supplementing missing data, and repeatedly scoring a target ship track diagram or a camera shooting diagram of the target ship through the scoring channel until the preset distribution requirement is met. And then normalizing all the camera shooting images, wherein the normalization process is to normalize all the camera shooting images to the range of 128 x 128. All target ship track diagrams and normalized camera shooting diagrams are then grouped according to the grouping number, wherein the grouping number can be set by a person skilled in the art in a self-defined manner, for example: and dividing the 32 images into a group, obtaining a target ship track image group and a camera shooting image group, and constructing a training data set based on the target ship track image group and the camera shooting image group. By obtaining the training data set, support is provided for the next step of model supervision training.
Step S520: constructing a convolution network structure, wherein the convolution network structure comprises a CNN feature extraction layer, a feature fusion layer, a convolution processing layer and a full connection layer;
step S530: and training the convolutional network structure by using the training data set, and obtaining the state evaluation model when the convergence target requirement is met.
In one embodiment, step S530 of the present application further includes:
step S531: inputting the training data set into a CNN feature extraction layer, wherein the CNN feature extraction layer comprises different CNN extraction layers, and performing feature extraction on the training data set by using different CNNs to obtain a first feature map and a second feature map, wherein the first feature map is AIS features, and the second feature map is camera shooting map features;
step S532: inputting the first feature map and the second feature map into the feature fusion layer to perform feature fusion, so as to obtain a fusion feature map;
step S533: inputting the fused feature images into the convolution processing layer, and convolving the fused feature images through the convolution layer to obtain new feature images;
step S534: inputting the new feature map into the full-connection layer, and integrating two layers of full-connection features to obtain a state prediction score result;
step S535: and carrying out loss function calculation based on the state prediction score result and the scoring result in the training data set, and carrying out model parameter correction and retraining according to the loss value until reaching the convergence target requirement.
In one embodiment, step S535 of the present application further comprises:
step S5351: by regression of the loss function:
a loss value calculation is performed, wherein,for scoring results, < >>Score results are predicted for states.
Specifically, a convolution network structure of the state evaluation model is constructed, wherein the convolution network structure comprises a CNN feature extraction layer and a feature fusion layerA convolution processing layer and a full connection layer. And then inputting the training data set into the CNN feature extraction layer to perform feature extraction to obtain a first feature map and a second feature map, wherein the first feature map and the second feature map are both 64 x 128 x 32 feature maps, the first feature map is an AIS feature, and the second feature map is a camera shooting image feature. And then inputting the first feature map and the second feature map into the feature fusion layer to conduct concat feature fusion to obtain 64×64×256×32 feature maps, and obtaining a fusion feature map. And inputting the fusion feature map into the convolution processing layer, and carrying out convolution of 3 x 512 on the fused feature map through the convolution layer to obtain a new feature map of 64 x 512. And inputting the new feature map into the full-connection layer, and integrating the full-connection features of 1 x 512 of the two layers to obtain a 1 x 32 predictive score result, namely the state predictive score result. And finally, calculating a mean square error regression loss function between the state prediction score result and a scoring result in the training data set, wherein the regression loss function is as follows: Wherein->For scoring results, < >>Score results are predicted for states. And carrying out model parameter correction according to the loss value calculated by the regression loss function, then continuing to carry out supervision training through a training data set, and stopping training when the model output result tends to a convergence state, namely when the output result is unchanged, so as to obtain the state evaluation model. By constructing a state evaluation model based on a convolutional neural network, the efficiency and accuracy of ship state evaluation can be improved. And finally, inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model to perform state evaluation, and obtaining ship state evaluation information.
Step S600: classifying according to preset classification requirements based on the ship state evaluation information, and determining ship behavior classification.
In one embodiment, step S600 of the present application further includes:
step S610: performing ship body characteristics and carrier risk assessment on the target ship to obtain a target ship risk level;
step S620: determining a safety classification adjustment coefficient according to the target ship risk level;
specifically, the ship body characteristics and the cargo of the target ship are obtained, the ship body characteristics and the cargo risk of the target ship are evaluated, and the risk level of the target ship is obtained according to the evaluation result. For example: when the risk assessment is carried out from the characteristics of the ship body, the risk assessment can be carried out from various aspects such as the specification of the ship body, the age of the ship, the history sailing record of the ship body and the like, the higher the age of the ship is, the higher the risk coefficient is, and the risk assessment can be carried out according to the type and the loading capacity of the carrier during the carrier risk assessment, for example: the risk coefficient of agricultural products such as grains, fruits and the like is low, and the risk coefficient of petroleum, natural gas and the like is high. The weight distribution coefficients are preset, and the weight distribution coefficients can be set by a person skilled in the art in a self-defined manner, for example: 40% and 60%. And carrying out weighted calculation on the ship feature risk assessment result and the carrier risk assessment result according to the preset weight distribution coefficient to obtain a risk assessment result, and obtaining a target ship risk level according to the risk assessment result. The target ship risk level can be set by a person skilled in the art in a customized manner, and the larger the value of the risk assessment result is, the higher the target ship risk level is.
And determining a safety classification adjustment coefficient according to the target ship risk level. The security class adjustment coefficients can be custom set by those skilled in the art based on the actual situation, for example: the target ship risk level is 7, wherein the target ship risk level sets the adjustment factor to-3 for one stage, and then adds 1 to each stage adjustment factor. By obtaining the security classification adjustment coefficient, support is provided for the next step of ship risk level adjustment.
Step S630: performing scoring numerical adjustment on the preset classification requirements by using the safety classification adjustment coefficient, and determining an execution rule of the preset classification requirements;
in one embodiment, step S630 of the present application further includes:
step S631: obtaining a scoring rule of a scoring channel, carrying out semantic analysis on the scoring rule, and determining a scoring parameter-score mapping relation;
step S632: performing scoring latitude calculation based on the scoring parameter-score mapping relation;
step S633: determining a scoring-classifying relation by combining the latitude of the scoring parameters with classifying requirements, and establishing the preset classifying requirement based on the scoring-classifying relation, wherein when the latitude of the scoring parameters is smaller than the classifying requirement, standard scores are determined to cluster based on the classifying requirement, and the scoring-classifying relation is determined; and when the latitude of the scoring parameter is larger than the classification requirement, splitting the scoring result based on the corresponding relation between the classification requirement and the scoring parameter, and determining the scoring-classification relation.
Specifically, an evaluation rule of a scoring channel is obtained, wherein the evaluation rule refers to a risk evaluation index of a professional. And carrying out semantic analysis on the scoring rules to determine the mapping relation between scoring parameters and scores. And performing scoring latitude calculation according to the scoring parameter-score mapping relation, wherein the scoring latitude calculation refers to judging the score span in the scoring parameter-score mapping relation, and the larger the score span is, the less strict the scoring is, and the larger the latitude is, and the smaller the latitude is otherwise. And combining classification requirements according to the latitude of the scoring parameters, wherein the classification requirements refer to indexes of classification, and a scoring-classifying relationship is determined. And determining the preset classification requirement according to the scoring-classifying relationship.
And when the tolerance of the scoring parameter is smaller than the classification requirement, determining a standard score based on the classification requirement for clustering, and determining the scoring-classification relation. When the classification requirement indexes are more or the similarity is higher, a plurality of values can be selected for clustering, the number of classifications is reduced, and the relationship of scoring classification is redetermined. For example: in the scoring classification relation, the classification is classified into one level every 0.02, and if the classification is more dense, 3 indexes can be aggregated, and the classification is classified into one level every 0.06. And when the latitude of the scoring parameter is larger than the classification requirement, splitting the scoring result based on the corresponding relation between the classification requirement and the scoring parameter, and determining the scoring-classification relation. For example: in the scoring classification relation, one level is divided every 0.2, the latitude is high, the 0.2 level can be divided into 2 levels, and the 0.1 level can be divided into 1 level. By establishing the preset classification requirements based on the scoring and classification relations, the scoring precision can be improved, and the accuracy of the ship state evaluation result is indirectly improved. And finally, performing scoring numerical adjustment on the preset classification requirements according to the safety classification adjustment coefficient, and determining an execution rule of the preset classification requirements, wherein the execution rule of the classification requirements comprises a plurality of score levels and corresponding ship behavior categories.
Step S640: and carrying out score matching on the ship state evaluation information and the execution rule required by the preset classification to obtain a corresponding ship behavior classification, wherein the ship behavior classification represents the safety level of the navigation state of the ship.
Specifically, the ship state evaluation information is subjected to score matching with the execution rule required by the preset classification, and a corresponding ship behavior classification is obtained, wherein the ship behavior classification is used for representing the safety level of the navigation state of the ship. The method solves the problem that the collision early warning judgment accuracy rate of the ship is low by solely depending on AIS information of surrounding ships in a complex weather environment, and can improve the accuracy rate of the collision early warning judgment of the ship in the complex environment.
In one embodiment, there is provided as shown in fig. 4 a ship behavior classification system based on AIS trace patterns and camera pattern sets, comprising: a target ship monitoring module 1, a ship monitoring information recording module 2, a ship plane diagram obtaining module 3, a ship AIS track diagram constructing module 4, a ship state evaluation information obtaining module 5, a ship behavior classification determining module 6, wherein:
the target ship monitoring module 1 is used for installing AIS equipment and a camera together, wherein the longitude and latitude of the AIS equipment are kept consistent with those of the camera, and when the AIS equipment detects that a target ship enters a detection range, a signal is synchronously activated to the camera to control the camera to monitor the target ship;
The ship monitoring information recording module 2 is used for recording monitoring information of all ships in the detection area according to preset interval time, and shooting a target ship picture through a camera, wherein the ship monitoring information comprises longitude and latitude coordinates of the ship, the ship course and the ship speed information, and a ship monitoring database is constructed;
the ship plane map obtaining module 3 is used for extracting parameters according to preset drawing rules based on the ship monitoring database, and drawing the ship plane map of the extracted parameters according to the preset drawing rules to obtain a ship plane map, wherein the ship plane map comprises a target ship, and the drawing pixels of the target ship are different from those of other ships;
the ship AIS track map construction module 4 is used for summarizing all ship plane maps according to a time sequence based on AIS information of all ships when the ship AIS track map construction module 4 meets the requirement of a preset shooting quantity to construct a ship AIS track map;
the ship state evaluation information acquisition module 5 is used for inputting target ship pictures and ship AIS track pictures of a preset shooting number into a state evaluation model to acquire ship state evaluation information;
And the ship behavior classification determining module 6 is used for classifying according to preset classification requirements based on the ship state evaluation information, and determining the ship behavior classification.
In one embodiment, the system further comprises:
the training data set building module is used for building a training data set and comprises a target ship track graph, a camera shooting graph and risk assessment scoring thereof;
the convolution network structure construction module is used for constructing a convolution network structure and comprises a CNN feature extraction layer, a feature fusion layer, a convolution processing layer and a full connection layer;
the state evaluation model obtaining module is used for training the convolution network structure by utilizing the training data set, and obtaining the state evaluation model when the convergence target requirement is met.
In one embodiment, the system further comprises:
the information collection module is used for collecting a target ship track diagram and a camera shooting diagram of the target ship according to the preset data quantity requirement, wherein the camera shooting diagram comprises various environmental features, and the environmental features represent shooting environmental conditions of the target ship;
The risk scoring module is used for sending the target ship track map and the camera shooting map of the target ship to a scoring channel to score the ship behavior risk degree, and obtaining scoring results of the target ship track map and the camera shooting map of the target ship;
the missing data supplementing module is used for analyzing the distribution uniformity of the scores based on the scoring result, supplementing the missing data when the distribution uniformity does not meet the preset distribution requirement, and repeatedly scoring until the distribution uniformity meets the preset distribution requirement;
the image normalization processing module is used for carrying out normalization processing on all the camera shooting pictures;
the training data set obtaining module is used for grouping all target ship track diagrams and normalized camera shooting diagrams according to the grouping number to obtain the training data set.
In one embodiment, the system further comprises:
the feature extraction module is used for inputting the training data set into a CNN feature extraction layer, wherein the feature extraction module comprises different CNN extraction layers, and the training data set is subjected to feature extraction by using different CNNs to obtain a first feature map and a second feature map, wherein the first feature map is an AIS feature, and the second feature map is a camera shooting map feature;
The feature fusion module is used for inputting the first feature map and the second feature map into the feature fusion layer to perform feature fusion, so as to obtain a fusion feature map;
the new feature map obtaining module is used for inputting the fused feature map into the convolution processing layer, and convolving the fused feature map through the convolution layer to obtain a new feature map;
the state prediction score result obtaining module is used for inputting the new feature map into the fully-connected layer, and integrating two layers of fully-connected features to obtain a state prediction score result;
and the loss function calculation module is used for carrying out loss function calculation based on the state prediction score result and the scoring result in the training data set, and carrying out model parameter correction and retraining according to the loss value until the convergence target requirement is reached.
In one embodiment, the system further comprises:
a loss value calculation module for calculating a loss function by regression:
a loss value calculation is performed, wherein,for scoring results, < >>Score results are predicted for states.
In one embodiment, the system further comprises:
the target ship risk level obtaining module is used for carrying out ship body characteristic and carrier risk assessment on the target ship to obtain a target ship risk level;
the safety classification adjustment coefficient determining module is used for determining a safety classification adjustment coefficient according to the risk level of the target ship;
the scoring value adjusting module is used for performing scoring value adjustment on the preset classification requirement by utilizing the safety classification adjusting coefficient and determining an execution rule of the preset classification requirement;
and the ship behavior classification obtaining module is used for carrying out score matching on the ship state evaluation information and the execution rule required by the preset classification to obtain a corresponding ship behavior classification, and the ship behavior classification represents the safety level of the navigation state of the ship.
In one embodiment, the system further comprises:
the mapping relation determining module is used for obtaining a scoring rule of the scoring channel, carrying out semantic analysis on the scoring rule and determining a scoring parameter-score mapping relation;
The scoring latitude calculating module is used for calculating scoring latitude based on the scoring parameter-score mapping relation;
the preset classification requirement calculation module is used for determining a scoring-classification relation by combining the latitude of the scoring parameters with the classification requirement, and establishing the preset classification requirement based on the scoring-classification relation, wherein when the latitude of the scoring parameters is smaller than the classification requirement, standard scores are determined for clustering based on the classification requirement, and the scoring-classification relation is determined; and when the latitude of the scoring parameter is larger than the classification requirement, splitting the scoring result based on the corresponding relation between the classification requirement and the scoring parameter, and determining the scoring-classification relation.
In summary, the ship behavior classification method based on the AIS track map and the camera map set has the following technical effects:
1. the problem that the collision early warning judgment accuracy rate of the ship is low by solely depending on AIS information of surrounding ships in a complex weather environment is solved, and the accuracy rate of the collision early warning judgment of the ship in the complex environment can be improved.
2. By constructing a state evaluation model based on a convolutional neural network, the efficiency and accuracy of ship state evaluation can be improved.
3. By establishing the preset classification requirements based on the scoring and classification relations, the scoring precision can be improved, and the accuracy of the ship state evaluation result is indirectly improved.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.
Claims (5)
1. The ship behavior classification method based on the AIS track diagram and the camera diagram group is characterized by comprising the following steps:
installing AIS equipment and a camera together, wherein the longitude and latitude of the AIS equipment are kept consistent with those of the camera, and when the AIS equipment detects that a target ship enters a detection range, activating signals are synchronized to the camera to control the camera to monitor the target ship;
Recording monitoring information of all ships in a detection area according to preset interval time, and shooting a target ship picture through a camera, wherein the ship monitoring information comprises longitude and latitude coordinates of the ship, ship heading and ship speed information, and constructing a ship monitoring database;
extracting parameters according to preset drawing rules based on the ship monitoring database, and drawing a ship plane graph of the extracted parameters according to the preset drawing rules to obtain a ship plane graph, wherein the ship plane graph comprises a target ship, and the drawing pixels of the target ship are different from those of other ships;
when the requirements of the preset shooting quantity are met, summarizing all ship plane graphs according to a time sequence based on AIS information of all ships, and constructing ship AIS track graphs;
inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model to obtain ship state evaluation information;
classifying according to preset classification requirements based on the ship state evaluation information, and determining ship behavior classification;
the method comprises the steps of inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model, and comprises the following steps:
constructing a training data set, wherein the training data set comprises a target ship track graph, a camera shooting graph and risk assessment scoring thereof;
Constructing a convolution network structure, wherein the convolution network structure comprises a CNN feature extraction layer, a feature fusion layer, a convolution processing layer and a full connection layer;
training the convolutional network structure by using the training data set, and obtaining the state evaluation model when the convergence target requirement is met;
wherein classifying according to a preset classification requirement based on the ship state evaluation information, determining a ship behavior classification, comprises:
performing ship body characteristics and carrier risk assessment on the target ship to obtain a target ship risk level;
determining a safety classification adjustment coefficient according to the target ship risk level;
performing scoring numerical adjustment on the preset classification requirements by using the safety classification adjustment coefficient, and determining an execution rule of the preset classification requirements;
performing score matching on the ship state evaluation information and the execution rule required by the preset classification to obtain a corresponding ship behavior classification, wherein the ship behavior classification represents the safety level of the navigation state of the ship;
obtaining a scoring rule of a scoring channel, carrying out semantic analysis on the scoring rule, and determining a scoring parameter-score mapping relation;
performing scoring latitude calculation based on the scoring parameter-score mapping relation;
Determining a scoring-classifying relationship by combining the latitude of the scoring parameter with classifying requirements, and establishing the preset classifying requirement based on the scoring-classifying relationship, wherein when the latitude of the scoring parameter is smaller than the classifying requirement, standard scores are determined for clustering based on the classifying requirement, and the scoring-classifying relationship is determined; and when the latitude of the scoring parameter is larger than the classification requirement, splitting the scoring result based on the corresponding relation between the classification requirement and the scoring parameter, and determining the scoring-classification relation.
2. The method of claim 1, wherein the constructing a training data set comprises:
collecting a target ship track diagram and a camera shooting diagram of a target ship according to the preset data quantity requirement, wherein the camera shooting diagram comprises various environmental features, and the environmental features represent shooting environmental conditions of the target ship;
transmitting the target ship track map and the camera shooting map of the target ship to a scoring channel for scoring the ship behavior risk degree, and obtaining scoring results of the target ship track map and the camera shooting map of the target ship;
performing score distribution uniformity analysis based on the scoring result, supplementing missing data when the distribution uniformity does not meet the preset distribution requirement, and repeatedly scoring until the distribution uniformity meets the preset distribution requirement;
Normalizing all the camera shooting pictures;
grouping all target ship track diagrams and normalized camera shooting diagrams according to the grouping number to obtain the training data set.
3. The method of claim 1, wherein the training the convolutional network structure with the training data set comprises:
inputting the training data set into a CNN feature extraction layer, wherein the CNN feature extraction layer comprises different CNN extraction layers, and performing feature extraction on the training data set by using different CNNs to obtain a first feature map and a second feature map, wherein the first feature map is AIS features, and the second feature map is camera shooting map features;
inputting the first feature map and the second feature map into the feature fusion layer to perform feature fusion, so as to obtain a fusion feature map;
inputting the fused feature images into the convolution processing layer, and convolving the fused feature images through the convolution layer to obtain new feature images;
inputting the new feature map into the full-connection layer, and integrating two layers of full-connection features to obtain a state prediction score result;
and carrying out loss function calculation based on the state prediction score result and the scoring result in the training data set, and carrying out model parameter correction and retraining according to the loss value until reaching the convergence target requirement.
4. The method of claim 3, wherein performing a loss function calculation based on the state prediction score result and a scoring result in a training dataset comprises:
by regression of the loss function:
performing loss value calculation, wherein y i In order to score the result of the score,score results are predicted for states.
5. The ship behavior classification system based on the AIS track graph and the camera graph group is characterized by comprising:
the target ship monitoring module is used for installing AIS equipment and the camera together, wherein the longitude and latitude of the AIS equipment are kept consistent with those of the camera, and when the AIS equipment detects that a target ship enters a detection range, a signal is synchronously activated to the camera to control the camera to monitor the target ship;
the ship monitoring information recording module is used for recording monitoring information of all ships in the detection area according to preset interval time, and shooting a target ship picture through the camera, wherein the ship monitoring information comprises longitude and latitude coordinates of the ship, ship course and ship speed information, and a ship monitoring database is constructed;
the ship plane map obtaining module is used for extracting parameters according to preset drawing rules based on the ship monitoring database, and drawing the ship plane map of the extracted parameters according to the preset drawing rules to obtain a ship plane map, wherein the ship plane map comprises a target ship, and the drawing pixels of the target ship are different from those of other ships;
The ship AIS track map construction module is used for summarizing all ship plane maps according to a time sequence based on AIS information of all ships when the ship AIS track map construction module meets the requirement of the preset shooting quantity to construct a ship AIS track map;
the ship state evaluation information acquisition module is used for inputting target ship pictures and ship AIS track pictures with preset shooting quantity into a state evaluation model to acquire ship state evaluation information;
the ship behavior classification determining module is used for classifying according to preset classification requirements based on the ship state evaluation information and determining ship behavior classification;
the training data set building module is used for building a training data set and comprises a target ship track graph, a camera shooting graph and risk assessment scoring thereof;
the convolution network structure construction module is used for constructing a convolution network structure and comprises a CNN feature extraction layer, a feature fusion layer, a convolution processing layer and a full connection layer;
the state evaluation model obtaining module is used for training the convolution network structure by utilizing the training data set, and obtaining the state evaluation model when the convergence target requirement is met;
The target ship risk level obtaining module is used for carrying out ship body characteristic and carrier risk assessment on the target ship to obtain a target ship risk level;
the safety classification adjustment coefficient determining module is used for determining a safety classification adjustment coefficient according to the risk level of the target ship;
the scoring value adjusting module is used for performing scoring value adjustment on the preset classification requirement by utilizing the safety classification adjusting coefficient and determining an execution rule of the preset classification requirement;
the ship behavior classification obtaining module is used for carrying out score matching on the ship state evaluation information and the execution rules required by the preset classification to obtain corresponding ship behavior classification, and the ship behavior classification represents the safety level of the navigation state of the ship;
the mapping relation determining module is used for obtaining a scoring rule of the scoring channel, carrying out semantic analysis on the scoring rule and determining a scoring parameter-score mapping relation;
the scoring latitude calculating module is used for calculating scoring latitude based on the scoring parameter-score mapping relation;
The preset classification requirement calculation module is used for determining a scoring-classification relation by combining the latitude of the scoring parameters with the classification requirement, and establishing the preset classification requirement based on the scoring-classification relation, wherein when the latitude of the scoring parameters is smaller than the classification requirement, standard scores are determined for clustering based on the classification requirement, and the scoring-classification relation is determined; and when the latitude of the scoring parameter is larger than the classification requirement, splitting the scoring result based on the corresponding relation between the classification requirement and the scoring parameter, and determining the scoring-classification relation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310564867.9A CN116310601B (en) | 2023-05-19 | 2023-05-19 | Ship behavior classification method based on AIS track diagram and camera diagram group |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310564867.9A CN116310601B (en) | 2023-05-19 | 2023-05-19 | Ship behavior classification method based on AIS track diagram and camera diagram group |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116310601A CN116310601A (en) | 2023-06-23 |
| CN116310601B true CN116310601B (en) | 2023-08-04 |
Family
ID=86796360
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310564867.9A Active CN116310601B (en) | 2023-05-19 | 2023-05-19 | Ship behavior classification method based on AIS track diagram and camera diagram group |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116310601B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117315573A (en) * | 2023-09-20 | 2023-12-29 | 亿海蓝（北京）数据技术股份公司 | Oil supply ship operation behavior identification method and system, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106816038A (en) * | 2017-03-17 | 2017-06-09 | 武汉理工大学 | A kind of inland waters abnormal behaviour ship automatic identification system and method |
| CN109766811A (en) * | 2018-12-31 | 2019-05-17 | 复旦大学 | The end-to-end detection and recognition methods of sea ship in a kind of satellite-borne SAR image |
| CN110796900A (en) * | 2019-12-03 | 2020-02-14 | 闽江学院 | Fishing boat safety monitoring command system |
| KR20220045762A (en) * | 2020-10-06 | 2022-04-13 | 주식회사 스카이시스 | System for automatic recognition and monitoring of vessel using artificial intelligence image processing and method for providing the same |
| KR20220055555A (en) * | 2020-10-26 | 2022-05-04 | 씨드로닉스(주) | Method and device for monitoring harbor and ship |
| CN114510961A (en) * | 2022-01-03 | 2022-05-17 | 中国电子科技集团公司第二十研究所 | Ship behavior intelligent monitoring algorithm based on recurrent neural network and Beidou positioning |
-
2023
- 2023-05-19 CN CN202310564867.9A patent/CN116310601B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106816038A (en) * | 2017-03-17 | 2017-06-09 | 武汉理工大学 | A kind of inland waters abnormal behaviour ship automatic identification system and method |
| CN109766811A (en) * | 2018-12-31 | 2019-05-17 | 复旦大学 | The end-to-end detection and recognition methods of sea ship in a kind of satellite-borne SAR image |
| CN110796900A (en) * | 2019-12-03 | 2020-02-14 | 闽江学院 | Fishing boat safety monitoring command system |
| KR20220045762A (en) * | 2020-10-06 | 2022-04-13 | 주식회사 스카이시스 | System for automatic recognition and monitoring of vessel using artificial intelligence image processing and method for providing the same |
| KR20220055555A (en) * | 2020-10-26 | 2022-05-04 | 씨드로닉스(주) | Method and device for monitoring harbor and ship |
| CN114510961A (en) * | 2022-01-03 | 2022-05-17 | 中国电子科技集团公司第二十研究所 | Ship behavior intelligent monitoring algorithm based on recurrent neural network and Beidou positioning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116310601A (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Vehicle-damage-detection segmentation algorithm based on improved mask RCNN | |
| KR102129893B1 (en) | Ship tracking method and system based on deep learning network and average movement | |
| KR101995107B1 (en) | Method and system for artificial intelligence based video surveillance using deep learning | |
| WO2021249071A1 (en) | Lane line detection method, and related apparatus | |
| CN110889324A (en) | Thermal infrared image target identification method based on YOLO V3 terminal-oriented guidance | |
| CN116310601B (en) | Ship behavior classification method based on AIS track diagram and camera diagram group | |
| CN109145836B (en) | Ship target video detection method based on deep learning network and Kalman filtering | |
| CN111222574A (en) | Ship and civil ship target detection and classification method based on multi-model decision-level fusion | |
| CN110619276B (en) | Anomaly and violence detection system and method based on unmanned aerial vehicle mobile monitoring | |
| CN113962282B (en) | Ship cabin fire real-time detection system and method based on improved yolov5l+deep | |
| CN110610165A (en) | Ship behavior analysis method based on YOLO model | |
| CN111414807A (en) | Tidal water identification and crisis early warning method based on YO L O technology | |
| CN111159924A (en) | Method and apparatus for predicting drift trajectory | |
| CN108960421A (en) | The unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network of improvement | |
| CN114564545A (en) | System and method for extracting ship experience course based on AIS historical data | |
| CN112149618B (en) | Pedestrian abnormal behavior detection method and device suitable for inspection vehicle | |
| US20200133283A1 (en) | Proceedable direction detection apparatus and proceedable direction detection method | |
| CN115995165A (en) | Ship navigation risk management method and system | |
| CN111163290A (en) | Device and method for detecting and tracking night navigation ship | |
| CN116206223A (en) | Fire detection method and system based on unmanned aerial vehicle edge calculation | |
| CN115937659A (en) | Mask-RCNN-based multi-target detection method in indoor complex environment | |
| CN114139373B (en) | Multi-sensor automatic collaborative management method for unmanned aerial vehicle reverse vehicle | |
| CN113420819B (en) | Lightweight underwater target detection method based on CenterNet | |
| CN115171006B (en) | Detection method for automatically identifying person entering electric power dangerous area based on deep learning | |
| CN115880620B (en) | Personnel counting method applied to cart early warning system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |