CN109117794A - A kind of moving target behavior tracking method, apparatus, equipment and readable storage medium storing program for executing - Google Patents
A kind of moving target behavior tracking method, apparatus, equipment and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN109117794A CN109117794A CN201810935776.0A CN201810935776A CN109117794A CN 109117794 A CN109117794 A CN 109117794A CN 201810935776 A CN201810935776 A CN 201810935776A CN 109117794 A CN109117794 A CN 109117794A
- Authority
- CN
- China
- Prior art keywords
- moving target
- target
- detection model
- behavior tracking
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a kind of moving target behavior tracking method, apparatus, equipment and computer readable storage mediums.Wherein, method includes that the target detection model constructed in advance is called to carry out target detection to the movement destination image of acquisition, to position moving target position in movement destination image.Wherein, target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for loss function training gained；And target detection model meets the object boundary frame of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm, and determines the centre coordinate of object boundary frame, to position the position of moving target.The application is positioned moving target using target detection iconic model in the movement destination image of acquisition, for the multiframe consecutive image continuously obtained, by positioning moving target in every frame image, to realize the tracing detection to moving target behavior in video.The application improves moving object detection speed and positional accuracy.
Description
Technical field
The present embodiments relate to technical field of image processing, more particularly to a kind of moving target behavior tracking method,
Device, equipment and computer readable storage medium.
Background technique
With the continuous development of information technology, network courses are also popularized therewith, and the research of quality of instruction is also more convenient.For
It is better to study the effect given lessons, the behaviortrace of student and teacher are particularly important, the effect of tracking can be used for
Teaching improving and real-time recorded broadcast control.Then numerous production firms and research institution see the huge quotient wherein contained
Machine, to start the development work of intelligent classroom video recording and broadcasting system, and many colleges and universities in the whole nation begin setting up respective reality
Test room.By the research and development of more than ten years, the performance of intelligent classroom video recording and broadcasting system is continuously improved, and function constantly improve, perhaps
The middle and primary schools in more schools especially basic education stage gradually see the value that intelligent recording and broadcasting system is contained.
In recent years, with the rise of artificial intelligence technology, Faster-RCNN algorithm is widely used in moving target behavior
Feature detection and tracking.The characteristic area that this method uses the convolutional layer of convolutional neural networks (CNN) that pond layer is added to extract picture first
Domain, these characteristic areas be shared for subsequent region choose network (Region Proposal Networks, RPN) and entirely
Articulamentum.Then candidate frame is generated using RPN network, which first passes through the full articulamentum of softmax and judge that anchor (anchors) belongs to
Then prospect or background are modifying anchor using bounding box (bounding box regression) is returned, to obtain essence
True candidate frame.The characteristic area of input is finally collected by the pond RoI layer, and these data are sent into full articulamentum and carry out target
The judgement of classification.
Although Faster-RCNN algorithm can realize behavior tracking and the detection of moving target, the framework ratio of the algorithm
It is more complex, it is higher to demanding terminal, in video or require to be difficult to carry out in the very fast application scenarios for identifying object.
In consideration of it, how in recorded broadcast by controlling the tracking of teacher the movement of video recorder, with realize completely nobody
Classroom video is recorded in the state of control video camera, is those skilled in the art's urgent problem to be solved.
Summary of the invention
It can the purpose of the embodiment of the present invention is that providing a kind of moving target behavior tracking method, apparatus, equipment and computer
Storage medium is read, it can be achieved that tracing detection to moving target behavior in video.
In order to solve the above technical problems, the embodiment of the present invention the following technical schemes are provided:
On the one hand the embodiment of the present invention provides a kind of moving target behavior tracking method, comprising:
Obtain movement destination image；
The target detection model constructed in advance is called, moving target position is positioned in the movement destination image；
Wherein, the target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for loss letter
Number training gained；The target detection model meets the mesh of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm
Bounding box is marked, and determines the center position coordinate of the moving target characteristic area, to position the position of the moving target.
Optionally, the loss function of the target detection model are as follows:
In formula, L is that two-value intersects loss function, and R is the loss function after regularization, ziFor predicted value, yiFor desired value,
ωiFor the weighted value of each layer.
Optionally, the target detection model uses 53 layers of convolutional network, and is formed by stacking by multiple residual units.
Optionally, the target detection model extracts the feature of the movement destination image using feature pyramid network,
The object detection results of output include the class label of bounding box information and the moving target.
Optionally, the target detection model is predicted in the characteristic area of the moving target using K-means algorithm
The length value and width value of heart point position coordinates and bounding box.
Optionally, the object detection results include the bounding box and the corresponding central point of each bounding box of three different scales
Position coordinates.
Optionally, the target detection model trains the spy of the moving target using quadratic sum range error loss method
Levy the center position coordinate in region and the length value and width value of bounding box.
On the other hand the embodiment of the present invention provides a kind of moving target behavior tracking device, comprising:
Image collection module, for obtaining movement destination image；
Target locating module positions in the movement destination image for calling the target detection model constructed in advance
Moving target position；The target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for damage
Lose function training gained；The target detection model meets preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm
Object boundary frame, and the center position coordinate of the moving target characteristic area is determined, to position the moving target
Position.
The embodiment of the invention also provides a kind of moving target behavior tracking equipment, including processor, the processor is used
The step of realizing the moving target behavior tracking method as described in preceding any one when executing the computer program stored in memory.
The embodiment of the present invention finally additionally provides a kind of computer readable storage medium, the computer readable storage medium
On be stored with moving target behavior tracking program, realize when the moving target behavior tracking program is executed by processor as former
The step of one moving target behavior tracking method.
The embodiment of the invention provides a kind of moving target behavior tracking methods, call the target detection model constructed in advance
Target detection is carried out to the movement destination image of acquisition, to position moving target position in movement destination image.Its
In, target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for loss function training gained；And mesh
Mark detection model meets the object boundary frame of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm, and determines mesh
The centre coordinate of bounding box is marked, to position the position of moving target.
The advantages of technical solution provided by the present application, is, using target detection iconic model by moving target in acquisition
It is positioned in movement destination image, for the multiframe consecutive image continuously obtained, namely for video, by every frame image
Moving target is positioned, to realize the tracing detection to moving target behavior in video.Due to YOLOv3 algorithm process picture
Speed quickly, under the same conditions, is based on YOLOv3 algorithm training objective detection model to image processing speed than existing volume
The processing speed of the model (1000 times such as faster than R-CNN, 100 times faster than Fast-RCNN) of product neural network algorithm training is fast.It adopts
It is loss function with regularization and two-value cross entropy, can further improve the detection speed and positional accuracy of moving target.This
Outside, the transplanting of YOLOv3 algorithm is convenient, can realize under each operating system, relatively low to the configuration requirement of terminal hardware,
The operation of target detection model can be easier realized in lightweight equipment.For the application scenarios of real-time recorded broadcast, this Shen
The accurate positionin and quickly identification to moving target feature please can be realized under identical precision, what raising was identified in video field
Speed and precision reduces the delay and Caton of recording and broadcasting system.
In addition, the embodiment of the present invention provides corresponding realization device, equipment also directed to moving target behavior tracking method
And computer readable storage medium, further such that the method has more practicability, described device, equipment and computer-readable
Storage medium has the advantages that corresponding.
Detailed description of the invention
It, below will be to embodiment or existing for the clearer technical solution for illustrating the embodiment of the present invention or the prior art
Attached drawing needed in technical description is briefly described, it should be apparent that, the accompanying drawings in the following description is only this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is a kind of flow diagram of moving target behavior tracking method provided in an embodiment of the present invention；
Fig. 2 is that a kind of bounding box provided in an embodiment of the present invention is handed over and compares schematic diagram；
Fig. 3 is a kind of display schematic diagram of real border frame and predicted boundary frame provided in an embodiment of the present invention；
Fig. 4 is feature pyramid network diagram provided in an embodiment of the present invention；
Fig. 5 is residual error schematic network structure provided in an embodiment of the present invention；
Fig. 6 is Darknet-53 network diagram provided in an embodiment of the present invention；
Fig. 7 is a kind of specific embodiment structure chart of moving target behavior tracking device provided in an embodiment of the present invention.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, with reference to the accompanying drawings and detailed description
The present invention is described in further detail.Obviously, described embodiments are only a part of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
The description and claims of this application and term " first ", " second ", " third " " in above-mentioned attached drawing
Four " etc. be for distinguishing different objects, rather than for describing specific sequence.Furthermore term " includes " and " having " and
Their any deformations, it is intended that cover and non-exclusive include.Such as contain a series of steps or units process, method,
System, product or equipment are not limited to listed step or unit, but may include the step of not listing or unit.
Present inventor has found after study, the fast speed of YOLOv3 algorithm process picture, and YOLOv3 algorithm
Transplanting is convenient, can realize under each operating system, be relatively low to the configuration requirement of hardware, can be easier light
It is realized in magnitude equipment.
YOLOv3 algorithm passes through feature extraction network first and extracts feature to input picture, obtains the spy of certain size size
Sign figure, such as 13*13, are then divided into 13*13 grid cell (grid cell) for input picture, if ground truth
Which grid cell the centre coordinate of some object (main body) falls in in (labeled data), then just by the grid cell
The object is predicted, because each grid cell can predict the bounding box of fixed quantity (in YOLO v3 algorithm
It is 3, the initial size of these bounding box is different), it is only maximum with the IOU of ground truth
Bounding box is only for predicting the object's.It can be seen that dimension is to mention there are two the output characteristic patterns that prediction obtains
The dimension for the feature got, such as 13*13 are B* (5+C) there are one dimension (depth), and wherein B indicates each grid cell
The quantity of the bounding box of prediction, for example be 2 in YOLO v1, it is 5 in YOLO v2, is 3 in YOLO v3, C table
Show the classification number of bounding box, 5 indicate 4 coordinate informations and a confidence level (objectness score).
With popularizing for recorded broadcast instrument, the approach of acquiring video information also becomes to be relatively easy to, and can then be easier to obtain
Instructional video resource.By controlling the tracking of teacher the movement of video recorder when recorded broadcast, achieves completely unmanned control and take the photograph
Classroom video is recorded in the state of camera.Relatively high occasion is required in this real-time detection, YOLOv3 algorithm is likely to be breached
Quick detection effect.By the adjusting of parameter, model optimization makes it that personage in video is quickly identified and be tracked, and
And guarantee to reach certain precision
After describing the technical solution of the embodiment of the present invention, the various non-limiting realities of detailed description below the application
Apply mode.
Referring first to Fig. 1, Fig. 1 is a kind of process signal of moving target behavior tracking method provided in an embodiment of the present invention
Figure, the embodiment of the present invention may include the following contents:
S101: movement destination image is obtained.
S102: calling the target detection model constructed in advance, and moving target position is positioned in movement destination image.
Movement destination image is the moving target image during the motion captured, and corresponding video is sayed, continuous capturing
Multiple images can reflect that moving target in the motion state of this period, positions moving target, Bian Keshi in every frame image
The tracing detection of existing moving target.
Target detection model for based on YOLOv3 algorithm, use regularization and two-value cross entropy for loss function training institute
?；Target detection model meets the object boundary frame of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm, and
The centre coordinate of object boundary frame is determined, to position the position of moving target.
It hands over and than (Intersection-over-Union, IoU), to refer to the candidate frame generated in target detection
The overlapping rate of (candidate bound) and former indicia framing (ground truth bound), i.e. their intersection and union
Ratio.
Non-maxima suppression (non maximum suppression) method, to generate detection based on object detection score
Frame, the highest detection block of score is selected, other have the detection block of obvious overlapping to be suppressed with selected detection block.The process quilt
It is constantly recursive to be applied to remaining detection block.Specific step is as follows:
Framed score is sorted, best result and its corresponding frame are chosen.
Remaining frame is traversed, if the overlapping area (IOU) with current best result frame is greater than certain threshold value, is just deleted frame
It removes.
Continue to select a highest scoring from untreated frame, repeat the above process.
The loss function of target detection model can are as follows:
In formula, L is that two-value intersects loss function, and R is the loss function after regularization, ziFor predicted value, yiFor desired value,
ωiFor the weighted value of each layer.
Two-value cross entropy after using Regularization is as loss function, by the training dataset provided, examines the target
Model learning is surveyed to the function that can identify object features, different images is tested, can check its accuracy rate.
In technical solution provided in an embodiment of the present invention, using target detection iconic model by moving target in acquisition
It is positioned in movement destination image, for the multiframe consecutive image continuously obtained, namely for video, by every frame image
Moving target is positioned, to realize the tracing detection to moving target behavior in video.Due to YOLOv3 algorithm process picture
Speed quickly, under the same conditions, is based on YOLOv3 algorithm training objective detection model to image processing speed than existing volume
The processing speed of the model (1000 times such as faster than R-CNN, 100 times faster than Fast-RCNN) of product neural network algorithm training is fast.It adopts
It is loss function with regularization and two-value cross entropy, can further improve the detection speed and positional accuracy of moving target.This
Outside, the transplanting of YOLOv3 algorithm is convenient, can realize under each operating system, relatively low to the configuration requirement of terminal hardware,
The operation of target detection model can be easier realized in lightweight equipment.For the application scenarios of real-time recorded broadcast, this Shen
The accurate positionin and quickly identification to moving target feature please can be realized under identical precision, what raising was identified in video field
Speed and precision reduces the delay and Caton of recording and broadcasting system.
After obtaining the optimal object boundary frame of effect, wherein the best those skilled in the art of effect can be according to specific
Application scenarios are determined, and the application does not do any restriction to this.It can predict to obtain object by K-means algorithm and coordinate transform
The position of characteristic area central point and the length and width of bounding box.
K-means algorithm is the clustering algorithm based on distance, the evaluation index using distance as similitude, two objects
Distance it is closer, similarity is bigger.The algorithm think cluster by forming apart from close object, therefore it is compact obtaining
And independent cluster is as final goal.Specific method can be as follows:
Randomly choose from the set of data points of input at one o'clock as first cluster centre.
For each of data set point x, calculate it and nearest cluster centre (referring to selected cluster centre) away from
From D (x), select a new data point as new cluster centre, the principle selected is the biggish point of D (x), is selected conduct
The probability of cluster centre is larger, repeats the above steps, and obtains k cluster centre until choosing.
Using this k initial cluster centres come the K-means algorithm of operation standard.
What the anchor frame (anchor boxes) of the application can be obtained by the method that K-means is clustered, with Faster-
RCNN is it is preferred that priori frame is different.YOLOv3 algorithm network is for 4 coordinates of each Boundary Prediction, tx, ty, tw, thCell
From the upper left angular variation (c of Fig. 2x；cy), and the height that has of frame before and width pw、ph, then predicted value respectively corresponds
In please referring to shown in Fig. 3, solid box is the block diagram of moving target characteristic area prediction, i.e. anchor frame, dotted line frame is true side
Frame:
bx=δ (tx)+cx；
by=δ (ty)+cy；
In formula, δ () is logistic function, is used for coordinate tx、tyIt normalizes between 0-1, bh、bwFor the side of prediction
The height value and width value of boundary's frame, bx、byCentered on put position coordinates.
At above-mentioned 4 coordinate values of YOLOv3 algorithm network training (height, width and the center position coordinate of bounding box)
When, can be used quadratic sum range error loss, so as to quickly calculate error.
The application target detection model to each frame by logistic regression predict an object score, if this
Prediction frame is largely overlapped relative to true frame value and gets well than other predicted values, then this value is exactly 1
Judge it for object features region.If lap does not reach a threshold value, for example, can given threshold be 0.5, then this
The frame of a prediction will be ignored, and can then be shown as no penalty values.
Optionally, feature pyramid network (FPN) can be used and obtain 3 kinds of different predictions by target of object features area image
Frame is realized across scale prediction, is please referred to shown in Fig. 4.
Three frame values of every kind of scale prediction, the design method of anchor can obtain 9 cluster centres with cluster to get to 9
Cluster.It is averagely given to 3 kinds of scales by size again.Scale 1: some convolutional layers are added after basic network and export frame letter again
Breath；Scale 2: from the convolutional layer of the layer second from the bottom in scale 1 up-sampling again with the characteristic pattern phase of the last one 16 × 16 size
Add, then frame information is exported by multiple convolutional layer, becomes larger twice compared to scale 1；Scale 3: it is similar with scale 2 using 32 ×
The characteristic pattern of 32 sizes.As shown, in 3 different size prediction frames, the Feature Selection Model that target detection model uses
By being changed on FPN (feature pyramid network) network, finally prediction obtains the tensor of one 3 dimension, includes
Bounding box information, object information and class predictive information (such as label classification belonging to moving target, in recording and broadcasting system,
Student or teacher can be shown on bounding box).
Object features zone marker good frame coordinate value and the corresponding label of every picture are inputted, with Darknet-53 net
Based on network structure (wherein, the structural model of Darket-53 can be as shown in Figure 6), it is constantly trained, realizes cancer
Region Feature Extraction, classification.
It again include residual error network structure in Darknet-53 network structure.
Residual error network has two layers, and expression formula is as follows, and wherein σ represents nonlinear function ReLU.
F=W2σ(W1x)；
In formula, x is the input value of current layer, and W indicates i-th layer of weighted value, W1For first layer weighted value, F is that network is defeated
Result out.
Then by shortcut and the 2nd ReLU, output y is obtained:
Y=F (x, { Wi})+x。
In formula, F (x, { Wi) it is the output of a upper layer network as a result, x is the input value of current layer.
When need to output and input dimension be changed when (as change number of active lanes), can be in shortcut to x
It is a linear transformation Ws, such as following formula, however experiments have shown that x is enough, it does not need to do a dimension transformation again.
Y=F (x, { Wi})+Wsx。
In formula, WsFor s layers of weighted value, x is the input value of current layer, F (x, { Wi) be a upper layer network output knot
Fruit.
It is demonstrated experimentally that residual block generally requires two layers or more, residual block y=F (x, { W of simple layeri)+x can not rise
To castering action.Residual error network can solve the problems, such as to degenerate, and on training set and checksum set, the deeper network all demonstrated is wrong
Accidentally rate is smaller.Specific structure is as shown in Figure 5.
An object of the application detection model can be used one 53 layers of convolutional network, this network be superimposed by residual unit and
At.By test of many times, it was demonstrated that in the balance in classification accuracy with efficiency, this model ratio ResNet-101, ResNet-
152 and Darknet-19 shows more preferably.
In order to make those skilled in the art that presently filed embodiment be more clearly understood, present invention also provides a tools
The example of body, specifically can include:
Cutting image is inputted into target detection model and writes comments on a document the label of completion, and image is splitted into final characteristic pattern
Multiple grids.
According to IOU value, the bounding box and center point coordinate of input picture are predicted.
By predicted value and label comparison after, the target fractional of generation in conjunction with classification confidence level NMS principle to bounding box
Screening, constantly executes the step, until effect is best.
It repeats the above steps until generating the prediction coordinate of three different scales.
After carrying out above-mentioned processing to image, export actually detected image result, i.e., it will fortune in movement destination image
Moving-target is outlined with bounding box come the just label belonging to standard on bounding box.
FPS (Frames Per Second)=7.96 can be obtained by experiment, it is seen that speed is than existing Faster-RCNN
It is fast very much.
The embodiment of the present invention provides corresponding realization device also directed to moving target behavior tracking method, further such that
The method has more practicability.Moving target behavior tracking device provided in an embodiment of the present invention is introduced below, under
The moving target behavior tracking device of text description can correspond to each other reference with above-described moving target behavior tracking method.
Referring to Fig. 7, Fig. 7 is moving target behavior tracking device provided in an embodiment of the present invention in a kind of specific embodiment
Under structure chart, the device can include:
Image collection module 701, for obtaining movement destination image；
Target locating module 702 positions fortune for calling the target detection model constructed in advance in movement destination image
Moving-target position；Target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for loss function
Training gained；Target detection model meets the object boundary of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm
Frame, and determine the centre coordinate of object boundary frame, to position the position of moving target.
It optionally, is target detection model in the target locating module 702 in some embodiments of the present embodiment
Loss function be following formula module:
In formula, L is that two-value intersects loss function, and R is the loss function after regularization, ziFor predicted value, yiFor desired value,
For the weighted value of each layer.
In addition, the target locating module 702 is the convolutional network that target detection model uses 53 layers, and by multiple residual errors
The module that unit is formed by stacking.
Specifically, the target locating module 702 can also extract institute using feature pyramid network for target detection model
The feature of movement destination image is stated, the object detection results of output include the class label of bounding box information and the moving target
Module.
Optionally, the target locating module 702 can also predict the fortune using K-means algorithm for target detection model
The center position coordinate of the characteristic area of moving-target and the length value and width value of bounding box obtain module.
Further, the target locating module 702 can include also for example three different scales for object detection results
The module of bounding box and the corresponding center position coordinate of each bounding box.
Optionally, the target locating module 702 for example can also be damaged for target detection model using quadratic sum range error
The center position coordinate of characteristic area of the mistake method training moving target and the mould of the length value of bounding box and width value
Block.
The function of each functional module of moving target behavior tracking device can be according to the above method described in the embodiment of the present invention
Method specific implementation in embodiment, specific implementation process are referred to the associated description of above method embodiment, herein not
It repeats again.
From the foregoing, it will be observed that the embodiment of the present invention can realize the tracing detection to moving target behavior in video.
The embodiment of the invention also provides a kind of moving target behavior tracking equipment, specifically can include:
Memory, for storing computer program；
Processor realizes moving target behavior tracking side described in any one embodiment as above for executing computer program
The step of method.
The function of each functional module of moving target behavior tracking equipment can be according to the above method described in the embodiment of the present invention
Method specific implementation in embodiment, specific implementation process are referred to the associated description of above method embodiment, herein not
It repeats again.
From the foregoing, it will be observed that the embodiment of the present invention can realize the tracing detection to moving target behavior in video.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored with moving target behavior tracking journey
Sequence, moving target behavior tracking described in any one embodiment as above when the moving target behavior tracking program is executed by processor
The step of method.
The function of each functional module of computer readable storage medium described in the embodiment of the present invention can be according to above method reality
The method specific implementation in example is applied, specific implementation process is referred to the associated description of above method embodiment, herein no longer
It repeats.
From the foregoing, it will be observed that the embodiment of the present invention can realize the tracing detection to moving target behavior in video.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with it is other
The difference of embodiment, same or similar part may refer to each other between each embodiment.For being filled disclosed in embodiment
For setting, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is referring to method part
Explanation.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
The step of method described in conjunction with the examples disclosed in this document or algorithm, can directly be held with hardware, processor
The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
It to a kind of moving target behavior tracking method, apparatus provided by the present invention, equipment and computer-readable deposits above
Storage media is described in detail.It is used herein that a specific example illustrates the principle and implementation of the invention,
The above description of the embodiment is only used to help understand the method for the present invention and its core ideas.It should be pointed out that for this technology
For the those of ordinary skill in field, without departing from the principle of the present invention, several improvement can also be carried out to the present invention
And modification, these improvements and modifications also fall within the scope of protection of the claims of the present invention.
Claims (10)
1. a kind of moving target behavior tracking method characterized by comprising
Obtain movement destination image；
The target detection model constructed in advance is called, moving target position is positioned in the movement destination image；
Wherein, the target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for loss function instruction
Practice gained；The target detection model meets the target side of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm
Boundary's frame, and determine the center position coordinate of the moving target characteristic area, to position the position of the moving target.
2. moving target behavior tracking method according to claim 1, which is characterized in that the damage of the target detection model
Lose function are as follows:
In formula, L is that two-value intersects loss function, and R is the loss function after regularization, ziFor predicted value, yiFor desired value, ωiFor
The weighted value of each layer.
3. moving target behavior tracking method according to claim 2, which is characterized in that the target detection model uses
53 layers of convolutional network, and be formed by stacking by multiple residual units.
4. according to claim 1 to moving target behavior tracking method described in 3 any one, which is characterized in that the target
Detection model extracts the feature of the movement destination image using feature pyramid network, and the object detection results of output include side
The class label of boundary frame information and the moving target.
5. moving target behavior tracking method according to claim 4, which is characterized in that the target detection model uses
K-means algorithm predicts the center position coordinate of the characteristic area of the moving target and the length value and width of bounding box
Value.
6. moving target behavior tracking method according to claim 5, which is characterized in that the object detection results include
The bounding box and the corresponding center position coordinate of each bounding box of three different scales.
7. moving target behavior tracking method according to claim 6, which is characterized in that the target detection model uses
Quadratic sum range error loses the center position coordinate of the characteristic area of the method training moving target and the length of bounding box
Angle value and width value.
8. a kind of moving target behavior tracking device characterized by comprising
Image collection module, for obtaining movement destination image；
Target locating module positions movement for calling the target detection model constructed in advance in the movement destination image
Target position；The target detection model is based on YOLOv3 algorithm, uses regularization and two-value cross entropy for loss letter
Number training gained；The target detection model meets the mesh of preset condition according to friendship and the screening of ratio and non-maxima suppression algorithm
Bounding box is marked, and determines the center position coordinate of the moving target characteristic area, to position the position of the moving target.
9. a kind of moving target behavior tracking equipment, which is characterized in that including processor, the processor is for executing memory
It is realized when the computer program of middle storage as described in any one of claim 1 to 7 the step of moving target behavior tracking method.
10. a kind of computer readable storage medium, which is characterized in that be stored with movement mesh on the computer readable storage medium
Behavior tracking program is marked, is realized when the moving target behavior tracking program is executed by processor such as any one of claim 1 to 7
The step of moving target behavior tracking method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810935776.0A CN109117794A (en) | 2018-08-16 | 2018-08-16 | A kind of moving target behavior tracking method, apparatus, equipment and readable storage medium storing program for executing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810935776.0A CN109117794A (en) | 2018-08-16 | 2018-08-16 | A kind of moving target behavior tracking method, apparatus, equipment and readable storage medium storing program for executing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109117794A true CN109117794A (en) | 2019-01-01 |
Family
ID=64853282
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810935776.0A Pending CN109117794A (en) | 2018-08-16 | 2018-08-16 | A kind of moving target behavior tracking method, apparatus, equipment and readable storage medium storing program for executing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109117794A (en) |
Cited By (60)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109613934A (en) * | 2019-01-09 | 2019-04-12 | 天津大学 | A kind of method that unmanned plane captures black winged unmanned plane |
| CN109815886A (en) * | 2019-01-21 | 2019-05-28 | 南京邮电大学 | A kind of pedestrian and vehicle checking method and system based on improvement YOLOv3 |
| CN109828578A (en) * | 2019-02-22 | 2019-05-31 | 南京天创电子技术有限公司 | A kind of instrument crusing robot optimal route planing method based on YOLOv3 |
| CN109871829A (en) * | 2019-03-15 | 2019-06-11 | 北京行易道科技有限公司 | A kind of detection model training method and device based on deep learning |
| CN109886085A (en) * | 2019-01-03 | 2019-06-14 | 四川弘和通讯有限公司 | People counting method based on deep learning target detection |
| CN109886286A (en) * | 2019-01-03 | 2019-06-14 | 武汉精测电子集团股份有限公司 | Object detection method, target detection model and system based on cascade detectors |
| CN109903310A (en) * | 2019-01-23 | 2019-06-18 | 平安科技（深圳）有限公司 | Method for tracking target, device, computer installation and computer storage medium |
| CN109919008A (en) * | 2019-01-23 | 2019-06-21 | 平安科技（深圳）有限公司 | Moving target detecting method, device, computer equipment and storage medium |
| CN109919025A (en) * | 2019-01-30 | 2019-06-21 | 华南理工大学 | Video scene Method for text detection, system, equipment and medium based on deep learning |
| CN109948617A (en) * | 2019-03-29 | 2019-06-28 | 南京邮电大学 | A kind of invoice image position method |
| CN110072064A (en) * | 2019-06-12 | 2019-07-30 | 睿魔智能科技（深圳）有限公司 | A kind of image capturing method, device, equipment and storage medium |
| CN110110650A (en) * | 2019-05-02 | 2019-08-09 | 西安电子科技大学 | Face identification method in pedestrian |
| CN110110684A (en) * | 2019-05-14 | 2019-08-09 | 深圳供电局有限公司 | Exotic recognition methods, device and the computer equipment of power transmission line equipment |
| CN110135266A (en) * | 2019-04-17 | 2019-08-16 | 浙江理工大学 | A kind of dual camera electrical fire preventing control method and system based on deep learning |
| CN110223320A (en) * | 2019-05-08 | 2019-09-10 | 北京百度网讯科技有限公司 | Object detection tracking and detecting and tracking device |
| CN110232350A (en) * | 2019-06-10 | 2019-09-13 | 哈尔滨工程大学 | A kind of real-time water surface multiple mobile object detecting and tracking method based on on-line study |
| CN110246180A (en) * | 2019-05-17 | 2019-09-17 | 福州瑞芯微电子股份有限公司 | A kind of teacher's method for detecting position and system based on deep learning |
| CN110287856A (en) * | 2019-06-21 | 2019-09-27 | 上海闪马智能科技有限公司 | A kind of security personnel's behavior analysis system, method and device |
| CN110322478A (en) * | 2019-06-10 | 2019-10-11 | 广州视源电子科技股份有限公司 | Characteristic point watch window processing method, tracking, device, equipment and medium |
| CN110363100A (en) * | 2019-06-24 | 2019-10-22 | 昆明理工大学 | A kind of video object detection method based on YOLOv3 |
| CN110398226A (en) * | 2019-05-21 | 2019-11-01 | 重庆邮电大学 | A kind of monocular vision distance measuring method for advanced DAS (Driver Assistant System) |
| CN110472572A (en) * | 2019-08-14 | 2019-11-19 | 西北工业大学 | The quick identification and classification method of naval target under a kind of complex environment |
| CN110490148A (en) * | 2019-08-22 | 2019-11-22 | 四川自由健信息科技有限公司 | A kind of recognition methods for behavior of fighting |
| CN110610159A (en) * | 2019-09-16 | 2019-12-24 | 天津通卡智能网络科技股份有限公司 | Real-time bus passenger flow volume statistical method |
| CN110738686A (en) * | 2019-10-12 | 2020-01-31 | 四川航天神坤科技有限公司 | Static and dynamic combined video man-vehicle detection method and system |
| CN110766894A (en) * | 2019-12-25 | 2020-02-07 | 恒大智慧科技有限公司 | Community fence crossing early warning method, system, server and computer storage medium |
| CN110807886A (en) * | 2019-10-28 | 2020-02-18 | 恒大智慧科技有限公司 | Community security early warning method and system |
| CN110866473A (en) * | 2019-11-04 | 2020-03-06 | 浙江大华技术股份有限公司 | Target object tracking detection method and device, storage medium and electronic device |
| CN110889324A (en) * | 2019-10-12 | 2020-03-17 | 南京航空航天大学 | Thermal infrared image target identification method based on YOLO V3 terminal-oriented guidance |
| CN111008994A (en) * | 2019-11-14 | 2020-04-14 | 山东万腾电子科技有限公司 | Moving target real-time detection and tracking system and method based on MPSoC |
| CN111008595A (en) * | 2019-12-05 | 2020-04-14 | 武汉大学 | Private car interior rear row baby/pet groveling window distinguishing and car interior atmosphere identifying method |
| CN111104965A (en) * | 2019-11-25 | 2020-05-05 | 河北科技大学 | Vehicle target identification method and device |
| CN111144236A (en) * | 2019-12-10 | 2020-05-12 | 华南师范大学 | Method, system and storage medium for analyzing mating behavior of cockroach |
| CN111144273A (en) * | 2019-12-24 | 2020-05-12 | 苏州奥易克斯汽车电子有限公司 | Non-motor vehicle detection method |
| CN111160248A (en) * | 2019-12-30 | 2020-05-15 | 北京每日优鲜电子商务有限公司 | Method and device for tracking articles, computer equipment and storage medium |
| CN111476827A (en) * | 2019-01-24 | 2020-07-31 | 曜科智能科技（上海）有限公司 | Target tracking method, system, electronic device and storage medium |
| CN111476817A (en) * | 2020-02-27 | 2020-07-31 | 浙江工业大学 | Multi-target pedestrian detection tracking method based on yolov3 |
| WO2020164282A1 (en) * | 2019-02-14 | 2020-08-20 | 平安科技（深圳）有限公司 | Yolo-based image target recognition method and apparatus, electronic device, and storage medium |
| WO2020177432A1 (en) * | 2019-03-07 | 2020-09-10 | 中国科学院自动化研究所 | Multi-tag object detection method and system based on target detection network, and apparatuses |
| CN111652903A (en) * | 2020-05-22 | 2020-09-11 | 重庆理工大学 | Pedestrian target tracking method based on convolution correlation network in automatic driving scene |
| CN111652835A (en) * | 2020-02-12 | 2020-09-11 | 天津大学 | Method for detecting insulator loss of power transmission line based on deep learning and clustering |
| WO2020206861A1 (en) * | 2019-04-08 | 2020-10-15 | 江西理工大学 | Yolo v3-based detection method for key object at transportation junction |
| CN111898511A (en) * | 2020-07-23 | 2020-11-06 | 北京以萨技术股份有限公司 | High-altitude parabolic detection method, device and medium based on deep learning |
| CN112070085A (en) * | 2020-09-04 | 2020-12-11 | 中山大学 | Unmanned aerial vehicle multi-feature-point detection method and device based on double-stage connection deep network |
| CN112396628A (en) * | 2019-08-14 | 2021-02-23 | 上海开域信息科技有限公司 | Anti-track-jitter method based on target height prediction |
| CN112489020A (en) * | 2020-12-01 | 2021-03-12 | 济南博观智能科技有限公司 | Image screening method and device, electronic equipment and storage medium |
| CN112506342A (en) * | 2020-12-04 | 2021-03-16 | 郑州中业科技股份有限公司 | Man-machine interaction method and system based on dynamic gesture recognition |
| WO2021077868A1 (en) * | 2019-10-26 | 2021-04-29 | 华为技术有限公司 | Target detection method and device |
| CN112882612A (en) * | 2021-01-12 | 2021-06-01 | 京东方科技集团股份有限公司 | Display method, display equipment and display system |
| CN112907625A (en) * | 2021-02-05 | 2021-06-04 | 齐鲁工业大学 | Target following method and system applied to four-footed bionic robot |
| WO2021139049A1 (en) * | 2020-01-10 | 2021-07-15 | 深圳中兴网信科技有限公司 | Detection method, detection apparatus, monitoring device, and computer readable storage medium |
| CN113256683A (en) * | 2020-12-30 | 2021-08-13 | 深圳云天励飞技术股份有限公司 | Target tracking method and related equipment |
| US11094070B2 (en) * | 2019-04-23 | 2021-08-17 | Jiangnan University | Visual multi-object tracking based on multi-Bernoulli filter with YOLOv3 detection |
| CN113283279A (en) * | 2021-01-25 | 2021-08-20 | 广东技术师范大学 | Deep learning-based multi-target tracking method and device in video |
| CN113627295A (en) * | 2021-07-28 | 2021-11-09 | 中汽创智科技有限公司 | Image processing method, device, equipment and storage medium |
| CN113850264A (en) * | 2019-06-10 | 2021-12-28 | 创新先进技术有限公司 | Method and system for evaluating target detection model |
| CN114049377A (en) * | 2021-10-29 | 2022-02-15 | 哈尔滨工业大学 | Method and system for detecting high-dynamic small target in air |
| CN114067564A (en) * | 2021-11-15 | 2022-02-18 | 武汉理工大学 | Traffic condition comprehensive monitoring method based on YOLO |
| CN114387219A (en) * | 2021-12-17 | 2022-04-22 | 依未科技(北京)有限公司 | Method, device, medium and equipment for detecting arteriovenous cross compression characteristics of eyeground |
| CN115457036A (en) * | 2022-11-10 | 2022-12-09 | 中国平安财产保险股份有限公司 | Detection model training method, intelligent counting method and related equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150178570A1 (en) * | 2012-06-28 | 2015-06-25 | International Business Machines Corporation | Detection of static object on thoroughfare crossings |
| CN105163121A (en) * | 2015-08-24 | 2015-12-16 | 西安电子科技大学 | Large-compression-ratio satellite remote sensing image compression method based on deep self-encoding network |
| CN107423760A (en) * | 2017-07-21 | 2017-12-01 | 西安电子科技大学 | Based on pre-segmentation and the deep learning object detection method returned |
| CN107527009A (en) * | 2017-07-11 | 2017-12-29 | 浙江汉凡软件科技有限公司 | A kind of remnant object detection method based on YOLO target detections |
| CN108090919A (en) * | 2018-01-02 | 2018-05-29 | 华南理工大学 | Improved kernel correlation filtering tracking method based on super-pixel optical flow and adaptive learning factor |
-
2018
- 2018-08-16 CN CN201810935776.0A patent/CN109117794A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150178570A1 (en) * | 2012-06-28 | 2015-06-25 | International Business Machines Corporation | Detection of static object on thoroughfare crossings |
| CN105163121A (en) * | 2015-08-24 | 2015-12-16 | 西安电子科技大学 | Large-compression-ratio satellite remote sensing image compression method based on deep self-encoding network |
| CN107527009A (en) * | 2017-07-11 | 2017-12-29 | 浙江汉凡软件科技有限公司 | A kind of remnant object detection method based on YOLO target detections |
| CN107423760A (en) * | 2017-07-21 | 2017-12-01 | 西安电子科技大学 | Based on pre-segmentation and the deep learning object detection method returned |
| CN108090919A (en) * | 2018-01-02 | 2018-05-29 | 华南理工大学 | Improved kernel correlation filtering tracking method based on super-pixel optical flow and adaptive learning factor |
Non-Patent Citations (2)
| Title |
|---|
| JOSEPH REDMON等: "YOLOv3: An Incremental Improvement", 《COMPUTER VISION AND PATTERN RECOGNITION》 * |
| 凌超等: "视频图像中运动目标跟踪算法研究综述", 《信息技术》 * |
Cited By (80)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109886085A (en) * | 2019-01-03 | 2019-06-14 | 四川弘和通讯有限公司 | People counting method based on deep learning target detection |
| CN109886286A (en) * | 2019-01-03 | 2019-06-14 | 武汉精测电子集团股份有限公司 | Object detection method, target detection model and system based on cascade detectors |
| CN109613934A (en) * | 2019-01-09 | 2019-04-12 | 天津大学 | A kind of method that unmanned plane captures black winged unmanned plane |
| CN109815886A (en) * | 2019-01-21 | 2019-05-28 | 南京邮电大学 | A kind of pedestrian and vehicle checking method and system based on improvement YOLOv3 |
| CN109903310A (en) * | 2019-01-23 | 2019-06-18 | 平安科技（深圳）有限公司 | Method for tracking target, device, computer installation and computer storage medium |
| CN109919008A (en) * | 2019-01-23 | 2019-06-21 | 平安科技（深圳）有限公司 | Moving target detecting method, device, computer equipment and storage medium |
| CN111476827B (en) * | 2019-01-24 | 2024-02-02 | 曜科智能科技（上海）有限公司 | Target tracking method, system, electronic device and storage medium |
| CN111476827A (en) * | 2019-01-24 | 2020-07-31 | 曜科智能科技（上海）有限公司 | Target tracking method, system, electronic device and storage medium |
| CN109919025A (en) * | 2019-01-30 | 2019-06-21 | 华南理工大学 | Video scene Method for text detection, system, equipment and medium based on deep learning |
| WO2020164282A1 (en) * | 2019-02-14 | 2020-08-20 | 平安科技（深圳）有限公司 | Yolo-based image target recognition method and apparatus, electronic device, and storage medium |
| CN109828578A (en) * | 2019-02-22 | 2019-05-31 | 南京天创电子技术有限公司 | A kind of instrument crusing robot optimal route planing method based on YOLOv3 |
| CN109828578B (en) * | 2019-02-22 | 2020-06-16 | 南京天创电子技术有限公司 | Instrument inspection robot optimal route planning method based on YOLOv3 |
| US11429818B2 (en) | 2019-03-07 | 2022-08-30 | Institute Of Automation, Chinese Academy Of Sciences | Method, system and device for multi-label object detection based on an object detection network |
| WO2020177432A1 (en) * | 2019-03-07 | 2020-09-10 | 中国科学院自动化研究所 | Multi-tag object detection method and system based on target detection network, and apparatuses |
| CN109871829A (en) * | 2019-03-15 | 2019-06-11 | 北京行易道科技有限公司 | A kind of detection model training method and device based on deep learning |
| CN109948617A (en) * | 2019-03-29 | 2019-06-28 | 南京邮电大学 | A kind of invoice image position method |
| WO2020206861A1 (en) * | 2019-04-08 | 2020-10-15 | 江西理工大学 | Yolo v3-based detection method for key object at transportation junction |
| CN110135266A (en) * | 2019-04-17 | 2019-08-16 | 浙江理工大学 | A kind of dual camera electrical fire preventing control method and system based on deep learning |
| US11094070B2 (en) * | 2019-04-23 | 2021-08-17 | Jiangnan University | Visual multi-object tracking based on multi-Bernoulli filter with YOLOv3 detection |
| CN110110650A (en) * | 2019-05-02 | 2019-08-09 | 西安电子科技大学 | Face identification method in pedestrian |
| CN110223320A (en) * | 2019-05-08 | 2019-09-10 | 北京百度网讯科技有限公司 | Object detection tracking and detecting and tracking device |
| CN110110684A (en) * | 2019-05-14 | 2019-08-09 | 深圳供电局有限公司 | Exotic recognition methods, device and the computer equipment of power transmission line equipment |
| CN110246180A (en) * | 2019-05-17 | 2019-09-17 | 福州瑞芯微电子股份有限公司 | A kind of teacher's method for detecting position and system based on deep learning |
| CN110398226A (en) * | 2019-05-21 | 2019-11-01 | 重庆邮电大学 | A kind of monocular vision distance measuring method for advanced DAS (Driver Assistant System) |
| CN113850264A (en) * | 2019-06-10 | 2021-12-28 | 创新先进技术有限公司 | Method and system for evaluating target detection model |
| CN110232350B (en) * | 2019-06-10 | 2022-12-13 | 哈尔滨工程大学 | Real-time water surface multi-moving-object detection and tracking method based on online learning |
| CN110322478B (en) * | 2019-06-10 | 2021-09-07 | 广州视源电子科技股份有限公司 | Feature point observation window processing method, tracking method, device, equipment and medium |
| CN110232350A (en) * | 2019-06-10 | 2019-09-13 | 哈尔滨工程大学 | A kind of real-time water surface multiple mobile object detecting and tracking method based on on-line study |
| CN110322478A (en) * | 2019-06-10 | 2019-10-11 | 广州视源电子科技股份有限公司 | Characteristic point watch window processing method, tracking, device, equipment and medium |
| CN110072064A (en) * | 2019-06-12 | 2019-07-30 | 睿魔智能科技（深圳）有限公司 | A kind of image capturing method, device, equipment and storage medium |
| WO2020248396A1 (en) * | 2019-06-12 | 2020-12-17 | 睿魔智能科技（深圳）有限公司 | Method, apparatus, and device for image capture, and storage medium |
| US11736800B2 (en) | 2019-06-12 | 2023-08-22 | Remo Tech Co., Ltd. | Method, apparatus, and device for image capture, and storage medium |
| CN110287856A (en) * | 2019-06-21 | 2019-09-27 | 上海闪马智能科技有限公司 | A kind of security personnel's behavior analysis system, method and device |
| CN110363100A (en) * | 2019-06-24 | 2019-10-22 | 昆明理工大学 | A kind of video object detection method based on YOLOv3 |
| CN112396628B (en) * | 2019-08-14 | 2024-05-17 | 上海开域信息科技有限公司 | Track jitter resistance method based on target height prediction |
| CN110472572A (en) * | 2019-08-14 | 2019-11-19 | 西北工业大学 | The quick identification and classification method of naval target under a kind of complex environment |
| CN112396628A (en) * | 2019-08-14 | 2021-02-23 | 上海开域信息科技有限公司 | Anti-track-jitter method based on target height prediction |
| CN110490148A (en) * | 2019-08-22 | 2019-11-22 | 四川自由健信息科技有限公司 | A kind of recognition methods for behavior of fighting |
| CN110610159A (en) * | 2019-09-16 | 2019-12-24 | 天津通卡智能网络科技股份有限公司 | Real-time bus passenger flow volume statistical method |
| CN110738686A (en) * | 2019-10-12 | 2020-01-31 | 四川航天神坤科技有限公司 | Static and dynamic combined video man-vehicle detection method and system |
| CN110738686B (en) * | 2019-10-12 | 2022-12-02 | 四川航天神坤科技有限公司 | Static and dynamic combined video man-vehicle detection method and system |
| CN110889324A (en) * | 2019-10-12 | 2020-03-17 | 南京航空航天大学 | Thermal infrared image target identification method based on YOLO V3 terminal-oriented guidance |
| WO2021077868A1 (en) * | 2019-10-26 | 2021-04-29 | 华为技术有限公司 | Target detection method and device |
| CN110807886A (en) * | 2019-10-28 | 2020-02-18 | 恒大智慧科技有限公司 | Community security early warning method and system |
| CN110866473B (en) * | 2019-11-04 | 2022-11-18 | 浙江大华技术股份有限公司 | Target object tracking detection method and device, storage medium and electronic device |
| CN110866473A (en) * | 2019-11-04 | 2020-03-06 | 浙江大华技术股份有限公司 | Target object tracking detection method and device, storage medium and electronic device |
| CN111008994A (en) * | 2019-11-14 | 2020-04-14 | 山东万腾电子科技有限公司 | Moving target real-time detection and tracking system and method based on MPSoC |
| CN111104965A (en) * | 2019-11-25 | 2020-05-05 | 河北科技大学 | Vehicle target identification method and device |
| CN111008595A (en) * | 2019-12-05 | 2020-04-14 | 武汉大学 | Private car interior rear row baby/pet groveling window distinguishing and car interior atmosphere identifying method |
| CN111144236B (en) * | 2019-12-10 | 2024-04-26 | 华南师范大学 | Cockroach mating behavior analysis method, system and storage medium |
| CN111144236A (en) * | 2019-12-10 | 2020-05-12 | 华南师范大学 | Method, system and storage medium for analyzing mating behavior of cockroach |
| CN111144273A (en) * | 2019-12-24 | 2020-05-12 | 苏州奥易克斯汽车电子有限公司 | Non-motor vehicle detection method |
| CN110766894A (en) * | 2019-12-25 | 2020-02-07 | 恒大智慧科技有限公司 | Community fence crossing early warning method, system, server and computer storage medium |
| CN111160248A (en) * | 2019-12-30 | 2020-05-15 | 北京每日优鲜电子商务有限公司 | Method and device for tracking articles, computer equipment and storage medium |
| WO2021139049A1 (en) * | 2020-01-10 | 2021-07-15 | 深圳中兴网信科技有限公司 | Detection method, detection apparatus, monitoring device, and computer readable storage medium |
| CN111652835A (en) * | 2020-02-12 | 2020-09-11 | 天津大学 | Method for detecting insulator loss of power transmission line based on deep learning and clustering |
| CN111476817A (en) * | 2020-02-27 | 2020-07-31 | 浙江工业大学 | Multi-target pedestrian detection tracking method based on yolov3 |
| CN111652903B (en) * | 2020-05-22 | 2023-09-08 | 重庆理工大学 | Pedestrian target tracking method based on convolution association network in automatic driving scene |
| CN111652903A (en) * | 2020-05-22 | 2020-09-11 | 重庆理工大学 | Pedestrian target tracking method based on convolution correlation network in automatic driving scene |
| CN111898511A (en) * | 2020-07-23 | 2020-11-06 | 北京以萨技术股份有限公司 | High-altitude parabolic detection method, device and medium based on deep learning |
| CN112070085B (en) * | 2020-09-04 | 2023-07-28 | 中山大学 | Unmanned aerial vehicle multi-feature point detection method and device based on two-stage cascade depth network |
| CN112070085A (en) * | 2020-09-04 | 2020-12-11 | 中山大学 | Unmanned aerial vehicle multi-feature-point detection method and device based on double-stage connection deep network |
| CN112489020B (en) * | 2020-12-01 | 2023-05-23 | 济南博观智能科技有限公司 | Image screening method and device, electronic equipment and storage medium |
| CN112489020A (en) * | 2020-12-01 | 2021-03-12 | 济南博观智能科技有限公司 | Image screening method and device, electronic equipment and storage medium |
| CN112506342B (en) * | 2020-12-04 | 2022-01-28 | 郑州中业科技股份有限公司 | Man-machine interaction method and system based on dynamic gesture recognition |
| CN112506342A (en) * | 2020-12-04 | 2021-03-16 | 郑州中业科技股份有限公司 | Man-machine interaction method and system based on dynamic gesture recognition |
| CN113256683B (en) * | 2020-12-30 | 2024-03-29 | 深圳云天励飞技术股份有限公司 | Target tracking method and related equipment |
| CN113256683A (en) * | 2020-12-30 | 2021-08-13 | 深圳云天励飞技术股份有限公司 | Target tracking method and related equipment |
| CN112882612A (en) * | 2021-01-12 | 2021-06-01 | 京东方科技集团股份有限公司 | Display method, display equipment and display system |
| CN112882612B (en) * | 2021-01-12 | 2024-01-23 | 京东方科技集团股份有限公司 | Display method, display device and display system |
| CN113283279B (en) * | 2021-01-25 | 2024-01-19 | 广东技术师范大学 | Multi-target tracking method and device in video based on deep learning |
| CN113283279A (en) * | 2021-01-25 | 2021-08-20 | 广东技术师范大学 | Deep learning-based multi-target tracking method and device in video |
| CN112907625A (en) * | 2021-02-05 | 2021-06-04 | 齐鲁工业大学 | Target following method and system applied to four-footed bionic robot |
| CN113627295A (en) * | 2021-07-28 | 2021-11-09 | 中汽创智科技有限公司 | Image processing method, device, equipment and storage medium |
| CN114049377B (en) * | 2021-10-29 | 2022-06-10 | 哈尔滨工业大学 | Method and system for detecting high-dynamic small target in air |
| CN114049377A (en) * | 2021-10-29 | 2022-02-15 | 哈尔滨工业大学 | Method and system for detecting high-dynamic small target in air |
| CN114067564B (en) * | 2021-11-15 | 2023-08-29 | 武汉理工大学 | Traffic condition comprehensive monitoring method based on YOLO |
| CN114067564A (en) * | 2021-11-15 | 2022-02-18 | 武汉理工大学 | Traffic condition comprehensive monitoring method based on YOLO |
| CN114387219A (en) * | 2021-12-17 | 2022-04-22 | 依未科技(北京)有限公司 | Method, device, medium and equipment for detecting arteriovenous cross compression characteristics of eyeground |
| CN115457036A (en) * | 2022-11-10 | 2022-12-09 | 中国平安财产保险股份有限公司 | Detection model training method, intelligent counting method and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109117794A (en) | A kind of moving target behavior tracking method, apparatus, equipment and readable storage medium storing program for executing | |
| Movshovitz-Attias et al. | How useful is photo-realistic rendering for visual learning? | |
| CN108229478A (en) | Image, semantic segmentation and training method and device, electronic equipment, storage medium and program | |
| CN107609459A (en) | A kind of face identification method and device based on deep learning | |
| CN110503074A (en) | Information labeling method, apparatus, equipment and the storage medium of video frame | |
| CN109740420A (en) | Vehicle illegal recognition methods and Related product | |
| CN110321885A (en) | A kind of acquisition methods and device of point of interest | |
| CN113034495B (en) | Spine image segmentation method, medium and electronic device | |
| CN109558862A (en) | The people counting method and system of attention refinement frame based on spatial perception | |
| CN111738355A (en) | Image classification method and device with attention fused with mutual information and storage medium | |
| CN109145871A (en) | Psychology and behavior recognition methods, device and storage medium | |
| CN106023148A (en) | Star image point position extraction method under sequence focusing observation mode | |
| US20230053911A1 (en) | Detecting an object in an image using multiband and multidirectional filtering | |
| CN108230355A (en) | Target following and neural network training method, device, storage medium and electronic equipment | |
| CN110444058A (en) | Micro- class manufacturing system and micro- teaching system | |
| JPWO2020240808A1 (en) | Learning device, classification device, learning method, classification method, learning program, and classification program | |
| CN108229432A (en) | Face calibration method and device | |
| CN115186473A (en) | Scene perception modeling and verifying method based on parallel intelligence | |
| Johnson et al. | Detecting pianist hand posture mistakes for virtual piano tutoring | |
| Liu et al. | An improved method of identifying learner's behaviors based on deep learning | |
| Zhang | Innovation of English teaching model based on machine learning neural network and image super resolution | |
| CN104008397B (en) | Target tracking algorithm based on image set | |
| Fahmi et al. | Identification and Sorting of waste using artificial intelligence based on convolutional neural network | |
| Mubarak et al. | A Step Ahead Students CGPA Prediction Based on Support Vector Machines | |
| CN108229514A (en) | Object detecting method, device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190101 |
|
| RJ01 | Rejection of invention patent application after publication |