CN114494398B - Processing method and device of inclined target, storage medium and processor - Google Patents
Processing method and device of inclined target, storage medium and processor Download PDFInfo
- Publication number
- CN114494398B CN114494398B CN202210056957.2A CN202210056957A CN114494398B CN 114494398 B CN114494398 B CN 114494398B CN 202210056957 A CN202210056957 A CN 202210056957A CN 114494398 B CN114494398 B CN 114494398B
- Authority
- CN
- China
- Prior art keywords
- target
- density map
- determining
- original image
- image
- 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
- 238000003672 processing method Methods 0.000 title description 10
- 238000000034 method Methods 0.000 claims abstract description 40
- 238000010801 machine learning Methods 0.000 claims abstract description 9
- 230000006870 function Effects 0.000 claims description 16
- 239000003550 marker Substances 0.000 claims description 6
- 238000010586 diagram Methods 0.000 description 9
- 238000004590 computer program Methods 0.000 description 8
- 238000002372 labelling Methods 0.000 description 8
- 230000007547 defect Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The application provides a method and a device for processing an inclined target, a storage medium and a processor. The method comprises the following steps: acquiring a plurality of marked original images, wherein the original images comprise inclined targets; determining a position density map and a direction density map corresponding to each marked original image; constructing an artificial intelligent model, wherein the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the method comprises the steps of an original image, a position density map corresponding to the original image and a direction density map corresponding to the original image; inputting a target image to be predicted into an artificial intelligent model for operation, and outputting a target position density map and a target direction density map corresponding to the target image; and determining the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map. The scheme obtains the accurate inclination angle of the inclination target.
Description
Technical Field
The present application relates to the field of image processing, and in particular, to a method and apparatus for processing an oblique target, a storage medium, and a processor.
Background
Under the condition that a large number of cameras are used at present, the functions of face detection, pedestrian detection and the like are widely applied to cameras at a mobile end. In some scenarios, such as ship target detection, it is necessary to detect an inclined target. If the conventional horizontal rectangular frame representation method is used, it is difficult to accurately acquire angle information of the tilting target.
Disclosure of Invention
The application mainly aims to provide a processing method, a processing device, a storage medium and a processor for an inclined target, so as to solve the problem that in the prior art, the angle information of the inclined target is difficult to accurately acquire.
In order to achieve the above object, according to one aspect of the present application, there is provided a processing method of an oblique target, comprising: acquiring a plurality of marked original images, wherein the original images comprise inclined targets; determining a position density map and a direction density map corresponding to each marked original image, wherein the value of the position density map at a central point of the inclined target is different from the value of points except the central point in the original image, the value of the direction density map at a direction mark point is different from the value of points except the direction mark point in the original image, and the direction mark point is a point which is not the central point on the inclined target; constructing an artificial intelligent model, wherein the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, the position density map corresponding to the original image and the direction density map corresponding to the original image; inputting a target image to be predicted into the artificial intelligent model for operation, and outputting a target position density map and a target direction density map corresponding to the target image; and determining the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map.
Optionally, determining the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map includes: determining coordinates of a center point of an inclined target in the target image according to the target position density map; determining coordinates of a direction marking point of an inclined target in the target image according to the target direction density map; and determining the inclination angle of the inclined target in the target image according to the coordinates of the central point and the coordinates of the direction mark points.
Optionally, determining coordinates of a center point of the tilted object in the object image according to the object position density map includes: acquiring coordinates corresponding to a first maximum value of a local area of the target position density map; determining whether the first maximum value is greater than a first threshold value; and under the condition that the first maximum value is larger than the first threshold value, determining the coordinate corresponding to the first maximum value as the coordinate of the central point of the inclined target.
Optionally, determining coordinates of a direction marker point of the tilted object in the object image according to the object direction density map includes: acquiring a coordinate corresponding to a second maximum value of the target direction density map; determining whether the second maximum is greater than a second threshold; determining whether the coordinates corresponding to the second maximum value are within a neighborhood range of the coordinates of the central point of the inclined target; and determining that the coordinate corresponding to the second maximum value is the coordinate of the direction mark point under the condition that the second maximum value is larger than the second threshold value and the coordinate corresponding to the second maximum value is in the neighborhood range.
Optionally, the training data further includes a size density map corresponding to the original image, and the method further includes: extracting the center point coordinates of the inclined targets in the original image from the marked original image; extracting the size of an inclined target in the original image from the marked original image; extracting the inclination angle of an inclination target in the original image from the marked original image; generating a size density map corresponding to the original image according to the center point coordinates of the inclined target, the size of the inclined target and the inclined angle of the inclined target; and inputting the target image to be predicted into the artificial intelligent model for operation, and outputting a target size density map corresponding to the target image.
Optionally, the method further comprises: and determining the length information and the width information of the inclined target in the target image according to the target size density map.
Optionally, the target size density map is represented by a two-dimensional gaussian function, and determining length information and width information of an inclined target in the target image according to the target size density map includes: acquiring standard deviation of a first dimension and standard deviation of a second dimension of the two-dimensional Gaussian function corresponding to the target size density map; determining the length information according to the standard deviation of the first dimension; and determining the width information according to the standard deviation of the second dimension.
According to another aspect of the present application, there is provided a processing apparatus of an incline target, comprising: the acquisition unit is used for acquiring a plurality of marked original images, wherein the original images comprise inclined targets; a first determining unit, configured to determine a position density map and a direction density map corresponding to each of the noted original images, where the position density map has a value at a center point of the tilted object that is different from a value at a point other than the center point in the original image, and the direction density map has a value at a direction mark point that is different from a value at a point other than the direction mark point in the original image, and the direction mark point is a point on the tilted object that is not the center point; the building unit is used for building an artificial intelligent model, the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, the position density map corresponding to the original image and the direction density map corresponding to the original image; the computing unit is used for inputting a target image to be predicted into the artificial intelligent model for operation and outputting a target position density map and a target direction density map corresponding to the target image; and the second determining unit is used for determining the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map.
According to still another aspect of the present application, there is provided a computer readable storage medium including a stored program, wherein the program when run controls a device in which the computer readable storage medium is located to perform any one of the methods.
According to yet another aspect of the present application, there is provided a processor for running a program, wherein the program when run performs any one of the methods.
By applying the technical scheme of the application, an artificial intelligent model is obtained by training an original image, a position density map corresponding to the original image and a direction density map corresponding to the original image as training sets, then a target image to be predicted is input into the artificial intelligent model for operation, a target position density map and a target direction density map corresponding to the target image are output, and then the inclination angle of an inclined target in the target image is determined according to the target position density map and the target direction density map. In the scheme, the mode of model training by adopting the density map avoids the defects of the corner marking and the dip marking, so that the inclination angle of the inclination target in the determined target image is more accurate.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:
FIG. 1 shows a flow chart of a method of processing a tilted object according to an embodiment of the application;
Fig. 2 shows a schematic view of a processing apparatus of a tilting target according to an embodiment of the present application.
Detailed Description
It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other. The application will be described in detail below with reference to the drawings in connection with embodiments.
In order that those skilled in the art will better understand the present application, a technical solution in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate in order to describe the embodiments of the application herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It will be understood that when an element such as a layer, film, region, or substrate is referred to as being "on" another element, it can be directly on the other element or intervening elements may also be present. Furthermore, in the description and in the claims, when an element is described as being "connected" to another element, the element may be "directly connected" to the other element or "connected" to the other element through a third element.
As described in the background art, in order to solve the problem that it is difficult to accurately obtain the angle information of the inclined object in the prior art, the embodiments of the present application provide a method, an apparatus, a storage medium, and a processor for processing the inclined object.
According to an embodiment of the present application, there is provided a processing method of an inclined object.
Fig. 1 is a flowchart of a processing method of an inclined object according to an embodiment of the present application. As shown in fig. 1, the method comprises the steps of:
step S101, obtaining a plurality of marked original images, wherein the original images comprise inclined targets;
Step S102, determining a position density map and a direction density map corresponding to each marked original image, wherein the position density map is different from the value of a point except the center point in the original image at the center point of the inclined object, the direction density map is different from the value of a point except the direction mark point in the original image at the direction mark point, and the direction mark point is a point except the center point on the inclined object;
Step S103, constructing an artificial intelligent model, wherein the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, a position density map corresponding to the original image and a direction density map corresponding to the original image;
Step S104, inputting a target image to be predicted into the artificial intelligent model for operation, and outputting a target position density map and a target direction density map corresponding to the target image;
step S105, determining the inclination angle of the inclined object in the object image according to the object position density map and the object direction density map.
Specifically, the value of the position density map at the center point of the inclined object is different from the value of points other than the center point in the original image, that is, the information of the center point of the inclined object is hidden in the position density map;
Specifically, the direction density map has a value of a direction mark point different from a value of a point other than the direction mark point in the original image, the direction mark point being a point other than the center point on the tilt target, that is, the direction density map has information of the direction mark point hidden therein, and the direction mark point is a point for representing a tilt angle of the tilt target. For example, the center point of the incline target is taken as the center point of the ship, the direction mark point is taken as a point on the bow, and the line connecting the two points can represent the angle of incline of the ship.
Specifically, the plurality of original images as the training set are all annotated images.
Specifically, the tilting target includes a tilting object such as a tilting ship. One original image includes at least one tilting target therein.
Specifically, the artificial intelligence model may be a neural network model.
In the prior art, a common labeling mode of an inclined target frame is generally to add a rotation angle on the basis of a positive frame, but the labeling mode generally has a plurality of different numerical representations, so that the actual numerical result difference of a similar range frame can be larger, and the training of a model is not facilitated. The problem of labeling redundancy exists in labeling methods such as corner points, and a predicted labeling frame may not conform to priori knowledge during prediction. When the labeling methods such as dip angles are used, when the target deviation is smaller, the actual numerical value result of the labeling of the similar labeling frames is larger in difference, and model training is not facilitated. In the scheme, the mode of model training by adopting the density map avoids the defects of corner marking and dip marking.
In the scheme, an artificial intelligent model is obtained by training an original image, a position density map corresponding to the original image and a direction density map corresponding to the original image as training sets, then a target image to be predicted is input into the artificial intelligent model for operation, a target position density map and a target direction density map corresponding to the target image are output, and then the inclination angle of an inclined target in the target image is determined according to the target position density map and the target direction density map. In the scheme, the mode of model training by adopting the density map avoids the defects of the corner marking and the dip marking, so that the inclination angle of the inclination target in the determined target image is more accurate.
It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is illustrated in the flowcharts, in some cases the steps illustrated or described may be performed in an order other than that illustrated herein.
In an alternative embodiment of the present application, the position density map may be expressed as:
Wherein, Representing the value of a gaussian function centered on x p and having standard deviation σ p at position x, where σ p uses a constant value σ, and the value of x p is the center point coordinate of the tilted object, and Rl represents the set of all tilted objects, for example, three tilted objects in one original image. Of course, the positional density map of the present application is not limited to the form of equation 1.
In an alternative embodiment of the present application, the directional density map may be expressed as:
Wherein, Representing the value of a Gaussian function with x p as the center and standard deviation sigma p at the position x, wherein sigma p uses a fixed value sigma, the value of x p is the coordinate of a direction mark point, rl is the set of all direction mark points, and a plurality of points can be selected as the direction mark points. Of course, the direction density map of the present application is not limited to the form of formula 2.
In one embodiment of the present application, determining an inclination angle of an inclined object in the object image according to the object position density map and the object direction density map includes: determining coordinates of a center point of an inclined target in the target image according to the target position density map; determining coordinates of a direction marking point of an inclined target in the target image according to the target direction density map; and determining the inclination angle of the inclined target in the target image according to the coordinates of the central point and the coordinates of the direction mark point.
Specifically, determining the tilt angle of the tilt target in the target image according to the coordinates of the center point and the coordinates of the direction flag point includes: and acquiring the inclination angle of the connecting line of the central point and the direction mark point, and determining the inclination angle of the connecting line as the inclination angle of the inclination target. If the coordinates of the center point and the coordinates of the direction mark point are two-dimensional coordinates, the inclination angle of the inclined target can be determined according to the two-dimensional coordinates.
In a specific embodiment of the present application, determining coordinates of a center point of an inclined object in the object image according to the object position density map includes: acquiring coordinates corresponding to a first maximum value of a local area of the target position density map; determining whether the first maximum value is greater than a first threshold value; and determining that the coordinate corresponding to the first maximum value is the coordinate of the center point of the tilting target when the first maximum value is greater than the first threshold value. That is, the target position density map has the largest value at the center point, and since one target image to be predicted may have a plurality of inclined targets, for example, three ships in one map, the target position density map is divided into different areas to be processed, the coordinates corresponding to the first maximum value of the local area are extracted, and the coordinates corresponding to the first maximum value are determined as the coordinates of the center point of the inclined target when the first maximum value is satisfied to be greater than the set first threshold value. To enable an accurate determination of the center point of the tilted object.
In a specific embodiment of the present application, determining coordinates of a direction marker point of an inclined target in the target image according to the target direction density map includes: acquiring coordinates corresponding to a second maximum value of the target direction density map; determining whether the second maximum is greater than a second threshold; determining whether the coordinates corresponding to the second maximum value are within a neighborhood range of the coordinates of the central point of the inclined object; and determining that the coordinate corresponding to the second maximum value is the coordinate of the direction marker point when the second maximum value is greater than the second threshold value and the coordinate corresponding to the second maximum value is within the neighborhood range. That is, since the target direction density map has the maximum value at the direction marker point, the coordinates corresponding to the second maximum value of the target direction density map are extracted, and when the second maximum value is greater than the second threshold value, the coordinates corresponding to the second maximum value are determined as the coordinates of the direction marker point. To enable accurate determination of the directional landmark points of the tilted object.
In an embodiment of the present application, the training data further includes a size density map corresponding to the original image, and the method further includes: extracting the center point coordinates of the inclined targets in the original image from the marked original image; extracting the size of an inclined target in the original image from the marked original image; extracting the inclination angle of an inclination target in the original image from the marked original image; generating a size density map corresponding to the original image according to the center point coordinates of the inclined object, the size of the inclined object and the inclined angle of the inclined object; and inputting the target image to be predicted into the artificial intelligent model for operation, and outputting a target size density map corresponding to the target image.
In one embodiment of the present application, the method further includes: and determining the length information and the width information of the inclined target in the target image according to the target size density map.
In one embodiment of the present application, the target size density map is represented by a two-dimensional gaussian function, and determining the length information and the width information of the tilted target in the target image according to the target size density map includes: acquiring standard deviation of a first dimension and standard deviation of a second dimension of the two-dimensional Gaussian function corresponding to the target size density map; determining the length information according to the standard deviation of the first dimension; and determining the width information according to the standard deviation of the second dimension.
Specifically, the standard deviation of the first dimension has a proportional relationship with the length information, and the standard deviation of the second dimension has a proportional relationship with the width information.
In an alternative embodiment of the present application, the dimension density map may be expressed as:
x' =x×cos (θ) +y×sin (θ) equation 5
Y' = -x×sin (θ) +y×cos (θ) equation 6
Wherein, sigma x',σy' is in direct proportion to the length and width of the target p, the proportion is r, θ is the inclination angle of the target p, and z xp,zyp is the two-dimensional expression of the central coordinate of the target p.
In a more specific embodiment, the following formula is minimized:
Wherein,
x′＝x*cos(θc)+y*sin(θc)
y′＝-x*sin(θc)+y*cos(θc)
X c,yc is the central coordinate value obtained from the position density diagram before, θ c is the inclination angle value obtained from the direction density diagram, so as to determine sigma x′,σy′, and then the length and width information of the target can be obtained through the proportionality coefficient r. Where loc_ densitymap (x, y) represents the result output by the neural network model.
The embodiment of the application also provides a processing device for the inclined object, and the processing device for the inclined object can be used for executing the processing method for the inclined object. The following describes a processing device for an inclined object provided by an embodiment of the present application.
Fig. 2 is a schematic view of a processing apparatus of an incline target according to an embodiment of the present application. As shown in fig. 2, the apparatus includes:
an acquiring unit 10, configured to acquire a plurality of annotated original images, where the original images include an oblique target;
A first determining unit 20 configured to determine a position density map and a direction density map corresponding to each of the noted original images, the position density map having a value at a center point of the oblique target different from a value at a point other than the center point in the original image, the direction density map having a value at a direction flag point different from a value at a point other than the direction flag point in the original image, the direction flag point being a point other than the center point on the oblique target;
a construction unit 30, configured to construct an artificial intelligence model, where the artificial intelligence model is obtained by machine learning training using a plurality of sets of training data, and each set of training data includes: the original image, a position density map corresponding to the original image and a direction density map corresponding to the original image;
a calculating unit 40, configured to input a target image to be predicted into the artificial intelligence model for operation, and output a target position density map and a target direction density map corresponding to the target image;
a second determining unit 50 for determining an inclination angle of the inclined object in the object image based on the object position density map and the object direction density map.
In the above scheme, the artificial intelligent model is obtained by training the original image, the position density map corresponding to the original image and the direction density map corresponding to the original image as training sets, the calculating unit inputs the target image to be predicted into the artificial intelligent model for operation, outputs the target position density map and the target direction density map corresponding to the target image, and the second determining unit determines the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map. In the scheme, the mode of model training by adopting the density map avoids the defects of the corner marking and the dip marking, so that the inclination angle of the inclination target in the determined target image is more accurate.
In one embodiment of the present application, the second determining unit includes a first determining module, a second determining module, and a third determining module, where the first determining module is configured to determine coordinates of a center point of the tilted object in the object image according to the object position density map; the second determining module is used for determining coordinates of a direction marking point of the inclined target in the target image according to the target direction density diagram; and the third determining module is used for determining the inclination angle of the inclined target in the target image according to the coordinates of the central point and the coordinates of the direction mark points.
In a specific embodiment of the present application, the first determining module includes a first obtaining sub-module, a first determining sub-module, and a second determining sub-module, where the first obtaining sub-module is configured to obtain coordinates corresponding to a first maximum value of a local area of the target position density map; the first determining submodule is used for determining whether the first maximum value is larger than a first threshold value or not; the second determining submodule is used for determining that the coordinate corresponding to the first maximum value is the coordinate of the central point of the inclined target when the first maximum value is larger than the first threshold value.
In a specific embodiment of the present application, the second determining module includes a second obtaining sub-module, a third determining sub-module, a fourth determining sub-module, and a fifth determining sub-module, where the second obtaining sub-module is configured to obtain coordinates corresponding to a second maximum value of the target direction density map; the third determining submodule is used for determining whether the second maximum value is larger than a second threshold value or not; a fourth determining submodule is used for determining whether the coordinates corresponding to the second maximum value are in a neighborhood range of the coordinates of the central point of the inclined target; and the fifth determining submodule is used for determining that the coordinate corresponding to the second maximum value is the coordinate of the direction marking point when the second maximum value is larger than the second threshold value and the coordinate corresponding to the second maximum value is in the neighborhood range.
In one embodiment of the present application, the training data further includes a size density map corresponding to the original image, and the apparatus further includes a first extraction unit, a second extraction unit, a third extraction unit, a generation unit, and an operation unit, where the first extraction unit is configured to extract, from the labeled original image, a center point coordinate of an oblique target in the original image; the second extracting unit is used for extracting the size of the inclined target in the original image from the marked original image; the third extracting unit is used for extracting the inclination angle of the inclination target in the original image from the marked original image; the generation unit is used for generating a size density map corresponding to the original image according to the center point coordinates of the inclined target, the size of the inclined target and the inclined angle of the inclined target; the operation unit is used for inputting the target image to be predicted into the artificial intelligent model for operation and outputting a target size density map corresponding to the target image.
In one embodiment of the present application, the apparatus further includes a third determining unit configured to determine length information and width information of the tilted object in the object image according to the object size density map.
In one embodiment of the present application, the target size density map is represented by a two-dimensional gaussian function, and the third determining unit includes an obtaining module, a fourth determining module, and a fifth determining module, where the obtaining module is configured to obtain a standard deviation of a first dimension and a standard deviation of a second dimension of the two-dimensional gaussian function corresponding to the target size density map; the fourth determining module is used for determining the length information according to the standard deviation of the first dimension; and the fifth determining module is used for determining the width information according to the standard deviation of the second dimension.
The processing device of the tilting target comprises a processor and a memory, wherein the acquisition unit, the first determination unit, the construction unit, the calculation unit, the second determination unit and the like are all stored in the memory as program units, and the processor executes the program units stored in the memory to realize corresponding functions.
The processor includes a kernel, and the kernel fetches the corresponding program unit from the memory. The kernel can be provided with one or more than one, and the inclination angle of the inclination target is obtained by adjusting kernel parameters.
The memory may include volatile memory, random Access Memory (RAM), and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), among other forms in computer readable media, the memory including at least one memory chip.
The embodiment of the invention provides a computer readable storage medium, which comprises a stored program, wherein when the program runs, a device where the computer readable storage medium is located is controlled to execute a processing method of the inclined target.
The embodiment of the invention provides a processor which is used for running a program, wherein the program runs to execute the processing method of the inclined target.
The embodiment of the invention provides equipment, which comprises a processor, a memory and a program stored in the memory and capable of running on the processor, wherein the processor realizes at least the following steps when executing the program:
step S101, obtaining a plurality of marked original images, wherein the original images comprise inclined targets;
Step S102, determining a position density map and a direction density map corresponding to each marked original image, wherein the position density map is different from the value of a point except the center point in the original image at the center point of the inclined object, the direction density map is different from the value of a point except the direction mark point in the original image at the direction mark point, and the direction mark point is a point except the center point on the inclined object;
Step S103, constructing an artificial intelligent model, wherein the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, a position density map corresponding to the original image and a direction density map corresponding to the original image;
Step S104, inputting a target image to be predicted into the artificial intelligent model for operation, and outputting a target position density map and a target direction density map corresponding to the target image;
step S105, determining the inclination angle of the inclined object in the object image according to the object position density map and the object direction density map.
The device herein may be a server, PC, PAD, cell phone, etc.
The application also provides a computer program product adapted to perform, when executed on a data processing device, a program initialized with at least the following method steps:
step S101, obtaining a plurality of marked original images, wherein the original images comprise inclined targets;
Step S102, determining a position density map and a direction density map corresponding to each marked original image, wherein the position density map is different from the value of a point except the center point in the original image at the center point of the inclined object, the direction density map is different from the value of a point except the direction mark point in the original image at the direction mark point, and the direction mark point is a point except the center point on the inclined object;
Step S103, constructing an artificial intelligent model, wherein the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, a position density map corresponding to the original image and a direction density map corresponding to the original image;
Step S104, inputting a target image to be predicted into the artificial intelligent model for operation, and outputting a target position density map and a target direction density map corresponding to the target image;
step S105, determining the inclination angle of the inclined object in the object image according to the object position density map and the object direction density map.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, etc., such as Read Only Memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.
From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:
1) According to the processing method of the inclined target, an artificial intelligent model is obtained by training an original image, a position density image corresponding to the original image and a direction density image corresponding to the original image as training sets, then a target image to be predicted is input into the artificial intelligent model for operation, a target position density image and a target direction density image corresponding to the target image are output, and then the inclination angle of the inclined target in the target image is determined according to the target position density image and the target direction density image. In the scheme, the mode of model training by adopting the density map avoids the defects of the corner marking and the dip marking, so that the inclination angle of the inclination target in the determined target image is more accurate.
2) According to the processing device for the inclined target, an artificial intelligent model is obtained by training an original image, a position density image corresponding to the original image and a direction density image corresponding to the original image as training sets, a target image to be predicted is input into the artificial intelligent model to be operated by a calculating unit, a target position density image and a target direction density image corresponding to the target image are output, and the inclination angle of the inclined target in the target image is determined by a second determining unit according to the target position density image and the target direction density image. In the scheme, the mode of model training by adopting the density map avoids the defects of the corner marking and the dip marking, so that the inclination angle of the inclination target in the determined target image is more accurate.
The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (10)
1. A method of processing an inclined object, comprising:
acquiring a plurality of marked original images, wherein the original images comprise inclined targets;
Determining a position density map and a direction density map corresponding to each marked original image, wherein the value of the position density map at a central point of the inclined target is different from the value of points except the central point in the original image, the value of the direction density map at a direction mark point is different from the value of points except the direction mark point in the original image, and the direction mark point is a point which is not the central point on the inclined target;
constructing an artificial intelligent model, wherein the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, the position density map corresponding to the original image and the direction density map corresponding to the original image;
Inputting a target image to be predicted into the artificial intelligent model for operation, and outputting a target position density map and a target direction density map corresponding to the target image;
And determining the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map.
2. The method of claim 1, wherein determining the tilt angle of the tilted object in the object image from the object position density map and the object direction density map comprises:
determining coordinates of a center point of an inclined target in the target image according to the target position density map;
determining coordinates of a direction marking point of an inclined target in the target image according to the target direction density map;
and determining the inclination angle of the inclined target in the target image according to the coordinates of the central point and the coordinates of the direction mark points.
3. The method of claim 2, wherein determining coordinates of a center point of an inclined object in the object image from the object position density map comprises:
acquiring coordinates corresponding to a first maximum value of a local area of the target position density map;
Determining whether the first maximum value is greater than a first threshold value;
and under the condition that the first maximum value is larger than the first threshold value, determining the coordinate corresponding to the first maximum value as the coordinate of the central point of the inclined target.
4. A method according to claim 3, wherein determining coordinates of a direction marker point of an inclined target in the target image from the target direction density map comprises:
Acquiring a coordinate corresponding to a second maximum value of the target direction density map;
Determining whether the second maximum is greater than a second threshold;
Determining whether the coordinates corresponding to the second maximum value are within a neighborhood range of the coordinates of the central point of the inclined target;
and determining that the coordinate corresponding to the second maximum value is the coordinate of the direction mark point under the condition that the second maximum value is larger than the second threshold value and the coordinate corresponding to the second maximum value is in the neighborhood range.
5. The method of claim 1, wherein the training data further comprises a size density map corresponding to the original image, the method further comprising:
Extracting the center point coordinates of the inclined targets in the original image from the marked original image;
Extracting the size of an inclined target in the original image from the marked original image;
extracting the inclination angle of an inclination target in the original image from the marked original image;
generating a size density map corresponding to the original image according to the center point coordinates of the inclined target, the size of the inclined target and the inclined angle of the inclined target;
and inputting the target image to be predicted into the artificial intelligent model for operation, and outputting a target size density map corresponding to the target image.
6. The method of claim 5, wherein the method further comprises:
and determining the length information and the width information of the inclined target in the target image according to the target size density map.
7. The method of claim 6, wherein the target size density map is represented using a two-dimensional gaussian function, and determining length information and width information of an inclined target in the target image from the target size density map comprises:
acquiring standard deviation of a first dimension and standard deviation of a second dimension of the two-dimensional Gaussian function corresponding to the target size density map;
determining the length information according to the standard deviation of the first dimension;
and determining the width information according to the standard deviation of the second dimension.
8. A processing apparatus for tilting an object, comprising:
the acquisition unit is used for acquiring a plurality of marked original images, wherein the original images comprise inclined targets;
A first determining unit, configured to determine a position density map and a direction density map corresponding to each of the noted original images, where the position density map has a value at a center point of the tilted object that is different from a value at a point other than the center point in the original image, and the direction density map has a value at a direction mark point that is different from a value at a point other than the direction mark point in the original image, and the direction mark point is a point on the tilted object that is not the center point;
The building unit is used for building an artificial intelligent model, the artificial intelligent model is obtained through machine learning training by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises: the original image, the position density map corresponding to the original image and the direction density map corresponding to the original image;
The computing unit is used for inputting a target image to be predicted into the artificial intelligent model for operation and outputting a target position density map and a target direction density map corresponding to the target image;
And the second determining unit is used for determining the inclination angle of the inclined target in the target image according to the target position density map and the target direction density map.
9. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored program, wherein the program, when run, controls a device in which the computer readable storage medium is located to perform the method of any one of claims 1 to 7.
10. A processor for running a program, wherein the program when run performs the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210056957.2A CN114494398B (en) | 2022-01-18 | 2022-01-18 | Processing method and device of inclined target, storage medium and processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210056957.2A CN114494398B (en) | 2022-01-18 | 2022-01-18 | Processing method and device of inclined target, storage medium and processor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114494398A CN114494398A (en) | 2022-05-13 |
| CN114494398B true CN114494398B (en) | 2024-05-07 |
Family
ID=81471947
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210056957.2A Active CN114494398B (en) | 2022-01-18 | 2022-01-18 | Processing method and device of inclined target, storage medium and processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114494398B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117162104B (en) * | 2023-11-02 | 2024-01-30 | 泓浒(苏州)半导体科技有限公司 | Transfer mechanical arm installation machine control early warning system and method in ultra-clean environment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112418344A (en) * | 2020-12-07 | 2021-02-26 | 汇纳科技股份有限公司 | Training method, target detection method, medium and electronic device |
| WO2021129528A1 (en) * | 2019-12-28 | 2021-07-01 | 广东拓斯达科技股份有限公司 | Sorting method and apparatus, and device and storage medium |
| CN113591967A (en) * | 2021-07-27 | 2021-11-02 | 南京旭锐软件科技有限公司 | Image processing method, device and equipment and computer storage medium |
-
2022
- 2022-01-18 CN CN202210056957.2A patent/CN114494398B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021129528A1 (en) * | 2019-12-28 | 2021-07-01 | 广东拓斯达科技股份有限公司 | Sorting method and apparatus, and device and storage medium |
| CN112418344A (en) * | 2020-12-07 | 2021-02-26 | 汇纳科技股份有限公司 | Training method, target detection method, medium and electronic device |
| CN113591967A (en) * | 2021-07-27 | 2021-11-02 | 南京旭锐软件科技有限公司 | Image processing method, device and equipment and computer storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于深度学习的结构化图像标注研究;姚义;王诗珂;陈希豪;林宇翩;;电脑知识与技术;20191125(第33期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114494398A (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109117848B (en) | Text line character recognition method, device, medium and electronic equipment | |
| EP3489856B1 (en) | Method and system for recognizing location information in two-dimensional code | |
| US9990546B2 (en) | Method and apparatus for determining target region in video frame for target acquisition | |
| US20200175700A1 (en) | Joint Training Technique for Depth Map Generation | |
| CN111091123A (en) | Text region detection method and equipment | |
| CN111091023B (en) | Vehicle detection method and device and electronic equipment | |
| CN110930453B (en) | Target object positioning method, target object positioning device and readable storage medium | |
| CN110264523B (en) | Method and equipment for determining position information of target image in test image | |
| CN109583393B (en) | Lane line end point identification method and device, equipment and medium | |
| CN111561906B (en) | Robot monocular distance measuring method, system, electronic device and computer storage medium | |
| CN111340877A (en) | Vehicle positioning method and device | |
| CN114494398B (en) | Processing method and device of inclined target, storage medium and processor | |
| CN111476902B (en) | Labeling method and device for objects in 3D point cloud, storage medium and processor | |
| CN109190662A (en) | A kind of three-dimensional vehicle detection method, system, terminal and storage medium returned based on key point | |
| CN112784639A (en) | Intersection detection, neural network training and intelligent driving method, device and equipment | |
| CN110223320B (en) | Object detection tracking method and detection tracking device | |
| Barra et al. | Can Existing 3D Monocular Object Detection Methods Work in Roadside Contexts? A Reproducibility Study | |
| US20240011792A1 (en) | Method and apparatus for updating confidence of high-precision map | |
| CN115236689A (en) | Method and device for determining relative positions of laser radar and image acquisition equipment | |
| CN110827259B (en) | Sample augmentation method and device for target key point detection | |
| CN115291624A (en) | Unmanned aerial vehicle positioning landing method, storage medium and computer equipment | |
| CN112101139B (en) | Human shape detection method, device, equipment and storage medium | |
| CN114332201A (en) | Model training and target detection method and device | |
| Park et al. | Real‐time robust 3D object tracking and estimation for surveillance system | |
| CN112164107B (en) | End-to-end camera modeling method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |