CN110136227A - Mask method, device, equipment and the storage medium of high-precision map - Google Patents
Mask method, device, equipment and the storage medium of high-precision map Download PDFInfo
- Publication number
- CN110136227A CN110136227A CN201910344593.6A CN201910344593A CN110136227A CN 110136227 A CN110136227 A CN 110136227A CN 201910344593 A CN201910344593 A CN 201910344593A CN 110136227 A CN110136227 A CN 110136227A
- Authority
- CN
- China
- Prior art keywords
- map datum
- marked
- map
- operating position
- component data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000012545 processing Methods 0.000 claims abstract description 31
- 238000004590 computer program Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 230000005055 memory storage Effects 0.000 claims description 3
- 230000006399 behavior Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 11
- 238000003709 image segmentation Methods 0.000 description 10
- 238000012937 correction Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 241000283070 Equus zebra Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
- 230000001755 vocal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/60—Editing figures and text; Combining figures or text
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The application provides mask method, device, equipment and the storage medium of a kind of high-precision map, this method comprises: obtaining map datum to be marked;Image dividing processing is carried out to the map datum to be marked, obtains component data;The component data are input to preparatory trained prediction model, obtain the corresponding target labeled data of each component data;According to each target labeled data, the map datum to be marked is labeled, the map datum after being marked.Effectively increase working efficiency.
Description
Technical field
This application involves field of computer technology more particularly to a kind of mask method of high-precision map, device, equipment and deposit
Storage media.
Background technique
With the rapid development of the advanced driving assistance system technology of vehicle and unmanned technology, to map precision is wanted
It asks and is gradually increased, the map of original road grade can no longer meet the demand of automatic driving vehicle, and one kind is needed to be capable of providing
In high precision, the high-precision map of detailed path information.
Currently, the marker in high-precision map, which usually relies on, manually carries out auxiliary mark, mark personnel are rule of thumb searched
Corresponding streetscape data out, and control map is labeled.Existing mask method is needed using multiple applicating cooperation operations, and
To the more demanding of mark personnel, need to mark the region that personnel are familiar with mark, correctly to transfer the streetscape data of reference,
Working efficiency is lower.
Summary of the invention
The application provides mask method, device, equipment and the storage medium of a kind of high-precision map, to solve prior art mark
Infuse the defects such as working efficiency is low.
The application first aspect provides a kind of mask method of high-precision map, comprising:
Obtain map datum to be marked;
Image dividing processing is carried out to the map datum to be marked, obtains component data;
The component data are input to preparatory trained prediction model, obtain the corresponding target mark of each component data
Data;
According to each target labeled data, the map datum to be marked is labeled, the map after being marked
Data.
The application the second aspect provides a kind of annotation equipment of high-precision map, comprising:
Module is obtained, for obtaining map datum to be marked;
Divide module, for carrying out image dividing processing to the map datum to be marked, obtains component data;
Prediction module obtains each component data for the component data to be input to preparatory trained prediction model
Corresponding target labeled data;
Processing module is obtained for being labeled to the map datum to be marked according to each target labeled data
Map datum after mark.
A kind of computer equipment is provided in terms of the application third, comprising: at least one processor and memory;
The memory stores computer program;At least one described processor executes the computer of the memory storage
Program, the method to realize first aspect offer.
The 4th aspect of the application provides a kind of computer readable storage medium, stores in the computer readable storage medium
There is computer program, the computer program is performed the method for realizing that first aspect provides.
Mask method, device, equipment and the storage medium of high-precision map provided by the present application, by map number to be marked
According to image dividing processing is carried out, at least two component data are obtained, and each component is predicted using trained prediction model in advance
The corresponding target labeled data of data, is labeled map datum to be marked, and the map datum after being marked effectively improves
Working efficiency.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this Shen
Some embodiments please for those of ordinary skill in the art without any creative labor, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow diagram of the mask method for the high-precision map that one embodiment of the application provides;
Fig. 2 is the flow diagram of the mask method for the high-precision map that another embodiment of the application provides;
Fig. 3 is the exemplary flow diagram of the mask method for the high-precision map that one embodiment of the application provides;
Fig. 4 is the structural schematic diagram of the annotation equipment for the high-precision map that one embodiment of the application provides;
Fig. 5 is the structural schematic diagram for the computer equipment that one embodiment of the application provides.
Through the above attached drawings, it has been shown that the specific embodiment of the application will be hereinafter described in more detail.These attached drawings
It is not intended to limit the scope of this disclosure concept by any means with verbal description, but is by referring to specific embodiments
Those skilled in the art illustrate the concept of the application.
Specific embodiment
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall in the protection scope of this application.
Noun involved in the application is explained first:
UTM:Universal Transverse Mercartor Grid System, Universal Trans Meridian grid system.
UTM coordinate is a kind of plane rectangular coordinates, this coordinate axiom system and its based on projection be widely used for topographic map,
As satellite image and natural resources database grid of reference and require pinpoint other application.In UTM system,
Earth surface product between 84 degree of north latitude and 80 degree of south latitude is divided into north and south longitudinal bands (projection zone) by 6 degree of longitude.From 180 degree warp
Start eastwards to number these projection zones, compile from 1 to 60 (Beijing is in the 50th band).Each band is further subdivided into the four of 8 degree of the meridional difference
Side shape.The row of quadrangle is since 80 degree of south latitude.With letter C to X (being free of I and O), successively (X row includes the Northern Hemisphere to label
From 72 degree to 84 degree of north latitude whole land areas, totally 12 degree) each quadrangle is digital and monogram marks.Grid of reference to
It is read in dextrad.Each quadrilateral partition is the cell that many side lengths are 1000000 meters, with monogram system marks.Every
In a projection zone, positioned at the warp with center, assigning abscissa value is 500000 meters.For the label coordinate value in Northern Hemisphere equator
It is 0, is 10000000 meters for the Southern Hemisphere, southward successively decreases.
The mask method of high-precision map provided by the embodiments of the present application, is labeled suitable for high-precision map.To be marked
Map datum carries out image segmentation, and map to be marked is divided into small image, and using preparatory trained prediction model, prediction is each
The corresponding labeled data of small image, is automatically labeled map datum to be marked according to the labeled data of prediction, is marked
Map datum afterwards, effectively improves working efficiency.Wherein, prediction model can be the model based on convolutional neural networks, specifically
The network architecture can be arranged according to actual needs.
In addition, term " first ", " second " etc. are used for description purposes only, it is not understood to indicate or imply relatively important
Property or implicitly indicate the quantity of indicated technical characteristic.In the description of following embodiment, the meaning of " plurality " is two
More than a, unless otherwise specifically defined.
These specific embodiments can be combined with each other below, may be at certain for the same or similar concept or process
It is repeated no more in a little embodiments.Below in conjunction with attached drawing, the embodiment of the present invention is described.
Embodiment one
The present embodiment provides a kind of mask methods of high-precision map, the mark for high-precision map.The execution of the present embodiment
Main body is the annotation equipment of high-precision map, which can be set in computer equipment.Computer equipment can be desktop
The equipment such as brain, laptop
As shown in Figure 1, the flow diagram of the mask method for high-precision map provided in this embodiment, this method comprises:
Step 101, map datum to be marked is obtained.
Specifically, original map data can be generated according to the map wait mark after using vehicle acquisition original map data
Map datum is infused, map to be marked is established.After establishing map to be marked, needs to be labeled map to be marked, can obtain
Map datum to be marked is taken, and is performed corresponding processing.
Map datum to be marked can be the map datum under radar fix (i.e. latitude and longitude coordinates), can specifically include through
Latitude coordinate and image.Map datum to be marked can also be the map datum under UTM coordinate.It specifically can be according to actual needs
Setting, the present embodiment does not limit.
Step 102, image dividing processing is carried out to map datum to be marked, obtains component data.
Specifically, image dividing processing can be carried out to map datum to be marked after getting map datum to be marked,
Obtain component data.Component data can be stored, and generate the storage path of each component data.Image segmentation refer to by
For the big image segmentation of map datum to be marked at multiple small images, specific image segmentation mode can be any in the prior art
Enforceable mode, the present embodiment does not limit.
Optionally, it in order to realize the scaling of map, can be map datum to be marked according to multiple pre-set zoom grades
It is split, obtains the component data under different zoom grade.Under different zoom grade, the map state of display is different, with certain
For city map, the whole map that 0 zoom level shows the city in display interface can be set, not with zoom level
Disconnected to increase, the practical map area that display interface is shown is smaller, but the markup information shown can be more detailed.That is different zoom grade
The markup information of display is also different.Such as in 0 zoom level, it is shown that the approximate region in the entire map in the city and the city
Range divides, and markup information shows the title in city name and each region, and highest zoom level can be shown in display interface to be referred to
Determine detailed path, building and the corresponding markup information of range.Such as on certain road lane (lane, 2 lanes or
3 lanes), the various buildings of lane line (solid line or dotted line) and road both sides and other objects (such as direction board) etc..
Step 103, component data are input to preparatory trained prediction model, obtain the corresponding target of each component data
Labeled data.
Specifically, component data can be input to preparatory trained prediction model, obtained after obtaining component data
The corresponding target labeled data of each component data.
Prediction model is that acquisition is trained by neural network of a large amount of training datas to foundation, and training data includes pre-
The image data first obtained and corresponding labeled data.Prediction model can be convolutional neural networks model.It may include input
Layer, convolutional layer, pond layer, output layer etc..The network structure of specific prediction model can be arranged according to actual needs, this implementation
Example is without limitation.
Target labeled data can specifically include type and attribute belonging to each component data, and type includes lane line, vehicle
Other correlation types can also be arranged in road, crossing, pavement, traffic lights etc. according to actual needs.Attribute refer to it is all types of under
Subclasses, such as lane line have solid line, dotted line etc. differentiation, lane has Emergency Vehicle Lane, ring road etc.;Crossing has straight trip
Crossing, T-type crossing, crossroad etc..It can specifically be arranged according to actual needs, the present embodiment does not limit.
It should be understood that all types of and attribute can be with different symbols and indicate in target labeled data.It can build
Vertical symbol and all types of and attribute corresponding relationships show all types of and attribute in map according to the preset style in display
In.
Step 104, according to each target labeled data, map datum to be marked is labeled, the map after being marked
Data.
Specifically, after obtaining the target labeled data of map datum to be marked then number can be marked according to each target
According to being labeled to map datum to be marked, the map datum after being marked.Specifically, mesh can be established according to coordinate
The corresponding relationship for marking labeled data and map datum to be marked shows mark letter according to corresponding relationship in display in map
Breath.Here markup information is all types of and display information of the attribute in map.
Illustratively, the corresponding type in certain region is pavement in target labeled data, then establishes map datum to be marked
In, the corresponding relationship of the area coordinate and pavement shows the corresponding mark letter in pavement in the region when showing map
Breath, such as zebra stripes.
The mask method of high-precision map provided in this embodiment, by being carried out at image segmentation to map datum to be marked
Reason obtains at least two component data, and predicts the corresponding target mark of each component data using trained prediction model in advance
Data are infused, map datum to be marked are labeled, the map datum after being marked effectively increases working efficiency.
Embodiment two
The method that the present embodiment provides embodiment one does further supplementary explanation.
As shown in Fig. 2, the flow diagram of the mask method for high-precision map provided in this embodiment.
As a kind of enforceable mode, on the basis of the above embodiment 1, optionally, map datum to be marked includes
At least one corresponding map datum of pre-set zoom grade;Correspondingly, step 102 can specifically include:
Step 1021, according to each pre-set zoom grade, Slice by slice cutting processing is carried out to map datum to be marked, is obtained each pre-
If the component data under zoom level.
Step 103 can specifically include:
Step 1031, the component data under each pre-set zoom grade are input to preparatory trained prediction model, obtained
Target labeled data corresponding to component data under each pre-set zoom grade.
Step 104 can specifically include:
According to target labeled data corresponding to the component data under each pre-set zoom grade, to map datum to be marked into
Rower note, the map datum after being marked.
Specifically, the scaling in order to realize map, can be map datum to be marked according to multiple pre-set zoom grades
It is split, obtains the component data under different zoom grade.Under different zoom grade, the map state of display is different, with certain
For city map, the whole map that 0 zoom level shows the city in display interface can be set, not with zoom level
Disconnected to increase, the practical map area that display interface is shown is smaller, but the markup information shown can be more detailed.That is different zoom grade
The markup information of display is also different.Such as in 0 zoom level, it is shown that the approximate region in the entire map in the city and the city
Range divides, and markup information shows the title in city name and each region, and highest zoom level can be shown in display interface to be referred to
Determine detailed path, building and the corresponding markup information of range.Such as on certain road lane (lane, 2 lanes or
3 lanes), the various buildings of lane line (solid line or dotted line) and road both sides and other objects (such as direction board) etc..
To map image carries out multi-zone supervision and is applied primarily to following formula:
Wherein, S indicates the pantograph ratio under current zoom grade.The component quantity that different zoom Multi-level segmentation obtains is different,
Zoom level is higher, and the component quantity for dividing acquisition is more.Illustratively, zoom level can be set to ten grades of 0-9.Tool
Body can be arranged according to actual needs.
As another enforceable mode, on the basis of the above embodiment 1, optionally, target labeled data includes
Type and attribute belonging to its corresponding component, wherein type include at least lane line, lane, road, crossing, pavement and
Traffic lights, attribute be it is all types of under subclasses.
Specifically, target labeled data can specifically include type and attribute belonging to each component data, type includes vehicle
Other correlation types can also be arranged in diatom, lane, crossing, pavement, traffic lights etc. according to actual needs.Attribute refers to
Subclasses under all types of, such as lane line have the differentiation such as solid line, dotted line, and lane has Emergency Vehicle Lane, ring road etc.;Crossing
Have straight trip crossing, T-type crossing, crossroad etc..It can specifically be arranged according to actual needs.
As another enforceable mode, on the basis of the above embodiment 1, optionally, marked according to each target
Data are labeled map datum to be marked, and after the map datum after being marked, this method can also include:
Step 105, the map datum after mark is subjected to display processing.
Optionally, the map datum after mark is subjected to display processing, comprising:
Map datum after mark is subjected to coordinate conversion, obtains the map datum under UTM coordinate;
Map datum under UTM coordinate is shown.
Specifically, can then be shown the map datum after mark after the map datum after being marked, show
Show that content includes the corresponding markup information of each position in map and map.The display of map is shown based on UTM coordinate system,
Therefore, if the map datum after mark is latitude and longitude coordinates system, need to convert thereof into UTM coordinate system.If the ground after mark
Diagram data has been UTM coordinate system, then can directly be shown.Default zoom grade, such as default contracting can be preset
Putting grade is 0 grade, then is that just can see entire map in display interface when being initially displayed.User can pass through interface operation
Enlarged map enters the state under other zoom levels.
Illustratively, the conversion process of latitude and longitude coordinates system to UTM coordinate system specifically includes:
Coordinates of targets to be converted is obtained, coordinates of targets is latitude and longitude coordinates (Lat, Lng), and Lat indicates dimension, Lng table
Show longitude.It, will according to the image pixel and actual range ratio P under the current zoom grade S of coordinates of targets and current zoom grade
Coordinates of targets is converted to corresponding UTM coordinate (utmx, utmy).It is converted with specific reference to following formula:
Optionally, after the map datum after mark is carried out display processing, this method further include:
User is obtained in first operating position at map denotation interface;
Corresponding real-time street view image is obtained according to the first operating position;
Real-time street view image is shown in the first operating position, so that mark of the user according to real-time street view image rectification mistake
Infuse data.
Specifically, after being shown the map datum after mark, it can be automatic for user, such as mark personnel inspection
Whether annotation results are correct, and user can operate in display interface, for example mouse clicks some position of display interface, then may be used
To obtain the first operating position of user's operation, can be determined according to the first operating position on the corresponding map of user's click location
UTM coordinate, the real-time street view image that shoots in the UTM coordinate position of vehicle is obtained according to the UTM coordinate.It can pre-establish
The corresponding relationship of UTM coordinate and real-time street view image.After getting the corresponding real-time street view image of the UTM coordinate, it can incite somebody to action
Near the first operating position or the first operating position that real-time street view image is displayed in the display interface, so that user checks reality
When street view image.The real-time street view image can be specifically shown according to the wheelpath of collecting vehicle, user can be according to real-time
Street view image finds the place of error label in time and is modified.
Optionally, it is shown after the first operating position by real-time street view image, this method further include:
User is obtained in the second operating position of display interface;
According to the second operating position, the corresponding UTM coordinate of the second operating position is obtained;
The corresponding UTM coordinate of second operating position is converted into the corresponding radar fix of the second operating position;
The corresponding radar fix of second operating position is converted into the corresponding camera coordinates of the second operating position;
According to the corresponding camera coordinates of the second operating position, the target area on real-time street view image is positioned, so that user
Check whether the second operating position is currently marked consistent with real-time street view image.
Specifically, after the first operating position of user shows the real-time street view image shot in the position, user can be with
According to wheelpath, check in front of the first operating position is with real-time street view with the markup information of real-time street view image corresponding position
No consistent, user can click each position by mouse, can get the second operating position of user, can be according to the second behaviour
Make the corresponding UTM coordinate of the second operating position of position acquisition, and be converted into radar fix, reconvert is at camera coordinates, according to phase
Target area of the second operating position of machine coordinate setting on real-time street view image, the mark currently shown according to the second operating position
The target area infused on information and real-time street view image compares, come judge mark whether mistake.
Illustratively, the corresponding real-time street view image taking of the first operating position has arrived traffic lights, and user can be first
The position of mark traffic lights is found in front of operating position, is carried out the second operation, is got the second operating position of user, converts
After camera coordinates, the target area on real-time street view image is navigated to, if target area is traffic lights, then it represents that traffic lights mark
Note is accurate.
Illustratively, UTM coordinate system is specifically included to the conversion process that latitude and longitude coordinates system arrives camera coordinates system again:
Camera extrinsic is obtained, the transposition R1=R of outer ginseng spin matrix is constructed according to Camera extrinsicT, obtain reference position UTM
Coordinate (i.e. the corresponding UTM coordinate of the first operating position) calculates intermediate parameters T1 according to the following formula:
T1=-RT*T
Another intermediary matrix P1 are as follows:
Target position UTM coordinate (i.e. the corresponding UTM coordinate of the second operating position) is converted into the seat under radar fix system
Mark:
After obtaining the coordinate under radar fix system, camera internal reference K is obtained:
Wherein, fx, fy are focal length, and under normal circumstances, the two is equal, and x0, y0 are principal point coordinate (relative to imaging plane),
0.) and Camera extrinsic includes spin matrix R, translation vector T s is reference axis tilt parameters, ideally for, according to such as
Lower formula obtains the coordinate under camera coordinates system:
The embodiment of property as an example, as shown in figure 3, being the mask method of high-precision map provided in this embodiment
Exemplary flow diagram.It specifically includes:
1, figure management is cut in layering.
Specifically, carrying out layering to map datum to be marked cuts figure management, that is, image segmentation is carried out, obtains different zoom etc.
The corresponding component data of grade.
2, component imports prediction data
Specifically, carrying out prediction using prediction model obtains target labeled data, target labeled data is imported into foundation
Map to be marked is shown on map to be marked.
3, mark personnel examine by real-time street view image.Wrong error correction, error-free progress the next item down markup information are examined
It looks into.
4, examine that errorless rear output high quality marks file.
Optionally, mark personnel examined after, can also by after examination map datum and labeled data be used to optimize
The accuracy of prediction model is continuously improved in prediction model.
It should be noted that each enforceable mode can individually be implemented in the present embodiment, it can also be in the feelings not conflicted
It is combined in any combination under condition and implements the application without limitation.
The mask method of high-precision map provided in this embodiment, by being carried out at image segmentation to map datum to be marked
Reason obtains at least two component data, and predicts the corresponding target mark of each component data using trained prediction model in advance
Data are infused, map datum to be marked are labeled, the map datum after being marked effectively increases working efficiency.And
After the map datum after being marked, the corresponding real-time street view of specified location in user can also be shown to user in display interface
Images for user examines that automatic marking as a result, with timely error correction, improves the accuracy of mark.
Embodiment three
The present embodiment provides a kind of annotation equipments of high-precision map, the method for executing above-described embodiment one.
As shown in figure 4, the structural schematic diagram of the annotation equipment for high-precision map provided in this embodiment.The high-precision map
Annotation equipment 30 includes obtaining module 31, segmentation module 32, prediction module 33 and processing module 34.
Wherein, module is obtained, for obtaining map datum to be marked;Divide module, for map datum to be marked into
Row image dividing processing obtains component data;Prediction module, for component data to be input to preparatory trained prediction mould
Type obtains the corresponding target labeled data of each component data;Processing module is used for according to each target labeled data, to be marked
Map datum is labeled, the map datum after being marked.
Device in this present embodiment is closed, wherein modules execute the concrete mode of operation in related this method
It is described in detail in embodiment, no detailed explanation will be given here.
According to the annotation equipment of high-precision map provided in this embodiment, by carrying out image segmentation to map datum to be marked
Processing obtains at least two component data, and predicts the corresponding target of each component data using trained prediction model in advance
Labeled data, is labeled map datum to be marked, and the map datum after being marked effectively increases working efficiency.
Example IV
The device that the present embodiment provides above-described embodiment three does further supplementary explanation, to execute above-described embodiment two
Method.
As a kind of enforceable mode, on the basis of above-described embodiment three, optionally, map datum to be marked includes
At least one corresponding map datum of pre-set zoom grade.Correspondingly, divide module, be specifically used for:
According to each pre-set zoom grade, Slice by slice cutting processing is carried out to map datum to be marked, obtains each pre-set zoom etc.
Component data under grade.
Prediction module is specifically used for:
Component data under each pre-set zoom grade are input to preparatory trained prediction model, obtain each pre-set zoom
Target labeled data corresponding to component data under grade.
As another enforceable mode, on the basis of above-described embodiment three, optionally, target labeled data includes
Type and attribute belonging to its corresponding component, wherein type include at least lane line, lane, road, crossing, pavement and
Traffic lights, attribute be it is all types of under subclasses.
As another enforceable mode, on the basis of above-described embodiment three, optionally, processing module is also used to:
Map datum after mark is subjected to display processing.
Optionally, processing module is specifically used for:
Map datum after mark is subjected to coordinate conversion, obtains the map datum under UTM coordinate;
Map datum under UTM coordinate is shown.
Optionally, processing module is also used to:
User is obtained in first operating position at map denotation interface;
Corresponding real-time street view image is obtained according to the first operating position;
Real-time street view image is shown in the first operating position, so that mark of the user according to real-time street view image rectification mistake
Infuse data.
Optionally, processing module is also used to:
User is obtained in the second operating position of display interface;
According to the second operating position, the corresponding UTM coordinate of the second operating position is obtained;
The corresponding UTM coordinate of second operating position is converted into the corresponding radar fix of the second operating position;
The corresponding radar fix of second operating position is converted into the corresponding camera coordinates of the second operating position;
According to the corresponding camera coordinates of the second operating position, the target area on real-time street view image is positioned, so that user
Check whether the second operating position is currently marked consistent with real-time street view image.
Device in this present embodiment is closed, wherein modules execute the concrete mode of operation in related this method
It is described in detail in embodiment, no detailed explanation will be given here.
It should be noted that each enforceable mode can individually be implemented in the present embodiment, it can also be in the feelings not conflicted
It is combined in any combination under condition and implements the application without limitation.
According to the annotation equipment of the high-precision map of the present embodiment, by being carried out at image segmentation to map datum to be marked
Reason obtains at least two component data, and predicts the corresponding target mark of each component data using trained prediction model in advance
Data are infused, map datum to be marked are labeled, the map datum after being marked effectively increases working efficiency.And
After the map datum after being marked, the corresponding real-time street view of specified location in user can also be shown to user in display interface
Images for user examines that automatic marking as a result, with timely error correction, improves the accuracy of mark.
Embodiment five
The present embodiment provides a kind of computer equipments, for executing method provided by the above embodiment.
As shown in figure 5, being the structural schematic diagram of computer equipment provided in this embodiment.The computer equipment 50 includes:
At least one processor 51 and memory 52;
Memory stores computer program;At least one processor executes the computer program of memory storage, to realize
Method provided by the above embodiment.
It is obtained extremely according to the computer equipment of the present embodiment by carrying out image dividing processing to map datum to be marked
Few two component data, and the corresponding target labeled data of each component data is predicted using trained prediction model in advance, it is right
Map datum to be marked is labeled, and the map datum after being marked effectively increases working efficiency.And it is being marked
After map datum afterwards, the corresponding real-time street view images for user of specified location in user can also be shown to user in display interface
Examine that automatic marking as a result, with timely error correction, improves the accuracy of mark.
Embodiment six
The present embodiment provides a kind of computer readable storage medium, computer is stored in the computer readable storage medium
Program, computer program are performed the method for realizing that any of the above-described embodiment provides.
According to the computer readable storage medium of the present embodiment, by being carried out at image segmentation to map datum to be marked
Reason obtains at least two component data, and predicts the corresponding target mark of each component data using trained prediction model in advance
Data are infused, map datum to be marked are labeled, the map datum after being marked effectively increases working efficiency.And
After the map datum after being marked, the corresponding real-time street view of specified location in user can also be shown to user in display interface
Images for user examines that automatic marking as a result, with timely error correction, improves the accuracy of mark.
In several embodiments provided herein, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that equipment (can be personal computer, server or the network equipment etc.) or processor (processor) execute the application
The part steps of embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. it is various
It can store the medium of program code.
Those skilled in the art can be understood that, for convenience and simplicity of description, only with above-mentioned each functional module
Division progress for example, in practical application, can according to need and above-mentioned function distribution is complete by different functional modules
At the internal structure of device being divided into different functional modules, to complete all or part of the functions described above.On
The specific work process for stating the device of description, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
Finally, it should be noted that the above various embodiments is only to illustrate the technical solution of the application, rather than its limitations;To the greatest extent
Pipe is described in detail the application referring to foregoing embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement;And these are modified or replaceed, each embodiment technology of the application that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (10)
1. a kind of mask method of high-precision map characterized by comprising
Obtain map datum to be marked;
Image dividing processing is carried out to the map datum to be marked, obtains component data;
The component data are input to preparatory trained prediction model, obtain the corresponding target mark number of each component data
According to;
According to each target labeled data, the map datum to be marked is labeled, the map datum after being marked.
2. the method according to claim 1, wherein the map datum to be marked includes at least one default contracting
Put the corresponding map datum of grade;
Correspondingly, carrying out image dividing processing to the map datum to be marked, component data are obtained, comprising:
According to each pre-set zoom grade, Slice by slice cutting processing is carried out to the map datum to be marked, obtains each default contracting
Put the component data under grade;
The component data are input to preparatory trained prediction model, obtain the corresponding target mark number of each component data
According to, comprising:
Component data under each pre-set zoom grade are input to preparatory trained prediction model, obtain each pre-set zoom
Target labeled data corresponding to component data under grade.
3. the method according to claim 1, wherein the target labeled data includes belonging to its corresponding component
Type and attribute, wherein type include at least lane line, lane, road, crossing, pavement and traffic lights, attribute be it is all kinds of
Subclasses under type.
4. the method according to claim 1, wherein according to each target labeled data, to described wait mark
Note map datum is labeled, after the map datum after being marked, the method also includes:
Map datum after the mark is subjected to display processing.
5. according to the method described in claim 4, it is characterized in that, the map datum by after the mark carries out at display
Reason, comprising:
Map datum after the mark is subjected to coordinate conversion, obtains the map datum under UTM coordinate;
Map datum under the UTM coordinate is shown.
6. according to the method described in claim 4, it is characterized in that, the map datum after the mark is carried out display processing
Later, the method also includes:
User is obtained in first operating position at map denotation interface;
Corresponding real-time street view image is obtained according to first operating position;
The real-time street view image is shown in first operating position, so that the user is according to the real-time street view image
Correct the labeled data of mistake.
7. according to the method described in claim 6, it is characterized in that, showing by the real-time street view image in first behaviour
After making position, the method also includes:
User is obtained in the second operating position of display interface;
According to second operating position, the corresponding UTM coordinate of second operating position is obtained;
The corresponding UTM coordinate of second operating position is converted into the corresponding radar fix of second operating position;
The corresponding radar fix of second operating position is converted into the corresponding camera coordinates of second operating position;
According to the corresponding camera coordinates of second operating position, the target area on the real-time street view image is positioned, so that
The user checks whether second operating position is currently marked consistent with real-time street view image.
8. a kind of annotation equipment of high-precision map characterized by comprising
Module is obtained, for obtaining map datum to be marked;
Divide module, for carrying out image dividing processing to the map datum to be marked, obtains component data;
It is corresponding to obtain each component data for the component data to be input to preparatory trained prediction model for prediction module
Target labeled data;
Processing module, for being labeled, being marked to the map datum to be marked according to each target labeled data
Map datum afterwards.
9. a kind of computer equipment characterized by comprising at least one processor and memory;
The memory stores computer program;At least one described processor executes the computer journey of the memory storage
Sequence, to realize method of any of claims 1-7.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer journey in the computer readable storage medium
Sequence, the computer program, which is performed, realizes method of any of claims 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910344593.6A CN110136227A (en) | 2019-04-26 | 2019-04-26 | Mask method, device, equipment and the storage medium of high-precision map |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910344593.6A CN110136227A (en) | 2019-04-26 | 2019-04-26 | Mask method, device, equipment and the storage medium of high-precision map |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110136227A true CN110136227A (en) | 2019-08-16 |
Family
ID=67575242
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910344593.6A Pending CN110136227A (en) | 2019-04-26 | 2019-04-26 | Mask method, device, equipment and the storage medium of high-precision map |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110136227A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111679282A (en) * | 2020-06-16 | 2020-09-18 | 长沙行深智能科技有限公司 | Map representation and positioning method and system for container storage environment |
CN112288017A (en) * | 2020-10-30 | 2021-01-29 | 常州微亿智造科技有限公司 | Method and device for identifying marking accuracy of image defect marking model |
CN112307144A (en) * | 2020-10-10 | 2021-02-02 | 江西裕丰智能农业科技有限公司 | System and method for automatically marking town information by using artificial intelligence |
CN112446916A (en) * | 2019-09-02 | 2021-03-05 | 北京京东乾石科技有限公司 | Method and device for determining parking position of unmanned vehicle |
CN112880692A (en) * | 2019-11-29 | 2021-06-01 | 北京市商汤科技开发有限公司 | Map data annotation method and device and storage medium |
CN113554871A (en) * | 2021-07-19 | 2021-10-26 | 联想(北京)有限公司 | Internet of vehicles data processing method and electronic equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103915036A (en) * | 2012-12-30 | 2014-07-09 | 高德信息技术有限公司 | Method and device for calculating road name mark points |
WO2015184996A1 (en) * | 2014-06-05 | 2015-12-10 | Tencent Technology (Shenzhen) Company Limited | Method and device for providing guidance to street view destination |
CN108694882A (en) * | 2017-04-11 | 2018-10-23 | 百度在线网络技术(北京)有限公司 | Method, apparatus and equipment for marking map |
CN109215487A (en) * | 2018-08-24 | 2019-01-15 | 宽凳(北京)科技有限公司 | A kind of high-precision cartography method based on deep learning |
CN109377509A (en) * | 2018-09-26 | 2019-02-22 | 深圳前海达闼云端智能科技有限公司 | Method, apparatus, storage medium and the equipment of image, semantic segmentation mark |
-
2019
- 2019-04-26 CN CN201910344593.6A patent/CN110136227A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103915036A (en) * | 2012-12-30 | 2014-07-09 | 高德信息技术有限公司 | Method and device for calculating road name mark points |
WO2015184996A1 (en) * | 2014-06-05 | 2015-12-10 | Tencent Technology (Shenzhen) Company Limited | Method and device for providing guidance to street view destination |
CN108694882A (en) * | 2017-04-11 | 2018-10-23 | 百度在线网络技术(北京)有限公司 | Method, apparatus and equipment for marking map |
CN109215487A (en) * | 2018-08-24 | 2019-01-15 | 宽凳(北京)科技有限公司 | A kind of high-precision cartography method based on deep learning |
CN109377509A (en) * | 2018-09-26 | 2019-02-22 | 深圳前海达闼云端智能科技有限公司 | Method, apparatus, storage medium and the equipment of image, semantic segmentation mark |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112446916A (en) * | 2019-09-02 | 2021-03-05 | 北京京东乾石科技有限公司 | Method and device for determining parking position of unmanned vehicle |
CN112880692A (en) * | 2019-11-29 | 2021-06-01 | 北京市商汤科技开发有限公司 | Map data annotation method and device and storage medium |
CN112880692B (en) * | 2019-11-29 | 2024-03-22 | 北京市商汤科技开发有限公司 | Map data labeling method and device and storage medium |
CN111679282A (en) * | 2020-06-16 | 2020-09-18 | 长沙行深智能科技有限公司 | Map representation and positioning method and system for container storage environment |
CN112307144A (en) * | 2020-10-10 | 2021-02-02 | 江西裕丰智能农业科技有限公司 | System and method for automatically marking town information by using artificial intelligence |
CN112288017A (en) * | 2020-10-30 | 2021-01-29 | 常州微亿智造科技有限公司 | Method and device for identifying marking accuracy of image defect marking model |
CN113554871A (en) * | 2021-07-19 | 2021-10-26 | 联想(北京)有限公司 | Internet of vehicles data processing method and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110136227A (en) | Mask method, device, equipment and the storage medium of high-precision map | |
CN108415032A (en) | A kind of point cloud semanteme map constructing method based on deep learning and laser radar | |
Müller et al. | Area-patch generalisation: a competitive approach | |
EP3951321A1 (en) | Method and system for rapid generation of reference driving route, terminal and storage medium | |
CN109064506A (en) | Accurately drawing generating method, device and storage medium | |
CN110132291B (en) | Grid map generation method, system, equipment and storage medium for port | |
CN110414526B (en) | Training method, training device, server and storage medium for semantic segmentation network | |
CN110163065A (en) | Processing Method of Point-clouds, point cloud data loading method and device and equipment | |
CN108388641A (en) | A kind of means of transportation based on deep learning ground drawing generating method and system | |
CN110244734A (en) | A kind of automatic driving vehicle paths planning method based on depth convolutional neural networks | |
WO2021051344A1 (en) | Method and apparatus for determining lane lines in high-precision map | |
CN109858374A (en) | Arrow class graticule extraction method and device in high-precision cartography | |
EP4134624A2 (en) | Method and apparatus for fusing road data to generate a map | |
CN113066112B (en) | Indoor and outdoor fusion method and device based on three-dimensional model data | |
CN112132232A (en) | Medical image classification labeling method and system and server | |
Li et al. | Exploring multiple crowdsourced data to learn deep convolutional neural networks for road extraction | |
CN105608714B (en) | A kind of framing image generating method and system based on CAD and GIS collaboration | |
CN103400125B (en) | Double-scanning double-labeling method for image connected domain | |
CN112525208B (en) | Method, device and equipment for quickly constructing urban road map | |
CN112799430B (en) | Programmable unmanned aerial vehicle-based road surface image intelligent acquisition method | |
CN114419338B (en) | Image processing method, image processing device, computer equipment and storage medium | |
CN108364456A (en) | Method, storage medium, device and system for determining highway pile number | |
CN117611759B (en) | Three-dimensional model-based scoring map generation method, device, terminal and storage medium | |
CN117079142B (en) | Anti-attention generation countermeasure road center line extraction method for automatic inspection of unmanned aerial vehicle | |
Xiong et al. | Light4Mars: A lightweight transformer model for semantic segmentation on unstructured environment like Mars |
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: 20190816 |