CN110136134A - A kind of deep learning method, apparatus, equipment and medium for road surface segmentation - Google Patents
A kind of deep learning method, apparatus, equipment and medium for road surface segmentation Download PDFInfo
- Publication number
- CN110136134A CN110136134A CN201910264997.4A CN201910264997A CN110136134A CN 110136134 A CN110136134 A CN 110136134A CN 201910264997 A CN201910264997 A CN 201910264997A CN 110136134 A CN110136134 A CN 110136134A
- Authority
- CN
- China
- Prior art keywords
- matrix
- filtering
- image
- predetermined filter
- road surface
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4038—Scaling the whole image or part thereof for image mosaicing, i.e. plane images composed of plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30256—Lane; Road marking
Abstract
The embodiment of the invention discloses a kind of deep learning method, apparatus, equipment and media for road surface segmentation, to improve the generalization ability of model, to promote the accuracy of separation of image in the segmentation of road surface.The deep learning method for road surface segmentation, it include: successively to be filtered using multiple Predetermined filters to primitive character image, at least one in the row and column of filtering matrix is greater than 1 in each Predetermined filter, and the primitive character image is generated after carrying out sampling processing to the original image comprising traffic route information;Characteristic image after filtering processing is spliced with the primitive character image, generates target signature image;Road surface separation calculation is carried out based on the target signature image.
Description
Technical field
The present invention relates to computer sciences, more particularly, to a kind of deep learning method for road surface segmentation, dress
It sets, equipment and medium.
Background technique
Road surface segmentation is an important technology in computer vision, it is in automatic Pilot, intelligent vehicle technology and road roadblock
Hindering in the application such as analyte detection all has important value.
In the prior art, in the characteristic image extraction process of road surface segmentation, using only the characteristic image after convolutional calculation
Gradient information can be lost when convolution number increases by carrying out road surface segmentation, cause road surface segmentation inaccuracy.
Summary of the invention
The embodiment of the present invention provides a kind of deep learning method, apparatus, equipment and medium for road surface segmentation, to mention
The generalization ability of high model, to promote the accuracy of separation of image in the segmentation of road surface.
In a first aspect, a kind of deep learning method for road surface segmentation, comprising:
Primitive character image is successively filtered using multiple Predetermined filters, filtering in each Predetermined filter
At least one in the row and column of matrix is greater than 1, and primitive character image is adopted to the original image comprising traffic route information
It is generated after sample processing;
Characteristic image after filtering processing is spliced with primitive character image, generates target signature image;
Road surface separation calculation is carried out based on target signature image.
Deep learning method provided in an embodiment of the present invention for road surface segmentation, using multiple Predetermined filters to original
Characteristic image is successively filtered, at least one in the row and column of filtering matrix is original greater than 1 in each Predetermined filter
Characteristic image is generated after carrying out sampling processing to the original image comprising traffic route information, by the feature after filtering processing
Image is spliced with primitive character image, generates target signature image, carries out road surface separation calculation based on target signature image.
Compared with directly carrying out road surface segmentation using the characteristic image after convolutional calculation in the prior art, by the characteristic pattern after filtering processing
As being spliced with primitive character image, it is equivalent to the branch characteristic image for adding one without filtering processing, is remained
Therefore passback gradient in characteristic image remains gradient in the dividing processing of road surface, improve the generalization ability of model, from
And when carrying out the deep learning calculating in the segmentation of road surface based on target signature image, it is able to ascend point of image in the segmentation of road surface
Cut accuracy.
In a kind of possible embodiment, in method provided in an embodiment of the present invention, multiple Predetermined filters include:
One Predetermined filter, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein the first default filtering
Filtering matrix is the matrix of N × 1 in device, and filtering matrix is the matrix of 1 × N, third Predetermined filter in the second Predetermined filter
Interior filtering matrix is the matrix of N × 1, and filtering matrix is the matrix of 1 × N in the 4th Predetermined filter, and N is the natural number greater than 1.
Deep learning method provided in an embodiment of the present invention for road surface segmentation, the filtering square of multiple filters of selection
Battle array is N × 1 and the filter that 1 × N occurs in groups, and letter has lacked the calculation amount of characteristic image, improved the place of processing feature image
Manage speed.
In a kind of possible embodiment, in method provided in an embodiment of the present invention, multiple Predetermined filters include:
One Predetermined filter, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein the first default filtering
Filtering matrix is the matrix of 1 × N in device, and filtering matrix is the matrix of N × 1, third Predetermined filter in the second Predetermined filter
Interior filtering matrix is the matrix of 1 × N, and filtering matrix is the matrix of N × 1 in the 4th Predetermined filter, and N is the natural number greater than 1.
Deep learning method provided in an embodiment of the present invention for road surface segmentation, the filtering square of multiple filters of selection
Battle array is N × 1 and the filter that 1 × N occurs in groups, and letter has lacked the calculation amount of characteristic image, improved the place of processing feature image
Manage speed.
In a kind of possible embodiment, in method provided in an embodiment of the present invention, by the characteristic pattern after filtering processing
As being spliced with primitive character image, target signature image is generated, comprising:
It will splice on channel dimension in characteristic image after filtering processing with primitive character image, it is special to generate target
Image is levied, the port number of target signature image is the sum of the characteristic image after primitive character image channel number and filtering processing.
Deep learning method provided in an embodiment of the present invention for road surface segmentation, is filtered by primitive character image
It after processing, will also splice on channel dimension in the characteristic image after filtering processing with primitive character image, generate target
Characteristic image increases the data volume of target signature image, remains the gradient information of target signature image, thus being based on mesh
When marking the deep learning calculating in characteristic image progress road surface segmentation, it is able to ascend the accuracy of separation of image in the segmentation of road surface.
Second aspect, the embodiment of the present invention provide a kind of deep learning device for road surface segmentation, comprising:
Processing unit, for being successively filtered using multiple Predetermined filters to primitive character image, and it is each
At least one in the row and column of filtering matrix is greater than 1 in Predetermined filter, and primitive character image is to comprising traffic route information
Original image carry out sampling processing after generate;
It is special to generate target for splicing the characteristic image after filtering processing with primitive character image for concatenation unit
Levy image;
Computing unit, for carrying out road surface separation calculation based on target signature image.
In a kind of possible embodiment, in device provided in an embodiment of the present invention, multiple Predetermined filters include:
One Predetermined filter, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein the first default filtering
Filtering matrix is the matrix of N × 1 in device, and filtering matrix is the matrix of 1 × N, third Predetermined filter in the second Predetermined filter
Interior filtering matrix is the matrix of N × 1, and filtering matrix is the matrix of 1 × N in the 4th Predetermined filter, and N is the natural number greater than 1.
In a kind of possible embodiment, in device provided in an embodiment of the present invention, multiple Predetermined filters include:
One Predetermined filter, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein the first default filtering
Filtering matrix is the matrix of 1 × N in device, and filtering matrix is the matrix of N × 1, third Predetermined filter in the second Predetermined filter
Interior filtering matrix is the matrix of 1 × N, and filtering matrix is the matrix of N × 1 in the 4th Predetermined filter, and N is the natural number greater than 1.
In a kind of possible embodiment, in device provided in an embodiment of the present invention, concatenation unit is specifically used for: will filter
Spliced on channel dimension in wave treated characteristic image with primitive character image, generates target signature image, target
The port number of characteristic image is the sum of the characteristic image after primitive character image channel number and filtering processing.
The third aspect, the embodiment of the invention provides a kind of deep learning equipment for road surface segmentation, comprising: at least one
A processor, at least one processor and computer program instructions stored in memory, when computer program instructions quilt
The method that first aspect of the embodiment of the present invention provides is realized when processor executes.
Fourth aspect, the embodiment of the invention provides a kind of computer readable storage mediums, are stored thereon with computer journey
The method that first aspect of the embodiment of the present invention provides is realized in sequence instruction when computer program instructions are executed by processor.
Detailed description of the invention
Fig. 1 is a kind of schematic flow diagram of deep learning method for road surface segmentation provided in an embodiment of the present invention;
Fig. 2 is a kind of structural schematic diagram of deep learning device for road surface segmentation provided in an embodiment of the present invention;
Fig. 3 is a kind of structural schematic diagram of deep learning equipment for road surface segmentation provided in an embodiment of the present invention.
Specific embodiment
With reference to the accompanying drawing, to the deep learning method, apparatus provided in an embodiment of the present invention for road surface segmentation, equipment
And the specific embodiment of medium is described in detail.
As shown in Figure 1, the embodiment of the present invention provides a kind of deep learning method for road surface segmentation, including following step
It is rapid:
S101 is successively filtered primitive character image using multiple Predetermined filters, and each default filtering
At least one in the row and column of filtering matrix is greater than 1 in device, and primitive character image is to the original graph comprising traffic route information
As being generated after carrying out sampling processing.
When it is implemented, when acquisition includes the original image of traffic route information, it can be real-time by the camera configured
Image is acquired, (for example, obtaining every 10 milliseconds primary) can also be periodically acquired, it is not limited in the embodiment of the present invention.
It should be noted that the convolutional calculation of arbitrary number of times can be used when carrying out sampling processing to original image, it can also
So as to carry out sampling processing in other ways, it is not limited in the embodiment of the present invention.
Primitive character image is successively filtered when it is implemented, more than one filter can be used, often
At least one in the row and column of filtering matrix is greater than 1 in a Predetermined filter.For example, two Predetermined filters can be selected,
Greater number of Predetermined filter can be selected, or the multiple Predetermined filters arranged in pairs or groups in pairs, the present invention couple can also be selected
This is without limitation.
In a kind of possible embodiment, four filters are selected, filtering matrix is respectively 3 × 1 in Predetermined filter
Matrix, 1 × 3 matrix, 3 × 1 matrix, filtering matrix is respectively 1 × 3 matrix in 1 × 3 matrix or filter,
3 × 1 matrix, 1 × 3 matrix, 3 × 1 matrix.Wherein the filtering data in four filters is chosen according to actual needs,
It is non-interference each other.
When it is implemented, may be incorporated into activation primitive after filtering each time and swash to filtered characteristic image
Processing living, activation primitive can be sigmoid function, Relu function and softmax function, and which is not limited by the present invention.
Preferably, it after being filtered each time, introduces Relu activation primitive and filtered characteristic image is activated
Processing.
When it is implemented, the step-length of filtering processing can be set to 2 n times power, wherein N is natural number.When step-length is 2
When, the resolution ratio of the image after filtering processing is the half of the resolution ratio of the characteristic image before filtering processing, when step-length is 4
When, the resolution ratio of the image after filtering processing is a quarter of the resolution ratio of the characteristic image before filtering processing, and so on.
When it is implemented, may be incorporated into empty convolution mechanism when being filtered to primitive character image, introduce
Empty convolution can increase receptive field in the case where not the largest loss pond information, allow filtering output every time all comprising larger
The information of range.It should be noted that empty convolution mechanism can be used the mode in existing method, the embodiment of the present invention to this not
It limits.
S102 splices the primitive character image after filtering processing with characteristic image, generates target signature image.
When it is implemented, the characteristic image after filtering processing is spliced on channel dimension with primitive character image,
Target signature image is generated, the port number of target signature image is the characteristic pattern after primitive character image channel number and filtering processing
The sum of as.
S103 carries out road surface separation calculation based on target signature image.
It should be noted that in the characteristic image processing method of road surface segmentation provided in an embodiment of the present invention, it can be to spy
Sign image carries out the characteristic image being successively filtered described in an above-described embodiment using multiple Predetermined filters
Filtering processing can also be carried out successively carrying out using multiple Predetermined filters described in multiple above-described embodiment by actual demand
The characteristic image of filtering processing is filtered, and it is not limited in the embodiment of the present invention.
As shown in Fig. 2, the embodiment of the present invention provides a kind of deep learning device for road surface segmentation, comprising:
Processing unit 201, for being successively filtered using multiple Predetermined filters to primitive character image, and it is every
At least one in the row and column of filtering matrix is greater than 1 in a Predetermined filter, and primitive character image is believed comprising traffic route
It is generated after the original image progress sampling processing of breath;
Concatenation unit 202 generates target for splicing the characteristic image after filtering processing with primitive character image
Characteristic image;
Computing unit 203, for carrying out road surface separation calculation based on target signature image.
In a kind of possible embodiment, in device provided in an embodiment of the present invention, multiple Predetermined filters include:
One Predetermined filter, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein the first default filtering
Filtering matrix is the matrix of N × 1 in device, and filtering matrix is the matrix of 1 × N, third Predetermined filter in the second Predetermined filter
Interior filtering matrix is the matrix of N × 1, and filtering matrix is the matrix of 1 × N in the 4th Predetermined filter, and N is the natural number greater than 1.
In a kind of possible embodiment, in device provided in an embodiment of the present invention, multiple Predetermined filters include:
One Predetermined filter, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein the first default filtering
Filtering matrix is the matrix of 1 × N in device, and filtering matrix is the matrix of N × 1, third Predetermined filter in the second Predetermined filter
Interior filtering matrix is the matrix of 1 × N, and filtering matrix is the matrix of N × 1 in the 4th Predetermined filter, and N is the natural number greater than 1.
In a kind of possible embodiment, in device provided in an embodiment of the present invention, concatenation unit 202 is specifically used for:
It will splice on channel dimension in characteristic image after filtering processing with primitive character image, generate target signature image,
The port number of target signature image is the sum of the characteristic image after primitive character image channel number and filtering processing.
In addition, in conjunction with the deep learning method and apparatus divided for road surface of Fig. 1-Fig. 2 the embodiment of the present application described
It can be realized by the deep learning equipment divided for road surface.Fig. 3 shows provided by the embodiments of the present application for road surface point
The hardware structural diagram for the deep learning equipment cut.
Deep learning equipment for road surface segmentation may include processor 301 and be stored with computer program instructions
Memory 302.
Specifically, above-mentioned processor 301 may include central processing unit (CPU) or specific integrated circuit
(Application Specific Integrated Circuit, ASIC), or may be configured to implement implementation of the present invention
One or more integrated circuits of example.
Memory 302 may include the mass storage for data or instruction.For example it rather than limits, memory
302 may include hard disk drive (Hard Disk Drive, HDD), floppy disk drive, flash memory, CD, magneto-optic disk, tape or logical
With the combination of universal serial bus (Universal Serial Bus, USB) driver or two or more the above.It is closing
In the case where suitable, memory 302 may include the medium of removable or non-removable (or fixed).In a suitable case, it stores
Device 302 can be inside or outside data processing equipment.In a particular embodiment, memory 302 is nonvolatile solid state storage
Device.In a particular embodiment, memory 302 includes read-only memory (ROM).In a suitable case, which can be mask
ROM, programming ROM (PROM), erasable PROM (EPROM), the electric erasable PROM (EEPROM), electrically-alterable ROM of programming
(EAROM) or the combination of flash memory or two or more the above.
Processor 301 is by reading and executing the computer program instructions stored in memory 302, to realize above-mentioned implementation
Deep learning method of one of the example for road surface segmentation.
In one example, the deep learning equipment for road surface segmentation may also include communication interface 303 and bus 310.
Wherein, as shown in figure 3, processor 301, memory 302, communication interface 303 are connected by bus 310 and complete mutual lead to
Letter.
Communication interface 303 is mainly used for realizing in the embodiment of the present invention between each module, device, unit and/or equipment
Communication.
Bus 310 includes hardware, software or both, and the component for being used for the deep learning equipment of road surface segmentation is coupled to each other
Together.For example it rather than limits, bus may include accelerated graphics port (AGP) or other graphics bus, enhancing industry mark
Quasi- framework (EISA) bus, front side bus (FSB), super transmission (HT) interconnection, Industry Standard Architecture (ISA) bus, infinite bandwidth
Interconnection, low pin count (LPC) bus, memory bus, micro- channel architecture (MCA) bus, peripheral component interconnection (PCI) bus,
PCI-Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association part (VLB) are total
The combination of line or other suitable buses or two or more the above.In a suitable case, bus 310 may include
One or more buses.Although specific bus has been described and illustrated in the embodiment of the present invention, the present invention considers any suitable
Bus or interconnection.
This can execute the depth for road surface segmentation in the embodiment of the present invention for the deep learning equipment that road surface is divided
Learning method is spent, to realize the deep learning method divided for road surface described in conjunction with Fig. 1.
In addition, in conjunction with the deep learning method divided for road surface in above-described embodiment, the embodiment of the present invention be can provide
A kind of computer readable storage medium is realized.Computer program instructions are stored on the computer readable storage medium;The meter
Calculation machine program instruction realizes deep learning method of one of the above-described embodiment for road surface segmentation when being executed by processor.
It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although the preferred embodiment of the application has been described, it is created once a person skilled in the art knows basic
Property concept, then additional changes and modifications can be made to these embodiments.So it includes excellent that the following claims are intended to be interpreted as
It selects embodiment and falls into all change and modification of the application range.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (10)
1. a kind of deep learning method for road surface segmentation characterized by comprising
Primitive character image is successively filtered using multiple Predetermined filters, filtering matrix in each Predetermined filter
Row and column at least one be greater than 1, the primitive character image is adopted to the original image comprising traffic route information
It is generated after sample processing;
Characteristic image after filtering processing is spliced with the primitive character image, generates target signature image;
Road surface separation calculation is carried out based on the target signature image.
2. the method according to claim 1, wherein the multiple Predetermined filter includes: the first default filtering
Device, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein filter in first Predetermined filter
Wave matrix is the matrix of N × 1, and filtering matrix is the matrix of 1 × N, the default filtering of the third in second Predetermined filter
Filtering matrix is the matrix of N × 1 in device, and filtering matrix is the matrix of 1 × N in the 4th Predetermined filter, and N is greater than 1
Natural number.
3. the method according to claim 1, wherein the multiple Predetermined filter includes: the first default filtering
Device, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein filter in first Predetermined filter
Wave matrix is the matrix of 1 × N, and filtering matrix is the matrix of N × 1, the default filtering of the third in second Predetermined filter
Filtering matrix is the matrix of 1 × N in device, and filtering matrix is the matrix of N × 1 in the 4th Predetermined filter, and N is greater than 1
Natural number.
4. the method according to claim 1, wherein the characteristic image by after filtering processing with it is described original
Characteristic image is spliced, and target signature image is generated, comprising:
It will splice on channel dimension in characteristic image after the filtering processing with the primitive character image, generate mesh
Characteristic image is marked, after the port number of the target signature image is the primitive character image channel number and the filtering processing
The sum of characteristic image.
5. a kind of deep learning device for road surface segmentation characterized by comprising
Processing unit, for being successively filtered using multiple Predetermined filters to primitive character image, and it is each default
At least one in the row and column of filtering matrix is greater than 1 in filter, and the primitive character image is to comprising traffic route information
Original image carry out sampling processing after generate;
It is special to generate target for splicing the characteristic image after filtering processing with the primitive character image for concatenation unit
Levy image;
Computing unit, for carrying out road surface separation calculation based on the target signature image.
6. device according to claim 5, which is characterized in that the multiple Predetermined filter includes: the first default filtering
Device, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein filter in first Predetermined filter
Wave matrix is the matrix of N × 1, and filtering matrix is the matrix of 1 × N, the default filtering of the third in second Predetermined filter
Filtering matrix is the matrix of N × 1 in device, and filtering matrix is the matrix of 1 × N in the 4th Predetermined filter, and N is greater than 1
Natural number.
7. device according to claim 5, which is characterized in that the multiple Predetermined filter includes: the first default filtering
Device, the second Predetermined filter, third Predetermined filter and the 4th Predetermined filter, wherein filter in first Predetermined filter
Wave matrix is the matrix of 1 × N, and filtering matrix is the matrix of N × 1, the default filtering of the third in second Predetermined filter
Filtering matrix is the matrix of 1 × N in device, and filtering matrix is the matrix of N × 1 in the 4th Predetermined filter, and N is greater than 1
Natural number.
8. device according to claim 5, which is characterized in that concatenation unit is specifically used for: after the filtering processing
Spliced on channel dimension in characteristic image with the primitive character image, generate target signature image, the target is special
The port number for levying image is the primitive character image channel number and the sum of the characteristic image after the filtering processing.
9. it is a kind of for road surface segmentation deep learning equipment characterized by comprising at least one processor, at least one
The computer program instructions of memory and storage in the memory, when the computer program instructions are by the processor
Such as method of any of claims 1-4 is realized when execution.
10. a kind of computer readable storage medium, is stored thereon with computer program instructions, which is characterized in that when the calculating
Such as method of any of claims 1-4 is realized when machine program instruction is executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910264997.4A CN110136134A (en) | 2019-04-03 | 2019-04-03 | A kind of deep learning method, apparatus, equipment and medium for road surface segmentation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910264997.4A CN110136134A (en) | 2019-04-03 | 2019-04-03 | A kind of deep learning method, apparatus, equipment and medium for road surface segmentation |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110136134A true CN110136134A (en) | 2019-08-16 |
Family
ID=67568882
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910264997.4A Pending CN110136134A (en) | 2019-04-03 | 2019-04-03 | A kind of deep learning method, apparatus, equipment and medium for road surface segmentation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110136134A (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2009017483A1 (en) * | 2007-08-01 | 2009-02-05 | The Trustees Of The University Of Penssylvania | Malignancy diagnosis using content-based image retreival of tissue histopathology |
CN106558058A (en) * | 2016-11-29 | 2017-04-05 | 北京图森未来科技有限公司 | Parted pattern training method, lane segmentation method, control method for vehicle and device |
CN108231190A (en) * | 2017-12-12 | 2018-06-29 | 北京市商汤科技开发有限公司 | Handle the method for image and nerve network system, equipment, medium, program |
CN108664993A (en) * | 2018-04-08 | 2018-10-16 | 浙江工业大学 | A kind of convolutional neural networks image classification method of intensive weight connection |
CN108986029A (en) * | 2018-07-03 | 2018-12-11 | 南京览笛信息科技有限公司 | Character image super resolution ratio reconstruction method, system, terminal device and storage medium |
CN109034210A (en) * | 2018-07-04 | 2018-12-18 | 国家新闻出版广电总局广播科学研究院 | Object detection method based on super Fusion Features Yu multi-Scale Pyramid network |
CN109284704A (en) * | 2018-09-07 | 2019-01-29 | 中国电子科技集团公司第三十八研究所 | Complex background SAR vehicle target detection method based on CNN |
CN109559799A (en) * | 2018-10-12 | 2019-04-02 | 华南理工大学 | The construction method and the model of medical image semantic description method, descriptive model |
-
2019
- 2019-04-03 CN CN201910264997.4A patent/CN110136134A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2009017483A1 (en) * | 2007-08-01 | 2009-02-05 | The Trustees Of The University Of Penssylvania | Malignancy diagnosis using content-based image retreival of tissue histopathology |
CN106558058A (en) * | 2016-11-29 | 2017-04-05 | 北京图森未来科技有限公司 | Parted pattern training method, lane segmentation method, control method for vehicle and device |
CN108231190A (en) * | 2017-12-12 | 2018-06-29 | 北京市商汤科技开发有限公司 | Handle the method for image and nerve network system, equipment, medium, program |
CN108664993A (en) * | 2018-04-08 | 2018-10-16 | 浙江工业大学 | A kind of convolutional neural networks image classification method of intensive weight connection |
CN108986029A (en) * | 2018-07-03 | 2018-12-11 | 南京览笛信息科技有限公司 | Character image super resolution ratio reconstruction method, system, terminal device and storage medium |
CN109034210A (en) * | 2018-07-04 | 2018-12-18 | 国家新闻出版广电总局广播科学研究院 | Object detection method based on super Fusion Features Yu multi-Scale Pyramid network |
CN109284704A (en) * | 2018-09-07 | 2019-01-29 | 中国电子科技集团公司第三十八研究所 | Complex background SAR vehicle target detection method based on CNN |
CN109559799A (en) * | 2018-10-12 | 2019-04-02 | 华南理工大学 | The construction method and the model of medical image semantic description method, descriptive model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108470159A (en) | Lane line data processing method, device, computer equipment and storage medium | |
US20180322230A1 (en) | Driverless vehicle simulation test method and apparatus, device and readable medium | |
CN109270927B (en) | Road data generation method and device | |
CN109472844B (en) | Method and device for marking lane lines in road junction and storage medium | |
JP6817384B2 (en) | Visual sensing methods for autonomous vehicles, visual sensing devices for autonomous vehicles, control devices and computer-readable storage media | |
CN106023622B (en) | A kind of method and apparatus of determining traffic lights identifying system recognition performance | |
CN102867418A (en) | Method and device for judging license plate identification accuracy | |
CN108133484A (en) | Automatic Pilot processing method and processing device based on scene cut, computing device | |
CN111539484A (en) | Method and device for training neural network | |
CN111914845A (en) | Character layering method and device in license plate and electronic equipment | |
CN110136134A (en) | A kind of deep learning method, apparatus, equipment and medium for road surface segmentation | |
CN114973687A (en) | Traffic information processing method, device, equipment and medium | |
CN111489568A (en) | Traffic signal lamp regulation and control method and device and computer readable storage medium | |
CN109784129A (en) | Information output method and device | |
CN110148135A (en) | A kind of road surface dividing method, device, equipment and medium | |
CN110414374B (en) | Method, device, equipment and medium for determining obstacle position and attitude | |
CN114067221B (en) | Remote sensing image woodland extraction method, system, device and medium | |
CN115393827A (en) | Traffic signal lamp state identification method and system, electronic equipment and storage medium | |
CN112734923A (en) | Method, device and equipment for constructing automatic driving three-dimensional virtual scene and storage medium | |
CN109656245B (en) | Method and device for determining brake position | |
CN113343817A (en) | Unmanned vehicle path detection method and device for target area and medium | |
CN110135247B (en) | Data enhancement method, device, equipment and medium in pavement segmentation | |
CN113504734A (en) | Image display-based lane line simulation test method and system | |
CN112288748A (en) | Semantic segmentation network training and image semantic segmentation method and device | |
CN108765258A (en) | Vehicle-mounted image processing method and device, electronic equipment, medium, program product |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190816 |