CN109409428A - Training method, device and the electronic equipment of plank identification and plank identification model - Google Patents
Training method, device and the electronic equipment of plank identification and plank identification model Download PDFInfo
- Publication number
- CN109409428A CN109409428A CN201811251848.6A CN201811251848A CN109409428A CN 109409428 A CN109409428 A CN 109409428A CN 201811251848 A CN201811251848 A CN 201811251848A CN 109409428 A CN109409428 A CN 109409428A
- Authority
- CN
- China
- Prior art keywords
- plank
- sample
- image
- region
- identification
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the present disclosure discloses training method, device and the electronic equipment of a kind of identification of plank and plank identification model.Wherein, the training method of plank identification model includes: the multiple sample images for obtaining sample plank；Wherein, multiple sample images are to shoot the image that the sample plank obtains from multiple and different orientation；Obtain the corresponding annotation results of sample image；Wherein, annotation results include at least first mark in distorted image region in sample image and/or second mark in normal picture region；Model training is carried out using multiple sample images and its corresponding annotation results, obtains plank identification model.Multiple sample images that this mode of the disclosure is obtained based on multiple images sensor are as training data, plank identification model is come out the feature extraction in distorted image region and/or normal picture region, and then can from multiple sample images automatic identification effective image data, and based on effective image data complete timber identification.
Description
Technical field
This disclosure relates to timber identification technology field, and in particular to the training side of a kind of identification of plank and plank identification model
Method, device and electronic equipment.
Background technique
In plank process automation equipment, the image data that plank is obtained using imaging sensor is a kind of common side
Method.With the development of artificial intelligence technology, the especially development of deep neural network, convolutional neural networks, the performance of machine vision
More and more applied.Timber process automation equipment can realize high-precision point based on the image recognition to timber
Choosing, defect recognition, processing planning.However, in some scenes, such as the plank after processing, or the plank using antique imitation process
In there may be the curved surface on surface and be covered with artificial spray painting.In this case, even if using scattering light source as outside
Light source also be easy to cause the surface reflection of plank, due to current machine vision and cannot be distinguished timber itself feature or by
In reflective bring characteristics of image, therefore this retroreflective feature will affect the recognition effect to timber.In addition, some processing technologys
It is to be realized by changing surface texture form, antique imitation process as escribed above processes in board surface certain sometimes
Raised grain, to generate specific visual effect.Therefore, this subtle difference can not also be captured by two dimensional image.
Summary of the invention
The embodiment of the present disclosure provide the training method of a kind of identification of plank and plank identification model, device, electronic equipment and
Computer readable storage medium.
In a first aspect, providing a kind of training method of plank identification model in the embodiment of the present disclosure.
Specifically, the training method of the plank identification model, comprising:
Obtain multiple sample images of sample plank；Wherein, the multiple sample image is to shoot from multiple and different orientation
The image that the sample plank obtains；
Obtain the corresponding annotation results of the sample image；Wherein, the annotation results include at least the sample image
First mark in middle distorted image region and/or second mark in normal picture region；
Model training is carried out using multiple sample images and its corresponding annotation results, obtains plank identification mould
Type.
Further, the normal picture region in multiple sample images can at least cover the sample plank
Whole surface region.
Further, the annotation results further include at least one third mark, can pass through the sample for identifying
The feature that the normal picture in the whole surface region of plank identifies.
Further, the distorted image region is the region influenced by light, and the normal picture region is not light
The region that line influences.
Second aspect provides a kind of plank recognition methods in the embodiment of the present disclosure.
Specifically, the plank recognition methods, comprising:
Obtain multiple images to be recognized of plank to be identified；Wherein, the multiple images to be recognized is from multiple and different sides
Position shoots the image that the plank to be identified obtains；
Multiple images to be recognized are input in plank identification model, identify the normogram in the images to be recognized
As region；
According to the normal picture region in multiple images to be recognized, obtained by the plank identification model described wait know
The recognition result of other plank.
Further, the plank identification model identifies the covering plank to be identified from the multiple images to be recognized
Whole surface region normal picture.
Further, the recognition result for obtaining the plank to be identified includes: the plank identification model from covering
The normal picture in the whole surface region of the plank to be identified identifies at least one feature of the plank to be identified.
The third aspect provides a kind of training device of plank identification model in the embodiment of the present disclosure.
Specifically, the training device of the plank identification model, comprising:
First obtains module, is configured as obtaining multiple sample images of sample plank；Wherein, the multiple sample image
To shoot the image that the sample plank obtains from multiple and different orientation；
Second obtains module, is configured as obtaining the corresponding annotation results of the sample image；Wherein, the annotation results
Second mark of the first mark and/or normal picture region including at least distorted image region in the sample image；
Training module is configured as carrying out model instruction using multiple sample images and its corresponding annotation results
Practice, obtains plank identification model.
Further, the normal picture region in multiple sample images can at least cover the sample plank
Whole surface region.
Further, the annotation results further include at least one third mark, can pass through the sample for identifying
The feature that the normal picture in the whole surface region of plank identifies.
Further, the distorted image region is the region influenced by light, and the normal picture region is not light
The region that line influences.
Fourth aspect provides a kind of plank identification device in the embodiment of the present disclosure.
Specifically, the plank identification device, comprising:
Third obtains module, is configured as obtaining multiple images to be recognized of plank to be identified；Wherein, the multiple wait know
Other image is to shoot the image that the plank to be identified obtains from multiple and different orientation；
First identification module is configured as multiple images to be recognized being input in plank identification model, identifies institute
State the normal picture region in images to be recognized；
Second identification module is configured as according to the normal picture region in multiple images to be recognized, by the wood
Plate identification model obtains the recognition result of the plank to be identified.
Further, the plank identification model identifies the covering plank to be identified from the multiple images to be recognized
Whole surface region normal picture.
Further, second identification module, comprising:
The plank identification model identifies institute from the normal picture in the whole surface region for covering the plank to be identified
State at least one feature of plank to be identified.
The function can also execute corresponding software realization by hardware realization by hardware.The hardware or
Software includes one or more modules corresponding with above-mentioned function.
In a possible design, wrapped in the training device of the plank identification model or the structure of plank identification device
Memory and processor are included, the memory is used to store the training device or plank of one or more support plank identification model
Identification device device executes plank recognition methods in the training method or second aspect of plank identification model in above-mentioned first aspect
Computer instruction, the processor is configured to for executing the computer instruction stored in the memory.The plank
The training device or plank identification device of identification model can also include communication interface, the training device for plank identification model
Or plank identification device and other equipment or communication.
5th aspect, the embodiment of the present disclosure provide a kind of electronic equipment, including memory and processor；Wherein, described
Memory is for storing one or more computer instruction, wherein one or more computer instruction is by the processor
It executes to realize method and step described in first aspect.
6th aspect, the embodiment of the present disclosure provide a kind of computer readable storage medium, for storing plank identification mould
Computer instruction used in the training device or plank identification device of type, it includes know for executing plank in above-mentioned first aspect
The training method of other model or or second aspect in computer instruction involved in plank recognition methods.
The technical solution that the embodiment of the present disclosure provides can include the following benefits:
The embodiment of the present disclosure obtains sample image by shooting sample plank from multiple orientation, and then marks to sample image
Distorted image region and/or normal picture region, and machine learning model is trained using sample image and annotation results,
Obtain the plank identification model that can at least identify distorted image region and/or normal picture region.This mode base of the disclosure
In multiple sample images that multiple images sensor obtains as training data, allow plank identification model by distorted image
The feature extraction in region and/or normal picture region comes out, so can from multiple sample images automatic identification effective image
Data, and the identification based on effective image data completion timber.
The embodiment of the present disclosure also obtains images to be recognized by shooting plank to be identified from multiple orientation, and then using in advance
The normal picture region in images to be recognized is identified in the plank identification model of train number, and obtains institute using normal picture region
State the recognition result of plank to be identified.This mode of the disclosure is carried out by the image of the plank to be identified obtained to multiple orientation
Identification can recognize that normal picture region namely effective image data, and then be completed according to effective image data to timber
Identification, discrimination are higher.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The disclosure can be limited.
Detailed description of the invention
In conjunction with attached drawing, by the detailed description of following non-limiting embodiment, the other feature of the disclosure, purpose and excellent
Point will be apparent.In the accompanying drawings:
Fig. 1 shows the flow chart of the training method of the plank identification model according to one embodiment of the disclosure；
Fig. 2 shows the acquisition structural schematic diagrams of the sample image according to involved in one embodiment of the disclosure；
Fig. 3 is shown according to the schematic diagram for obtaining sample image in one embodiment of the disclosure；
Fig. 4 is shown to be illustrated according to the effect of the different sample images of the same timber obtained in one embodiment of the disclosure
Figure；
Fig. 5 (a)~(b) is shown to be shown according to the effect of the sample image obtained and annotation results in one embodiment of the disclosure
It is intended to；
Fig. 6 shows the flow chart of the plank recognition methods according to one embodiment of the disclosure；
Fig. 7 is shown according to the effect diagram for identifying plank defect to be identified in one embodiment of the disclosure；
Fig. 8 shows the structural block diagram of the training device of the plank identification model according to one embodiment of the disclosure；
Fig. 9 shows the structural block diagram of the plank identification device according to one embodiment of the disclosure；
Figure 10 is adapted for for realizing according to the plank recognition methods of one embodiment of the disclosure or plank identification model
The structural schematic diagram of the electronic equipment of training method.
Specific embodiment
Hereinafter, the illustrative embodiments of the disclosure will be described in detail with reference to the attached drawings, so that those skilled in the art can
Easily realize them.In addition, for the sake of clarity, the portion unrelated with description illustrative embodiments is omitted in the accompanying drawings
Point.
In the disclosure, it should be appreciated that the term of " comprising " or " having " etc. is intended to refer to disclosed in this specification
Feature, number, step, behavior, the presence of component, part or combinations thereof, and be not intended to exclude other one or more features,
A possibility that number, step, behavior, component, part or combinations thereof exist or are added.
It also should be noted that in the absence of conflict, the feature in embodiment and embodiment in the disclosure
It can be combined with each other.The disclosure is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Fig. 1 shows the flow chart of the training method of the plank identification model according to one embodiment of the disclosure.Such as Fig. 1 institute
Show, the training method of the plank identification model includes the following steps S101-S103:
In step s101, multiple sample images of sample plank are obtained；Wherein, the multiple sample image is from multiple
Different direction shoots the image that the sample plank obtains；
In step s 102, the corresponding annotation results of the sample image are obtained；Wherein, the annotation results include at least
First mark in distorted image region and/or second mark in normal picture region in the sample image；
In step s 103, model training is carried out using multiple sample images and its corresponding annotation results,
Obtain plank identification model.
In the present embodiment, multiple images sensor can be used from multiple and different orientation and obtain the more of same sample floor
A sample image.Fig. 2 shows a kind of acquisition structural schematic diagram of sample image involved in disclosure, one of timber,
Such as timber floor is sample floor.It is provided with multiple images sensor above it for obtaining the image data of timber, image
Sensor is equipped with lens assembly and picture processing chip (image processing system), wherein picture processing chip
The image data tentatively obtained can be pre-processed, such as white balance, noise reduction, go the processing such as distortion.
After obtaining corresponding multiple sample images for each sample plank, each sample image can be marked
Note.The content of mark includes the first mark and/or the second mark, and first is noted for indicating the distorted image area in sample image
Domain, and second is noted for indicating the normal picture region in sample image.In some embodiments, the first mark can only be marked
Note or the second mark, and some other embodiment can mark the first mark and the second mark simultaneously, with specific reference to actual conditions
Setting.Distorted image region can be due to the uncertain factors such as shooting angle, environmental parameter cause obtain image data with
The true biggish region of data differences, such as retroreflective regions.
In some embodiments, some certain types of timber are to have passed through specific surface treatment, such as specifically spray
Paint or surface engraving technique, so that the image that an imaging sensor obtains can not be suitable for subsequent processing.As shown in figure 3,
When imaging sensor obtain an archaized wood floor image after, a portion region due to floor surface itself texture and
Spray painting, shows as retroreflective regions.In subsequent treatment process, the image data of the retroreflective regions can not then carry the spy of plank
Sign, therefore be unfavorable for any identification mission and result is accurately identified.From image data angle, retroreflective regions correspond to figure
As data value reaches the maximum value of dynamic range, therefore, it is intended that data lack the statistical difference opposite sex, effective use can not be carried
In the data of identification mission.However, due to it is this it is reflective be due to artificially processing bring effect to timber, such as in order to reach
The texture of certain smooth surface or certain consistency, so that its retroreflective feature has certain space consistency.This consistency
It is also utilized when traditional artificial identification, such as observer is by adjusting the angle of observation timber, so that it may will
Retroreflective regions remove the region currently paid close attention to, and then are segmented the identification for completing monolith timber.In the disclosure, as shown in Figure 2
The sensor that space is distributed by different location will obtain similar multiple images data, position of each image data due to sensor
It sets different and has obtained the effective image data of the different corresponding regions of the same plank.That is, although board surface adds
Some image data carries inefficient sensing data in the image that work technique causes a certain imaging sensor to obtain, still
This inefficient data corresponding position in different sensors is different.As shown in figure 4, same timber is in imaging sensor 1 and figure
As the retroreflective regions generated in sensor 2 are respectively retroreflective regions 1 and retroreflective regions 2.The reason of this effect occur is, wood
Material surface processing technique has certain consistency, so cause to obtain different light distribution at different spatial,
At first space, incident light is largely concentrated on first area by the first retroreflective regions, and identical incident light is largely concentrated on
The second area of second space position.This light distribution forms corresponding different images by different image capture sensors
Different retroreflective regions.Therefore, the disclosure marks distortion map therein after the multiple sample images for obtaining same sample plank
As region and normal picture region, so that the plank identification model that training obtains can at least can recognize that the distortion in image
Image-region and normal picture region facilitate the subsequent identification that other tasks are carried out to plank.
It in some embodiments, can only include distorted image region and/or normogram in the annotation results of sample image
As the mark in region, so that the plank identification model that training obtains only identifies distorted image region and/or normogram in image
As region.In further embodiments, the annotation results of sample image are in addition to distorted image region and/or normal picture region
It can also include the corresponding annotation results of other tasks except mark, such as identification mission can be the inspection of the defect to timber
Survey, to the classification of the color and vein of timber, to planning of timber cutting scheme etc., be labeled with specific reference to different purposes；
In these embodiments, plank identification model in addition to can recognize that in image other than distorted image region and normal picture region,
Other tasks can also be executed based on normal picture region.
In the present embodiment, by using the sample image obtained based on multiple images sensor as training data, machine
Model can come out for example reflective feature extraction in distorted image region, allow machine learning model from multiple sample graphs
Automatic identification effective image data namely normal image data as in, and the identification based on effective image data completion plank.Such as
Shown in Fig. 5 (a)~(b), each training sample includes multiple sample images of the same plank obtained from multiple images sensor
Data, annotation results mark each sample image.In some embodiments, annotation results can also be comprising to the wood
The mark of one identification mission of plate, the corresponding plank of first group of data is classification 1 in Fig. 5 (a), second group in Fig. 5 (b)
The corresponding plank of data is classification 2.Wherein, the mark in region can for by way of artificial cognition, select picture quality compared with
High region is labeled as normal picture region.For example, the retroreflective regions in a sample image are set as normogram
As except region.The first mark or the second mark of multiple sample images are determined by image pattern itself, that is, each sample
The first mark or the second mark of image may not be identical.After obtaining a large amount of training samples, a depth mind can be trained
Through machine learning models such as network, convolutional neural networks, and then obtain the plank identification model for being directed to a certain task.Such as
A model for the identification of plank classification as shown in Fig. 5 (a)~(b).
The embodiment of the present disclosure obtains sample image by shooting sample plank from multiple orientation, and then marks to sample image
Distorted image region and/or normal picture region, and machine learning model is trained using sample image and annotation results,
Obtain the plank identification model that can at least identify distorted image region and/or normal picture region.This mode base of the disclosure
In multiple sample images that multiple images sensor obtains as training data, allow plank identification model by distorted image
The feature extraction in region and/or normal picture region comes out, so can from multiple sample images automatic identification effective image
Data, and the identification based on effective image data completion timber.
In an optional implementation of the present embodiment, the normal picture region in multiple sample images is extremely
The whole surface region of the sample plank can be covered less.
In the optional implementation, in multiple sample images acquired in same sample plank, normal picture
Region, which stacks up, can at least cover the whole surface of the sample plank, and such machine learning model can have been identified therefrom
Whole board surface image.In some embodiments, the normal picture in the corresponding multiple sample images of same sample plank
Region can not be overlapped (sample 1 in such as Fig. 5 (a)), and can cover the surface of entire sample plank, and in other realities
It applies in example, the normal raised areas in the corresponding multiple sample images of same sample plank can be overlapped (in such as Fig. 5 (b)
Sample 2), and the surface of entire sample plank can be covered.
In an optional implementation of the present embodiment, the annotation results further include at least one third mark, are used
In the feature that mark can be identified by the normal picture in the whole surface region of the sample plank.
In the optional implementation, in the annotation results of sample plank in addition to include first mark and/or second
Mark is outer, can also include that third marks, for indicating other features of the sample plank, such as classification, defect area, defect
Classification etc. can specifically be determined according to the task of plank identification model.Plank identification model can be identified first from image
Normal picture region, and multiple normal picture regions are stacked up and form the whole surface region of plank, and the whole surface
The image data in region is valid data, later again by being identified to obtain recognition result to whole surface region.
In an optional implementation of the present embodiment, the distorted image region is the region influenced by light, institute
Stating normal picture region is the region not influenced by light.
In the optional implementation, normal picture region is the image-region for capableing of actual response board surface feature,
Namely the image data in the normal picture region is valid data；And the image data in distorted image region is invalid number
According to, can not the feature of actual response board surface influenced to lead as reflective by light for example, when shooting board surface image
Cause the region of image data distortion.
Description through the foregoing embodiment is it is found that a machine learning model can be obtained based on multiple images sensor
Multiple sample images of same sample plank, and extract from multiple sample images to the contributive picture number of identification mission automatically
According to (normal picture region).One trained plank identification model can be given up automatically since Wood surface processing technology is brought
The reflective equal effect influenced by light, while whole surface images can be always extracted from multiple images again, Jin Ershi
Now accurately recognition effect.By similar methods, machine learning model can obtain different type based on multiple images sample
Evident characteristics.For example, being chosen by specific mask method and model, machine learning model can capture wood surface engraving
Cubic texture feature, and then to the timber carry out accurately identification etc..
Fig. 6 shows the flow chart of the plank recognition methods according to one embodiment of the disclosure.As shown in fig. 6, the plank
Recognition methods includes the following steps S601-S603:
In step s 601, multiple images to be recognized of plank to be identified are obtained；Wherein, the multiple images to be recognized is
The image that the plank to be identified obtains is shot from multiple and different orientation；
In step S602, multiple images to be recognized are input in plank identification model, are identified described to be identified
Normal picture region in image；
In step S603, according to the normal picture region in multiple images to be recognized, mould is identified by the plank
Type obtains the recognition result of the plank to be identified.
In the present embodiment, when identifying plank to be identified, shoot to obtain wood to be identified by multiple and different orientation
Multiple images to be recognized of plate surface, and multiple images to be recognized are input in trained plank identification model, by
Plank identification model obtains normal picture region namely effective image data from multiple images to be recognized, and then further according to just
Normal image-region obtains the recognition result of the plank to be identified.Recognition result can according to actual needs depending on, and according to reality
Border demand trains plank identification model, such as classification, defect etc. that plank identification model can be trained to identify plank.Tool
Body details can be found in the description of the training method of above-mentioned plank identification model, and details are not described herein.
In an optional implementation of the present embodiment, the plank identification model is known from the multiple images to be recognized
The normal picture in the whole surface region of the plank to be identified Chu not covered.
In the optional implementation, plank identification model first can identify normal picture from multiple images to be recognized
Region, and then spliced this normal picture region to obtain the normal image data in plank whole surface to be identified region again,
Enable the subsequent identification to plank to be identified more accurate, avoids causing since image data is influenced by factors such as light
Identify the problem of mistake.
In an optional implementation of the present embodiment, the recognition result for obtaining the plank to be identified includes:
The plank identification model identifies institute from the normal picture in the whole surface region for covering the plank to be identified
State at least one feature of plank to be identified.
In the optional implementation, plank identification model is in the normal picture number for obtaining the whole surface of plank to be identified
According to rear, at least one feature of plank to be identified, such as classification, the defect of plank to be identified are identified from normal image data
Deng.
As shown in fig. 7, in a plank to be identified, it is understood that there may be generated in a defect, such as process recessed
It falls into.However, in the first image data (image 1), due to the angle of imaging sensor or reflective, the characteristics of image of the defect is simultaneously
Do not captured by image.And in the second image data (image 2), then due to different imaging sensor angles, this feature
It is captured by image.If image 1 is directly input to plank identification model, plank identification model, which provides, does not have defective knot
Fruit.However, according to the method that the disclosure provides, since how the study of plank identification model is to extracting from the multiple images data
The ability of key feature, therefore output the result of defect.Similarly, defect shown in Fig. 7 is also possible to a specific texture,
Such as the Logo protrusion specially carved.Corresponding, trained machine learning model can be obtained by multiple images data
The recognition result of the texture, for example, the texture whether qualification etc..
Following is embodiment of the present disclosure, can be used for executing embodiments of the present disclosure.
Fig. 8 shows the structural block diagram of the training device of the plank identification model according to one embodiment of the disclosure, the device
It being implemented in combination with as some or all of of electronic equipment by software, hardware or both.As shown in figure 8, described
The training device of plank identification model includes that the first acquisition module 801, second obtains module 802 and training module 803:
First obtains module 801, is configured as obtaining multiple sample images of sample plank；Wherein, the multiple sample
Image is to shoot the image that the sample plank obtains from multiple and different orientation；
Second obtains module 802, is configured as obtaining the corresponding annotation results of the sample image；Wherein, the mark
As a result first mark in distorted image region in the sample image and/or second mark in normal picture region are included at least；
Training module 803 is configured as carrying out mould using multiple sample images and its corresponding annotation results
Type training obtains plank identification model.
In an optional implementation of the present embodiment, the normal picture region in multiple sample images is extremely
The whole surface region of the sample plank can be covered less.
In an optional implementation of the present embodiment, the annotation results further include at least one third mark, are used
In the feature that mark can be identified by the normal picture in the whole surface region of the sample plank.
In an optional implementation of the present embodiment, the distorted image region is the region influenced by light, institute
Stating normal picture region is the region not influenced by light.
It is proposed in the training device and embodiment illustrated in fig. 1 and related embodiment of the plank identification model that the present embodiment proposes
Plank identification model training method it is corresponding consistent, detail can be found in the above-mentioned training method to plank identification model
Description, details are not described herein.
Fig. 9 shows the structural block diagram of the plank identification device according to one embodiment of the disclosure, which can be by soft
Part, hardware or both are implemented in combination with as some or all of of electronic equipment.As shown in figure 9, the plank identifies mould
The training device of type includes that third obtains module 901, the first identification module 902 and the second identification module 903:
Third obtains module 901, is configured as obtaining multiple images to be recognized of plank to be identified；Wherein, the multiple
Images to be recognized is to shoot the image that the plank to be identified obtains from multiple and different orientation；
First identification module 902 is configured as multiple images to be recognized being input in plank identification model, identification
Normal picture region in the images to be recognized；
Second identification module 903 is configured as according to the normal picture region in multiple images to be recognized, by described
Plank identification model obtains the recognition result of the plank to be identified.
In an optional implementation of the present embodiment, the plank identification model is known from the multiple images to be recognized
The normal picture in the whole surface region of the plank to be identified Chu not covered.
In an optional implementation of the present embodiment, second identification module 903, comprising:
The plank identification model identifies institute from the normal picture in the whole surface region for covering the plank to be identified
State at least one feature of plank to be identified.
The plank proposed in the plank identification device and embodiment illustrated in fig. 6 and related embodiment that the present embodiment proposes identifies
The training method of model is corresponding consistent, and detail can be found in the above-mentioned description to plank recognition methods, and details are not described herein.
Figure 10 is adapted for for realizing the plank recognition methods according to disclosure embodiment or the instruction of plank identification model
Practice the structural schematic diagram of the electronic equipment of method.
As shown in Figure 10, electronic equipment 1000 includes central processing unit (CPU) 1001, can be read-only according to being stored in
Program in memory (ROM) 1002 is loaded into the journey in random access storage device (RAM) 1003 from storage section 1008
Sequence and execute the various processing in above-mentioned embodiment shown in FIG. 1.In RAM1003, it is also stored with the behaviour of electronic equipment 1000
Various programs and data needed for making.CPU1001, ROM1002 and RAM1003 are connected with each other by bus 1004.Input/defeated
(I/O) interface 1005 is also connected to bus 1004 out.
I/O interface 1005 is connected to lower component: the importation 1006 including keyboard, mouse etc.；Including such as cathode
The output par, c 1007 of ray tube (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.；Storage section including hard disk etc.
1008；And the communications portion 1009 of the network interface card including LAN card, modem etc..Communications portion 1009 passes through
Communication process is executed by the network of such as internet.Driver 1010 is also connected to I/O interface 1005 as needed.It is detachable to be situated between
Matter 1011, such as disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on as needed on driver 1010, so as to
In being mounted into storage section 1008 as needed from the computer program read thereon.
Particularly, according to embodiment of the present disclosure, it is soft to may be implemented as computer above with reference to Fig. 1 method described
Part program.For example, embodiment of the present disclosure includes a kind of computer program product comprising be tangibly embodied in and its readable
Computer program on medium, the computer program include for executing Fig. 1 ... the program code of method.Such
In embodiment, which can be downloaded and installed from network by communications portion 1009, and/or from detachable
Medium 1011 is mounted.
Flow chart and block diagram in attached drawing illustrate system, method and computer according to the various embodiments of the disclosure
The architecture, function and operation in the cards of program product.In this regard, each box in course diagram or block diagram can be with
A part of a module, section or code is represented, a part of the module, section or code includes one or more
Executable instruction for implementing the specified logical function.It should also be noted that in some implementations as replacements, institute in box
The function of mark can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are practical
On can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it wants
It is noted that the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart, Ke Yiyong
The dedicated hardware based system of defined functions or operations is executed to realize, or can be referred to specialized hardware and computer
The combination of order is realized.
Being described in unit or module involved in disclosure embodiment can be realized by way of software, can also
It is realized in a manner of through hardware.Described unit or module also can be set in the processor, these units or module
Title do not constitute the restriction to the unit or module itself under certain conditions.
As on the other hand, the disclosure additionally provides a kind of computer readable storage medium, the computer-readable storage medium
Matter can be computer readable storage medium included in device described in above embodiment；It is also possible to individualism,
Without the computer readable storage medium in supplying equipment.Computer-readable recording medium storage has one or more than one journey
Sequence, described program is used to execute by one or more than one processor is described in disclosed method.
Above description is only the preferred embodiment of the disclosure and the explanation to institute's application technology principle.Those skilled in the art
Member is it should be appreciated that invention scope involved in the disclosure, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from the inventive concept, it is carried out by above-mentioned technical characteristic or its equivalent feature
Any combination and the other technical solutions formed.Such as features described above has similar function with (but being not limited to) disclosed in the disclosure
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (10)
1. a kind of training method of plank identification model characterized by comprising
Obtain multiple sample images of sample plank；Wherein, the multiple sample image is from described in the shooting of multiple and different orientation
The image that sample plank obtains；
Obtain the corresponding annotation results of the sample image；Wherein, the annotation results include at least loses in the sample image
First mark of true image-region and/or second mark in normal picture region；
Model training is carried out using multiple sample images and its corresponding annotation results, obtains plank identification model.
2. the method according to claim 1, wherein the normal picture region in multiple sample images
The whole surface region of the sample plank can at least be covered.
3. the method according to claim 1, wherein the annotation results further include at least one third mark,
For identifying the feature that can be identified by the normal picture in the whole surface region of the sample plank.
4. the method according to claim 1, wherein the distorted image region is the region influenced by light,
The normal picture region is the region not influenced by light.
5. a kind of plank recognition methods characterized by comprising
Obtain multiple images to be recognized of plank to be identified；Wherein, the multiple images to be recognized is to clap from multiple and different orientation
Take the photograph the image that the plank to be identified obtains；
Multiple images to be recognized are input in plank identification model, identify the normal picture area in the images to be recognized
Domain；
According to the normal picture region in multiple images to be recognized, the wood to be identified is obtained by the plank identification model
The recognition result of plate.
6. according to the method described in claim 5, it is characterized in that, the plank identification model is from the multiple images to be recognized
Identify the normal picture for covering the whole surface region of the plank to be identified.
7. a kind of training device of plank identification model characterized by comprising
First obtains module, is configured as obtaining multiple sample images of sample plank；Wherein, the multiple sample image be from
Multiple and different orientation shoot the image that the sample plank obtains；
Second obtains module, is configured as obtaining the corresponding annotation results of the sample image；Wherein, the annotation results are at least
Second mark of the first mark and/or normal picture region including distorted image region in the sample image；
Training module is configured as carrying out model training using multiple sample images and its corresponding annotation results,
Obtain plank identification model.
8. a kind of plank identification device characterized by comprising
Third obtains module, is configured as obtaining multiple images to be recognized of plank to be identified；Wherein, the multiple figure to be identified
As being to shoot the image that the plank to be identified obtains from multiple and different orientation；
First identification module is configured as multiple images to be recognized being input in plank identification model, identification it is described to
Identify the normal picture region in image；
Second identification module is configured as being known according to the normal picture region in multiple images to be recognized by the plank
Other model obtains the recognition result of the plank to be identified.
9. a kind of electronic equipment, which is characterized in that including memory and processor；Wherein,
The memory is for storing one or more computer instruction, wherein one or more computer instruction is by institute
Processor is stated to execute to realize method and step described in any one of claims 1-6.
10. a kind of computer readable storage medium, is stored thereon with computer instruction, which is characterized in that the computer instruction quilt
Processor realizes method and step described in any one of claims 1-6 when executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811251848.6A CN109409428A (en) | 2018-10-25 | 2018-10-25 | Training method, device and the electronic equipment of plank identification and plank identification model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811251848.6A CN109409428A (en) | 2018-10-25 | 2018-10-25 | Training method, device and the electronic equipment of plank identification and plank identification model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109409428A true CN109409428A (en) | 2019-03-01 |
Family
ID=65469602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811251848.6A Pending CN109409428A (en) | 2018-10-25 | 2018-10-25 | Training method, device and the electronic equipment of plank identification and plank identification model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409428A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110059549A (en) * | 2019-03-11 | 2019-07-26 | 齐鲁工业大学 | A kind of thin wood plate categorizing system and algorithm based on deep learning |
| CN110068578A (en) * | 2019-05-17 | 2019-07-30 | 苏州图迈蓝舸智能科技有限公司 | A kind of visual defects detection method, device and the terminal device of PVC floor |
| CN111144244A (en) * | 2019-12-13 | 2020-05-12 | 唐美(北京)文化科技发展有限公司 | Furniture wood type identification method and device |
| CN111921885A (en) * | 2020-08-05 | 2020-11-13 | 东南数字经济发展研究院江山分院 | Intelligent detection method and device for wood door plate size |
| CN111951128A (en) * | 2020-08-31 | 2020-11-17 | 江苏工程职业技术学院 | Energy-saving and environment-friendly building construction method and device |
| CN112560537A (en) * | 2020-12-29 | 2021-03-26 | 云南易见纹语科技有限公司 | Article texture processing device, method, system, electronic device and storage medium |
| CN112907498A (en) * | 2019-11-18 | 2021-06-04 | 中国商用飞机有限责任公司 | Pore identification method, device, equipment and storage medium |
| CN113610184A (en) * | 2021-08-19 | 2021-11-05 | 江西应用技术职业学院 | Wood texture classification method based on transfer learning |
| CN114218423A (en) * | 2022-02-21 | 2022-03-22 | 广东联邦家私集团有限公司 | 5G-based non-labeling solid wood board identity digitalization method, device and system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103477861A (en) * | 2013-10-16 | 2014-01-01 | 福建省农业科学院植物保护研究所 | Method for processing export potted chrysalidocarpus lutescens |
| KR20170027580A (en) * | 2015-09-02 | 2017-03-10 | 김석배 | Glare preventing type road sign board and traffic lane discerning device for motors |
| CN106650743A (en) * | 2016-09-12 | 2017-05-10 | 北京旷视科技有限公司 | Strong light reflection detection method and device of image |
| CN107607554A (en) * | 2017-09-26 | 2018-01-19 | 天津工业大学 | A kind of Defect Detection and sorting technique of the zinc-plated stamping parts based on full convolutional neural networks |
| CN107944504A (en) * | 2017-12-14 | 2018-04-20 | 北京木业邦科技有限公司 | Plank identifies and machine learning method, device and the electronic equipment of plank identification |
| CN107967491A (en) * | 2017-12-14 | 2018-04-27 | 北京木业邦科技有限公司 | Machine learning method, device, electronic equipment and the storage medium again of plank identification |
-
2018
- 2018-10-25 CN CN201811251848.6A patent/CN109409428A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103477861A (en) * | 2013-10-16 | 2014-01-01 | 福建省农业科学院植物保护研究所 | Method for processing export potted chrysalidocarpus lutescens |
| KR20170027580A (en) * | 2015-09-02 | 2017-03-10 | 김석배 | Glare preventing type road sign board and traffic lane discerning device for motors |
| CN106650743A (en) * | 2016-09-12 | 2017-05-10 | 北京旷视科技有限公司 | Strong light reflection detection method and device of image |
| CN107607554A (en) * | 2017-09-26 | 2018-01-19 | 天津工业大学 | A kind of Defect Detection and sorting technique of the zinc-plated stamping parts based on full convolutional neural networks |
| CN107944504A (en) * | 2017-12-14 | 2018-04-20 | 北京木业邦科技有限公司 | Plank identifies and machine learning method, device and the electronic equipment of plank identification |
| CN107967491A (en) * | 2017-12-14 | 2018-04-27 | 北京木业邦科技有限公司 | Machine learning method, device, electronic equipment and the storage medium again of plank identification |
Non-Patent Citations (1)
| Title |
|---|
| 杨薇丹: "复杂场景下的阴影检测", 《中国优秀硕士学位论文全文数据库(电子期刊) 信息科技辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110059549A (en) * | 2019-03-11 | 2019-07-26 | 齐鲁工业大学 | A kind of thin wood plate categorizing system and algorithm based on deep learning |
| CN110068578A (en) * | 2019-05-17 | 2019-07-30 | 苏州图迈蓝舸智能科技有限公司 | A kind of visual defects detection method, device and the terminal device of PVC floor |
| CN110068578B (en) * | 2019-05-17 | 2022-02-22 | 苏州图迈蓝舸智能科技有限公司 | Apparent defect detection method and device for PVC (polyvinyl chloride) floor and terminal equipment |
| CN112907498A (en) * | 2019-11-18 | 2021-06-04 | 中国商用飞机有限责任公司 | Pore identification method, device, equipment and storage medium |
| CN112907498B (en) * | 2019-11-18 | 2024-04-30 | 中国商用飞机有限责任公司 | Pore identification method, device, equipment and storage medium |
| CN111144244A (en) * | 2019-12-13 | 2020-05-12 | 唐美(北京)文化科技发展有限公司 | Furniture wood type identification method and device |
| CN111921885A (en) * | 2020-08-05 | 2020-11-13 | 东南数字经济发展研究院江山分院 | Intelligent detection method and device for wood door plate size |
| CN111951128A (en) * | 2020-08-31 | 2020-11-17 | 江苏工程职业技术学院 | Energy-saving and environment-friendly building construction method and device |
| CN111951128B (en) * | 2020-08-31 | 2022-01-28 | 江苏工程职业技术学院 | Energy-saving and environment-friendly building construction method and device |
| CN112560537A (en) * | 2020-12-29 | 2021-03-26 | 云南易见纹语科技有限公司 | Article texture processing device, method, system, electronic device and storage medium |
| CN113610184A (en) * | 2021-08-19 | 2021-11-05 | 江西应用技术职业学院 | Wood texture classification method based on transfer learning |
| CN114218423A (en) * | 2022-02-21 | 2022-03-22 | 广东联邦家私集团有限公司 | 5G-based non-labeling solid wood board identity digitalization method, device and system |
| CN114218423B (en) * | 2022-02-21 | 2022-05-20 | 广东联邦家私集团有限公司 | 5G-based non-labeling solid wood board identity digitalization method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109409428A (en) | Training method, device and the electronic equipment of plank identification and plank identification model | |
| CN103914802B (en) | For the image selection using the depth information imported and the System and method for of masking | |
| CN111723691B (en) | Three-dimensional face recognition method and device, electronic equipment and storage medium | |
| CN108701234A (en) | Licence plate recognition method and cloud system | |
| CN109344804A (en) | A kind of recognition methods of laser point cloud data, device, equipment and medium | |
| CN110570435B (en) | Method and device for carrying out damage segmentation on vehicle damage image | |
| CN106408037A (en) | Image recognition method and apparatus | |
| CN108765333B (en) | Depth map perfecting method based on depth convolution neural network | |
| CN111160328B (en) | Automatic extraction method of traffic marking based on semantic segmentation technology | |
| CN107358184A (en) | The extracting method and extraction element of document word | |
| Yu et al. | Artificial intelligence for Dunhuang cultural heritage protection: the project and the dataset | |
| CN108230376A (en) | Remote sensing image processing method, device and electronic equipment | |
| CN113808098A (en) | Road disease identification method and device, electronic equipment and readable storage medium | |
| CN111259891B (en) | Method, device, equipment and medium for identifying identity card in natural scene | |
| CN107463659A (en) | Object search method and its device | |
| CN109993806A (en) | A kind of color identification method, device and electronic equipment | |
| CN109522807A (en) | Satellite image identifying system, method and electronic equipment based on self-generating feature | |
| CN114972646B (en) | Method and system for extracting and modifying independent ground objects of live-action three-dimensional model | |
| CN111209783A (en) | Object recognition system using simulated object image and method thereof | |
| CN105701496A (en) | Go board surface identification method based on artificial intelligence technology | |
| CN110414522A (en) | A kind of character identifying method and device | |
| CN113628170B (en) | Laser line extraction method and system based on deep learning | |
| CN104915641B (en) | The method that facial image light source orientation is obtained based on Android platform | |
| CN108764248A (en) | The extracting method and device of image characteristic point | |
| CN112132845A (en) | Three-dimensional model unitization method and device, electronic equipment and readable medium |
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: 20190301 |
|
| RJ01 | Rejection of invention patent application after publication |