CN110334716A - Characteristic pattern processing method, image processing method and device - Google Patents
Characteristic pattern processing method, image processing method and device Download PDFInfo
- Publication number
- CN110334716A CN110334716A CN201910600515.8A CN201910600515A CN110334716A CN 110334716 A CN110334716 A CN 110334716A CN 201910600515 A CN201910600515 A CN 201910600515A CN 110334716 A CN110334716 A CN 110334716A
- Authority
- CN
- China
- Prior art keywords
- individual character
- obtains
- image
- weight
- processing method
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
Present disclose provides characteristic pattern processing method, image processing method and devices.Wherein feature extracting method includes: to obtain characteristic pattern step, obtains secondary features figure；Parameter generation step is based on secondary features figure, obtains personality by pond；Parameter tuning step obtains individual character guide parameters by carrying out a convolution to personality；Weight set-up procedure is adjusted weight according to individual character guide parameters, obtains individual character weight；Characteristic extraction step is based on secondary features figure and individual character weight, obtains advanced features figure.The feature extracting method that the disclosure provides passes through the different characteristic figure for input, generates different individual character weights, each feature carries out feature extraction with the individual character weight by generating according to reapective features, to improve the precision of result.
Description
Technical field
This invention relates generally to field of image recognition, and in particular to a kind of characteristic pattern processing method, image processing method
And device.
Background technique
With the development of computer technology, more and more scenes need to carry out target such as by computer technology and examine
The image processing works such as survey, target identification.Wherein convolutional neural networks (CNN) model is the core of modern deep visual identifying system
The heart.Convolutional neural networks are generally in the form of Y=conv (X, W), and wherein X is input feature vector, and Y is output feature, and W is weight.
By neural network backpropagation, gradient updating scheduling algorithm, update the value of weight, the training of neural network namely to weight more
Newly.
And by completing training to a neural network, all data by the weight of this trained network,
This set weight of i.e. all data sharings；Although, can be with a public weight however, all data have general character, data be also
Respective feature only will lead to the inaccuracy of output feature with a weight.
Summary of the invention
In order to solve the above-mentioned problems in the prior art, the first aspect of the present invention characteristic pattern processing method, wherein
Method includes: to obtain characteristic pattern step, obtains secondary features figure；Parameter generation step is based on secondary features figure, by pond
To personality；Parameter tuning step obtains individual character guide parameters by carrying out a convolution to personality；Weight adjustment step
Suddenly, according to individual character guide parameters, weight is adjusted, obtains individual character weight；Characteristic extraction step is based on secondary features figure
And individual character weight, obtain advanced features figure.
In one example, personality includes: convolution nuclear parameter and/or channel parameters.
In one example, convolution nuclear parameter is obtained by carrying out the pond to secondary features figure.
In one example, channel parameters are obtained by carrying out pond to convolution nuclear parameter.
In one example, channel parameters are obtained by carrying out pond to secondary features figure.
In one example, pond turns to average pond.
In one example, parameter tuning step includes: to obtain individual character guidance ginseng by carrying out multilayer point convolution to personality
Number.
In one example, parameter tuning step further include: after by carrying out a convolution to personality, through activation primitive into
Row nonlinear transformation obtains individual character guide parameters.
The second aspect of the present invention provides a kind of image processing method, comprising: image acquisition step obtains image；Feature
Extraction step extracts the characteristics of image of image according to the characteristic pattern processing method of such as first aspect；Image recognizing step, according to
Characteristics of image carries out image recognition.
The third aspect of the present invention provides a kind of characteristic pattern processing unit, and device includes: to obtain feature module, for obtaining
Take secondary features figure；Parameter generation module obtains personality by pond for being based on secondary features figure；Parameter adjusts mould
Block, for obtaining individual character guide parameters by carrying out a convolution to personality；Weight adjusts module, for being referred to according to individual character
Parameter is led, weight is adjusted, obtains individual character weight；Characteristic extracting module, for being based on secondary features figure and individual character
Weight obtains advanced features figure.
The fourth aspect of the present invention provides a kind of image processing apparatus, and image processing apparatus includes: image collection module, uses
In acquisition image；Characteristic extracting module, for according to the characteristic pattern processing method such as first aspect, the image for extracting image to be special
Sign；Picture recognition module, for carrying out image recognition according to characteristics of image.
The fifth aspect of the present invention provides a kind of electronic equipment, comprising: memory, for storing instruction；And processor,
The characteristic pattern processing method of instruction execution first aspect for calling memory to store or the image processing method of second aspect.
The sixth aspect of the present invention provides a kind of computer readable storage medium, wherein being stored with instruction, instructs processed
When device executes, the characteristic pattern processing method such as first aspect or the image processing method such as second aspect are executed.
Characteristic pattern processing method, image processing method and device provided by the invention pass through in convolutional neural networks convolutional layer
In, by the different characteristic figure for input, different individual character weights is generated, each feature according to reapective features the same as by generating
Individual character weight carry out feature extraction, to improve the precision of result.
Detailed description of the invention
The following detailed description is read with reference to the accompanying drawings, above-mentioned and other purposes, the feature of embodiment of the present invention
It will become prone to understand with advantage.In the accompanying drawings, several implementations of the invention are shown by way of example rather than limitation
Mode, in which:
Fig. 1 shows the flow diagram of an embodiment characteristic pattern processing method according to the present invention；
Fig. 2 shows the schematic diagrames of an embodiment characteristic pattern processing unit according to the present invention；
Fig. 3 is a kind of electronic equipment schematic diagram provided in an embodiment of the present invention.
In the accompanying drawings, identical or corresponding label indicates identical or corresponding part.
Specific embodiment
The principle and spirit of the invention are described below with reference to several illustrative embodiments.It should be appreciated that providing this
A little embodiments are used for the purpose of making those skilled in the art can better understand that realizing the present invention in turn, and be not with any
Mode limits the scope of the invention.
Although being noted that the statements such as " first " used herein, " second " to describe embodiments of the present invention not
Same module, step and data etc., still the statement such as " first ", " second " is merely in different modules, step and data etc.
Between distinguish, and be not offered as specific sequence or significance level.In fact, the statements such as " first ", " second " are complete
It may be used interchangeably.
With the development of computer technology and the extensive use of principle of computer vision, computer image processing technology is utilized
Target in image is detected, is tracked, identifies increasingly hot topic, dynamic realtime tracking and positioning is carried out in intelligence to target
Surgical instrument positioning etc. has extensive in traffic system, intelligent monitor system, military target detection and medical navigation operation
Application value；And very important effect is also played in field of security guarantee such as public security, anti-terrorisms to the identification of target.Convolution
Neural network is the core of above-mentioned technology, has been largely fixed image processing work to the operational efficiency of feature extraction
Efficiency.
In order to improve the accuracy of convolutional neural networks output result, Fig. 1 shows one kind provided in an embodiment of the present invention
Characteristic pattern processing method 100, comprising: obtain characteristic pattern step 110, parameter generation step 120, parameter tuning step 130, weight
Set-up procedure 140 and characteristic extraction step 150.Characteristic pattern processing method 100 can be applied to a certain of convolutional neural networks
Convolutional layer also can be applied in multilayer or whole convolutional layers.Above-mentioned steps are described in detail below.
Characteristic pattern step 110 is obtained, secondary features figure is obtained.
Wherein, the parameter that secondary features figure has may include characteristic pattern batch size, characteristic pattern number of channels, characteristic pattern
Height and characteristic pattern width.
Wherein, the secondary features figure of acquisition can be original graph, be also possible to the characteristic pattern by other convolution layer operations.
Characteristic pattern input feature vector is identified with X herein, can have four dimensional tensor (tensor), having a size of (n, c, h, w), wherein n
For batch size, c is number of channels, and h is characterized the height of figure, and w is characterized the width of figure.
In one example, when application neural network model carries out feature extraction, an only picture is admitted to model every time,
Therefore n=1；In another example, when being trained to neural network model, n be input characteristic pattern quantity, for more than or equal to
1 positive integer.
Parameter generation step 120 is based on secondary features figure, obtains personality by pond.
By carrying out pond to characteristic pattern, the personality with characteristic pattern is obtained.In one example, pond turns to average
Pond (Average Pooling).
In one example, personality may include: convolution nuclear parameter and/or channel parameters.
In one example, convolution nuclear parameter can directly carry out pond by secondary features figure and obtain, for example, by sliding window
Size (window size) is set as h*w, and edge filling size (padding) is set as (k-1)/2, and wherein k is that convolution kernel is big
It is small, 3 are generally taken, by pond, generation obtains convolution nuclear parameter, and tensor is (n, c, k, k).
In another example, channel parameters can be obtained by carrying out pond to convolution nuclear parameter.To obtained convolution kernel ginseng
Number (n, c, k, k) does further pond and obtains channel parameters, that is, on the basis of obtaining convolution nuclear parameter (n, c, k, k), further
By pond layer, k*k is set by sliding window size, edge filling is sized to 0, obtains channel parameters, tensor is
(n, c, 1,1).
In another example, channel parameters can also be obtained by carrying out pond to secondary features figure.For example, by sliding window
Size (window size) is set as h*w, and edge filling is sized to 0, directly obtains channel parameters, tensor be (n, c,
1,1).
Parameter tuning step 130 obtains individual character guide parameters by carrying out a convolution to personality.
In one example, the point of the 1 × 1 of multilayer can be carried out to personality, i.e. convolution nuclear parameter and/or channel parameters
Convolution obtains individual character guide parameters.Point convolution by multilayer, the individual character guide parameters enabled to have more special
Sign.
In one example, after passing point convolution, nonlinear transformation is carried out through activation primitive, obtains the individual character guidance ginseng
Number.Wherein activation primitive can be sigmoid nonlinear function.
By the above-mentioned means, can be adjusted to obtain convolution kernel guide parameters to convolution nuclear parameter and/or channel parameters
And/or channel guide parameters, the different dimensions of weight can be instructed respectively.Wherein, the convolution kernel guide parameters obtained
Tensor can be (n, c, k, k), the tensor of channel guide parameters can be (n, 0,1,1).
Weight set-up procedure 140 is adjusted weight according to individual character guide parameters, obtains individual character weight.
By obtained individual character guide parameters, to script, shared weight is adjusted, to obtain for different inputs
Characteristic pattern personalization weight.
In one example, individual character guide parameters include convolution kernel guide parameters and/or channel guide parameters, respectively to weight
Corresponding dimension is adjusted, specifically, convolution kernel guide parameters are adjusted the convolution kernel dimension of weight, channel guidance
Parameter is adjusted the channel dimension of weight.Wherein adjustment can carry out in such a way that transformation (reshape) is multiplied afterwards, example
Such as, the tensor of original weight W be (0, c, k, k), by being transformed to (1,0, c, k, k), by convolution kernel guide parameters tensor (n, c,
K, k) it is transformed to (n, 1, c, k, k), the tensor (n, 0,1,1) of channel guide parameters is transformed to (n, 0,1,1,1), by above-mentioned
Five-tensor is multiplied, the individual character weight after being adjusted, and tensor is (n, 0, c, k, k).
Characteristic extraction step 150 is based on secondary features figure and individual character weight, obtains advanced features figure.
Secondary features figure based on input, and the corresponding individual character weight generated according to the secondary features figure, are extracted
To advanced features figure.For example, secondary features figure input characteristic of field X tensor (n, c, h, w), by variation for (1, n*c, h,
W), then by individual character weight (n, 0, c, k, k) be grouped convolution (groupwise conv), obtain tensor be (1, n, 0, h,
W) as a result, by transformation obtain the output characteristic Y of tensor (n, 0, h, w).
Any of the above-described embodiment provided by the disclosure generates guide parameters according to the feature of each characteristic pattern respectively,
Weight is adjusted by guide parameters again, makes weight that each respective feature of characteristic pattern be added, carries out again on this basis
When feature extraction, the precision of feature extraction can be improved.
A kind of image processing method that the embodiment of the present invention also provides, comprising: image acquisition step obtains image；Feature
Extraction step extracts the characteristics of image of image according to the characteristic pattern processing method 100 of aforementioned any embodiment；Image recognition step
Suddenly, image recognition is carried out according to characteristics of image.It wherein, can be in a certain convolutional layer using aforementioned in characteristic extraction step
The characteristic pattern processing method 100 of any embodiment, can also be in multiple or whole convolutional layers using aforementioned any embodiment
Characteristic pattern processing method 100.Often in a convolutional layer in application, can be according to the secondary features figure progress that current layer inputs
Property adjust its parameter, therefore enable to current layer export advanced features figure it is more accurate.
Fig. 2 shows a kind of characteristic pattern processing units 200 provided in an embodiment of the present invention, as shown in Fig. 2, characteristic pattern is handled
Device 200 includes: to obtain feature module 210, for obtaining characteristic pattern；Parameter generation module 220, for being based on characteristic pattern,
Personality is obtained by pond；Parameter adjustment module 230, for obtaining individual character and referring to by carrying out a convolution to personality
Lead parameter；Weight adjusts module 240, for being adjusted to weight, obtaining individual character weight according to individual character guide parameters；It is special
Extraction module 250 is levied, for being based on characteristic pattern and individual character weight, obtains the feature of characteristic pattern.
In one example, personality includes: convolution nuclear parameter and/or channel parameters.
In one example, channel parameters are obtained by carrying out pond to characteristic pattern, or by carrying out pond to convolution nuclear parameter
It obtains.
In one example, pond turns to average pond.
In one example, parameter adjustment module 230 is used for: by carrying out multilayer point convolution to personality, being obtained individual character and is referred to
Lead parameter.
In one example, parameter adjustment module 230 is also used to: after by carrying out a convolution to personality, through activating letter
Number carries out nonlinear transformation, obtains individual character guide parameters.
A kind of image processing apparatus that the embodiment of the present invention also provides, comprising: image collection module, for obtaining image；
Characteristic extracting module, for extracting the characteristics of image of image according to the characteristic pattern processing method such as aforementioned any embodiment；Image
Identification module, for carrying out image recognition according to characteristics of image.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, no detailed explanation will be given here.
As shown in figure 3, an embodiment of the invention provides a kind of electronic equipment 300.Wherein, the electronic equipment
300 include memory 301, processor 302, input/output (Input/Output, I/O) interface 303.Wherein, memory 301,
For storing instruction.Processor 302, for call memory 301 store the instruction execution embodiment of the present invention characteristic pattern at
Reason method.Wherein, processor 302 is connect with memory 301, I/O interface 303 respectively, for example, can by bus system and/or its
He is attached bindiny mechanism's (not shown) of form.Memory 301 can be used for storing program and data, including the present invention is implemented
The program of characteristic pattern processing method involved in example, processor 302 by operation be stored in the program of memory 301 thereby executing
The various function application and data processing of electronic equipment 300.
Processor 302 can use digital signal processor (Digital Signal in the embodiment of the present invention
Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), programmable patrol
At least one of volume array (Programmable Logic Array, PLA) example, in hardware realizes, the processor 302
It can be central processing unit (Central Processing Unit, CPU) or there is data-handling capacity and/or instruction
The combination of one or more of the processing unit of other forms of executive capability.
Memory 301 in the embodiment of the present invention may include one or more computer program products, the computer
Program product may include various forms of computer readable storage mediums, such as volatile memory and/or non-volatile deposit
Reservoir.The volatile memory for example may include random access memory (Random Access Memory, RAM) and/
Or cache memory (cache) etc..The nonvolatile memory for example may include read-only memory (Read-Only
Memory, ROM), flash memory (Flash Memory), hard disk (Hard Disk Drive, HDD) or solid state hard disk
(Solid-State Drive, SSD) etc..
In the embodiment of the present invention, I/O interface 303 can be used for receiving input instruction (such as number or character information, and
Generate key signals input related with the user setting of electronic equipment 300 and function control etc.), it can also be output to the outside various
Information (for example, image or sound etc.).In the embodiment of the present invention I/O interface 303 may include physical keyboard, function button (such as
Volume control button, switch key etc.), mouse, operating stick, trace ball, microphone, one in loudspeaker and touch panel etc.
It is a or multiple.
It is understood that although description operation in a particular order in the accompanying drawings in the embodiment of the present invention, is not answered
It is understood as requiring particular order or serial order shown in execute these operations, or requires to execute whole institutes
The operation shown is to obtain desired result.In specific environment, multitask and parallel processing may be advantageous.
The present embodiments relate to method and apparatus can be completed using standard programming technology, utilization is rule-based
Logic or other logics realize various method and steps.It should also be noted that herein and used in claims
Word " device " and " module " are intended to include using the realization of a line or multirow software code and/or hardware realization and/or use
In the equipment for receiving input.
One or more combined individually or with other equipment can be used in any step, operation or program described herein
A hardware or software module are executed or are realized.In one embodiment, software module use includes comprising computer program
The computer program product of the computer-readable medium of code is realized, can be executed by computer processor any for executing
Or whole described step, operation or programs.
For the purpose of example and description, the preceding description that the present invention is implemented is had been presented for.Preceding description is not poor
Also not the really wanting of act property limits the invention to exact form disclosed, according to the above instruction there is likely to be various modifications and
Modification, or various changes and modifications may be obtained from the practice of the present invention.Select and describe these embodiments and be in order to
Illustrate the principle of the present invention and its practical application, so that those skilled in the art can be to be suitable for the special-purpose conceived
Come in a variety of embodiments with various modifications and utilize the present invention.
Claims (13)
1. a kind of characteristic pattern processing method, wherein the described method includes:
Characteristic pattern step is obtained, secondary features figure is obtained；
Parameter generation step is based on the secondary features figure, obtains personality by pond；
Parameter tuning step obtains individual character guide parameters by carrying out a convolution to the personality；
Weight set-up procedure is adjusted weight according to the individual character guide parameters, obtains individual character weight；
Characteristic extraction step is based on the secondary features figure and the individual character weight, obtains advanced features figure.
2. according to the method described in claim 1, wherein, the personality includes: convolution nuclear parameter and/or channel parameters.
3. according to the method described in claim 2, wherein, the convolution nuclear parameter passes through to described in secondary features figure progress
Pond obtains.
4. according to the method described in claim 3, wherein, the channel parameters are by carrying out the pond to the convolution nuclear parameter
Change obtains.
5. according to the method described in claim 2, wherein, the channel parameters are by carrying out the pond to the secondary features figure
Change obtains.
6. method according to claim 1-5, wherein the pond turns to average pond.
7. according to the method described in claim 1, wherein, the parameter tuning step include: by the personality into
Convolution is put described in row multilayer, obtains the individual character guide parameters.
8. according to the method described in claim 1, wherein, the parameter tuning step further include: by joining to the individual character
After number carries out described convolution, nonlinear transformation is carried out through activation primitive, obtains the individual character guide parameters.
9. a kind of image processing method, comprising:
Image acquisition step obtains image；
Characteristic extraction step, characteristic pattern processing method according to claim 1-8, extracts the image of described image
Feature；
Image recognizing step carries out image recognition according to described image feature.
10. a kind of characteristic pattern processing unit, wherein described device includes:
Feature module is obtained, for obtaining secondary features figure；
Parameter generation module obtains personality by pond for being based on the secondary features figure；
Parameter adjustment module, for obtaining individual character guide parameters by carrying out a convolution to the personality；
Weight adjusts module, for being adjusted to weight, obtaining individual character weight according to the individual character guide parameters；
Characteristic extracting module, for being based on the secondary features figure and the individual character weight advanced features figure.
11. a kind of image processing apparatus, wherein described image processing unit includes:
Image collection module, for obtaining image；
Characteristic extracting module is used for characteristic pattern processing method according to claim 1-8, extracts described image
Characteristics of image；
Picture recognition module, for carrying out image recognition according to described image feature.
12. a kind of electronic equipment, wherein the electronic equipment includes:
Memory, for storing instruction；And
Processor, for calling the instruction execution characteristic pattern for example described in any item of the claim 1 to 8 of the memory storage
Processing method or image processing method as claimed in claim 9.
13. a kind of computer readable storage medium when described instruction is executed by processor, is executed as weighed wherein being stored with instruction
Benefit require any one of 1 to 8 described in characteristic pattern processing method or image processing method as claimed in claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910600515.8A CN110334716B (en) | 2019-07-04 | 2019-07-04 | Feature map processing method, image processing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910600515.8A CN110334716B (en) | 2019-07-04 | 2019-07-04 | Feature map processing method, image processing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110334716A true CN110334716A (en) | 2019-10-15 |
| CN110334716B CN110334716B (en) | 2022-01-11 |
Family
ID=68143150
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910600515.8A Active CN110334716B (en) | 2019-07-04 | 2019-07-04 | Feature map processing method, image processing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110334716B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110852348A (en) * | 2019-10-18 | 2020-02-28 | 北京迈格威科技有限公司 | Feature map processing method, image processing method and device |
| CN111210017A (en) * | 2019-12-24 | 2020-05-29 | 北京迈格威科技有限公司 | Method, device, equipment and storage medium for determining layout sequence and processing data |
| CN113239899A (en) * | 2021-06-17 | 2021-08-10 | 阿波罗智联(北京)科技有限公司 | Method for processing image and generating convolution kernel, road side equipment and cloud control platform |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110249886A1 (en) * | 2010-04-12 | 2011-10-13 | Samsung Electronics Co., Ltd. | Image converting device and three-dimensional image display device including the same |
| CN106529670A (en) * | 2016-10-27 | 2017-03-22 | 中国科学院计算技术研究所 | Neural network processor based on weight compression, design method, and chip |
| CN107766894A (en) * | 2017-11-03 | 2018-03-06 | 吉林大学 | Remote sensing images spatial term method based on notice mechanism and deep learning |
| CN108182452A (en) * | 2017-12-29 | 2018-06-19 | 哈尔滨工业大学（威海） | Aero-engine fault detection method and system based on grouping convolution self-encoding encoder |
| CN108364023A (en) * | 2018-02-11 | 2018-08-03 | 北京达佳互联信息技术有限公司 | Image-recognizing method based on attention model and system |
| CN108388900A (en) * | 2018-02-05 | 2018-08-10 | 华南理工大学 | The video presentation method being combined based on multiple features fusion and space-time attention mechanism |
| CN108510012A (en) * | 2018-05-04 | 2018-09-07 | 四川大学 | A kind of target rapid detection method based on Analysis On Multi-scale Features figure |
| US10242289B2 (en) * | 2015-12-21 | 2019-03-26 | Nokia Technologies Oy | Method for analysing media content |
| CN109753903A (en) * | 2019-02-27 | 2019-05-14 | 北航（四川）西部国际创新港科技有限公司 | A kind of unmanned plane detection method based on deep learning |
| CN109858419A (en) * | 2019-01-23 | 2019-06-07 | 广州智慧城市发展研究院 | It is a kind of from bottom to top-top-down Activity recognition system |
-
2019
- 2019-07-04 CN CN201910600515.8A patent/CN110334716B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110249886A1 (en) * | 2010-04-12 | 2011-10-13 | Samsung Electronics Co., Ltd. | Image converting device and three-dimensional image display device including the same |
| US10242289B2 (en) * | 2015-12-21 | 2019-03-26 | Nokia Technologies Oy | Method for analysing media content |
| CN106529670A (en) * | 2016-10-27 | 2017-03-22 | 中国科学院计算技术研究所 | Neural network processor based on weight compression, design method, and chip |
| CN107766894A (en) * | 2017-11-03 | 2018-03-06 | 吉林大学 | Remote sensing images spatial term method based on notice mechanism and deep learning |
| CN108182452A (en) * | 2017-12-29 | 2018-06-19 | 哈尔滨工业大学（威海） | Aero-engine fault detection method and system based on grouping convolution self-encoding encoder |
| CN108388900A (en) * | 2018-02-05 | 2018-08-10 | 华南理工大学 | The video presentation method being combined based on multiple features fusion and space-time attention mechanism |
| CN108364023A (en) * | 2018-02-11 | 2018-08-03 | 北京达佳互联信息技术有限公司 | Image-recognizing method based on attention model and system |
| CN108510012A (en) * | 2018-05-04 | 2018-09-07 | 四川大学 | A kind of target rapid detection method based on Analysis On Multi-scale Features figure |
| CN109858419A (en) * | 2019-01-23 | 2019-06-07 | 广州智慧城市发展研究院 | It is a kind of from bottom to top-top-down Activity recognition system |
| CN109753903A (en) * | 2019-02-27 | 2019-05-14 | 北航（四川）西部国际创新港科技有限公司 | A kind of unmanned plane detection method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| SANGHYUN WOO等: "CBAM: Convolutional Block Attention Module", 《ARXIV:1807.06521V2》 * |
| XIANG LI: "Spatial Group-wise Enhance: Improving Semantic Feature Learning in Convolutional Networks", 《ARXIV:1905.09646V2》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110852348A (en) * | 2019-10-18 | 2020-02-28 | 北京迈格威科技有限公司 | Feature map processing method, image processing method and device |
| CN110852348B (en) * | 2019-10-18 | 2022-09-30 | 北京迈格威科技有限公司 | Feature map processing method, image processing method and device |
| CN111210017A (en) * | 2019-12-24 | 2020-05-29 | 北京迈格威科技有限公司 | Method, device, equipment and storage medium for determining layout sequence and processing data |
| CN111210017B (en) * | 2019-12-24 | 2023-09-26 | 北京迈格威科技有限公司 | Method, device, equipment and storage medium for determining layout sequence and data processing |
| CN113239899A (en) * | 2021-06-17 | 2021-08-10 | 阿波罗智联(北京)科技有限公司 | Method for processing image and generating convolution kernel, road side equipment and cloud control platform |
| CN113239899B (en) * | 2021-06-17 | 2024-05-28 | 阿波罗智联(北京)科技有限公司 | Method for processing image and generating convolution kernel, road side equipment and cloud control platform |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110334716B (en) | 2022-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10963783B2 (en) | Technologies for optimized machine learning training | |
| CN110334716A (en) | Characteristic pattern processing method, image processing method and device | |
| WO2019106136A1 (en) | Training a neural network model | |
| CN105701120A (en) | Method and apparatus for determining semantic matching degree | |
| CN110210513A (en) | Data classification method, device and terminal device | |
| KR20220088943A (en) | Memristor-based neural network parallel acceleration method, processor and device | |
| CN112990444B (en) | Hybrid neural network training method, system, equipment and storage medium | |
| US11514315B2 (en) | Deep neural network training method and apparatus, and computer device | |
| CN113298152B (en) | Model training method, device, terminal equipment and computer readable storage medium | |
| CN110443241A (en) | Car license recognition model training method, licence plate recognition method and device | |
| CN110059804A (en) | Network training method, data processing method and device to be searched | |
| TW202207091A (en) | Network model quantization method, device, and electronic apparatus | |
| CN112508190A (en) | Method, device and equipment for processing structured sparse parameters and storage medium | |
| Jafar et al. | Hyperparameter optimization for deep residual learning in image classification | |
| Wang et al. | Light attention embedding for facial expression recognition | |
| CN111738269A (en) | Model training method, image processing device, model training apparatus, and storage medium | |
| CN110222730A (en) | Method for identifying ID and identification model construction method based on inertial sensor | |
| Zeng et al. | A novel tensor decomposition-based efficient detector for low-altitude aerial objects with knowledge distillation scheme | |
| CN112420125A (en) | Molecular attribute prediction method and device, intelligent equipment and terminal | |
| CN111126358A (en) | Face detection method, face detection device, storage medium and equipment | |
| CN108875901B (en) | Neural network training method and universal object detection method, device and system | |
| Chaitra et al. | Deep learning model for image-based plant diseases detection on edge devices | |
| CN110334576A (en) | A kind of hand method for tracing and device | |
| CN110852348B (en) | Feature map processing method, image processing method and device | |
| CN117999560A (en) | Hardware-aware progressive training of machine learning models |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |