CN109117703A - A fine-grained identification-based method for identification of promiscuous cell types - Google Patents
A fine-grained identification-based method for identification of promiscuous cell types Download PDFInfo
- Publication number
- CN109117703A CN109117703A CN201810608329.4A CN201810608329A CN109117703A CN 109117703 A CN109117703 A CN 109117703A CN 201810608329 A CN201810608329 A CN 201810608329A CN 109117703 A CN109117703 A CN 109117703A
- Authority
- CN
- China
- Prior art keywords
- cell
- image
- identification
- fine granularity
- convolutional neural
- 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 OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
- G06V20/695—Preprocessing, e.g. image segmentation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
- G06V20/698—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention particularly relates to a method for identifying the types of mixed cells based on fine-grained identification, which comprises the following steps: pre-establishing a fine-grained recognition convolutional neural network model and a cell image database, wherein the cell image database comprises a mixed cell image which is an image comprising multiple types of cells; s1, collecting mixed cell images; s2, inputting the mixed cell image into a fine-grained identification convolutional neural network model to obtain a cell type heat map; s3, thresholding the mixed cell image to obtain a cell region binary image; and S4, combining the cell region binary image and the cell type heat map to obtain a cell type identification result. The invention accurately identifies the cell types according to the specificity of the cell morphological characteristics, and avoids the defects of long time consumption and complicated process of the traditional cell type identification method. The model can learn the morphological characteristics of fine-grained cells, identify the cell types through the information such as textures and the like, and has high identification accuracy and robustness.
Description
Technical field
The present invention relates to Biomedical Image processing and machine learning field, in particular to specific admixture cell category identifications
Method.
Background technique
In biomedicine experiment, cell line is occurred often by the case where misidentification or cross contamination, is distinguished using mistake
Know or the cell line of cross contamination will lead to experimental result cannot reappear, research conclusion mistake, clinical cytology treatment disaster etc.
Serious consequence, while also wasting a large amount of manpowers, energy, money etc..Conventional cell system identification method uses cell sample DNA information
Mode compared with cell bank locus pair, determines cell line classification and whether by cross contamination, higher cost, it is time-consuming compared with
It is long.
Recently, depth convolutional neural networks have been achieved with immense success on many visual tasks.Compared to traditional machine
Learning method, convolutional neural networks be not necessarily to expertise, can automatically extract suitable characteristics of image be applied to classification, detection,
The tasks such as semantic segmentation, thus show good performance.More and more researchers are by depth convolutional neural networks application
To field of medical image processing, and obtain good result.In terms of cell image recognition, most of prior art is all first from image
In be partitioned into individual cells, classified later according to cell morphology characteristic to it.These methods for only including in image
The performance of the case where single, individual cells is good, but the case where cell growth is more intensive, detection zone includes various kinds of cell
Under, the cell segmentation in image becomes difficult, cell morphology characteristic is easy the interference by other cell types, leads to cell
Recognition accuracy reduces.
Fine granularity identification refers to the identification between of a sort different subclasses or example, such as Poodle, shepherd dog, Bulldog
Etc. belonging to canine, the form difference between them is smaller, needs to be distinguished by coat color, Texture eigenvalue.Particulate
Degree identification can substantially be divided into two class method of partial model and world model.Partial model is first to the higher portion of object discrimination
Position is positioned, and the feature for extracting these positions later judges the object category, such method can reduce position, posture and view
Angle changes the influence to classification results.The feature that world model then passes through extraction entire image classifies to image, vision word
The classics image representation such as related mutation of allusion quotation and its texture analysis belongs to such method.Researcher both domestic and external is
Fine grit classification feature is extracted using convolutional neural networks, and is shown in the tasks such as texture recognition, scene Recognition, fine grit classification
Excellent performance is shown.
Summary of the invention
In order to overcome the drawbacks of the prior art, the present invention provides a kind of easy to operate, result and accurately mixes cell category
Identification method carries out accurately identifying for cell category according to the specificity of cell morphological characteristic, avoids conventional cell type mirror
The disadvantage that the method for determining takes a long time, process is cumbersome.The present invention mix using fine granularity identification convolutional neural networks model thin
The identification of born of the same parents' image pixel-class cell category, model can learn to fine granularity cell morphological characteristic, be identified by information such as textures
Cell category.Compared to general depth convolutional neural networks, this method cell is smaller, growth is intensive, in image comprising a variety of thin
Recognition accuracy with higher and robustness in the case where born of the same parents.
It is of the invention the specific scheme is that
It is a kind of that cell category identification method is mixed based on fine granularity identification, include the following steps:
Fine granularity identification convolutional neural networks model and cell image database are pre-established, is wrapped in cell image database
Miscegenation cell image is included, miscegenation cell image is the image for including multiple types cell；
S1, miscegenation cell image is collected；
S2, miscegenation cell image input fine granularity is identified in convolutional neural networks model, obtains cell category thermal map；
S3, thresholding is carried out to miscegenation cell image, obtains cell compartment bianry image；
S4, combination cell region bianry image and cell category thermal map, obtain cell category qualification result.
The present invention collects miscegenation cell image under the microscope, carries out cell by the specificity of cell morphological characteristic
Type accurately identifies, avoid conventional cell Identification of Species method take a long time, the disadvantage that process is cumbersome.What is pre-established is thin
Granularity identifies convolutional neural networks model, wherein study has fine granularity cell morphological characteristic, identifies cell by information such as textures
Type, compares general depth convolutional neural networks model, this method cell is smaller, growth is intensive, in image comprising a variety of thin
Recognition accuracy with higher and robustness in the case where born of the same parents.Cell compartment bianry image after thresholding can be by background area
It is separated with cell compartment, in order to remove the misrecognition in background area in subsequent processes, improves recognition accuracy.
Further, cell image database further includes the single cell image for having been marked with cell category label, single
Cell image is the image for including single type cell；The step of pre-establishing fine granularity identification convolutional neural networks model packet
It includes:
It constructs fine granularity and identifies convolutional neural networks model；
Collect single cell image；
Convolutional neural networks model, which is trained, to be identified to fine granularity by single cell image.
Fine granularity identification convolutional neural networks used in the present invention only need to provide single cytological map in the training process
The training data of picture and its corresponding cell category label avoid thin to Pixel-level used in cell image progress semantic segmentation
Born of the same parents' type label, can save a large amount of human and material resources.
Further, before training, data amplification is carried out to single cell image.The method expanded using data, energy one
Determine to improve data capacity in degree, to improve the training effect of model.
Further, data amplification procedure includes: translation, scaling, rotation and color channel offset.
Further, the image in cell image database is taking out before use, being pre-processed.It is pretreated
Image is more advantageous to fine granularity identification convolutional neural networks extraction cell morphological characteristic and is identified, improved thin compared to original image
Granularity identifies the training effectiveness of the accuracy rate of convolutional neural networks model identification, fine granularity identification convolutional neural networks.
Further, preprocessing process includes: that background illumination normalization, brightness normalization and contrast are promoted.
Further, it by before image input fine granularity identification convolutional neural networks model, needs image cropping to be multiple
Image block；The multiple images block formed by miscegenation cell image obtains after input fine granularity identification convolutional neural networks model
Multiple cell category labels form cell category thermal map by this multiple cell category label in conjunction with miscegenation cell image.Specifically
By way of sliding window by image cropping be multiple images block；Multiple cell categories are label converting to be indicated with pixel value,
It is mapped to the center of corresponding image block, then multiple images block carries out rearranging group further according to former miscegenation cell image
It closes, to form final cell category thermal map.
Further, after thresholding, it is also necessary to using morphological operation removal noise and hole, can just obtain cell compartment
Bianry image.
Further, the step S4 specifically: for the connected region in cell compartment bianry image, count connected region
Cell category label corresponding to pixel in domain will be corresponding with the most cell category label of pixel quantity as connected region
The cell category label of all pixels point, obtains cell category qualification result in domain.Cell category corresponding to above-mentioned pixel
Label is substantially mapped to cell category is label converting on pixel for pixel value.
Further, fine granularity identification convolutional neural networks model includes five convolution blocks, a bilinearity outer lamination, one
A full articulamentum.
Compared with the prior art, the invention has the benefit that
(1) present invention can be inputted using the cell image under microscope as system, and data acquisition is convenient.User of service
Cell image under clearly microscope only need to be acquired, uploads to system, cellular identification work can be completed.It avoids traditional
Cellular identification method needs for cell sample to be sent to evaluating center, extracts sample gene information, and then carries out the numerous of cellular identification
Trivial process.
(2) method proposed by the invention has stronger robustness.Preprocessing process of the present invention can effectively eliminate unevenness
Even background illumination, normalized image brightness, enhancing picture contrast.It is inclined using translation, scaling, rotation, color channel simultaneously
The data amplification methods such as shifting increase training samples number, avoid model over-fitting, improve the robustness of model.
(3) method proposed by the invention is suitable for the identification of the cell image of several scenes.Most prior art is all
The case where just for detection zone including single, individual cells, is detected.These methods grow more intensive, detection in cell
Region in the case where various kinds of cell comprising performing poor.Method proposed by the invention includes a variety of, Duo Gexi in detection zone
When born of the same parents, it can effectively avoid other cells to the interference of model, generate accurate prediction result.
(4) method proposed by the invention can generate accurate Pixel-level cell category prediction result.The present invention is based on thin
Bilinearity pond layer is added in a model, models to image to the interaction between grade feature, energy for granularity convolutional neural networks
Enough extract cell fine granularity morphological feature.It changes in cell position, form etc. and detection zone includes other cell types
When, it remains to generate accurate classification results.
(5) fine granularity convolutional neural networks model of the invention is able to carry out end-to-end training, training process simplicity.It compares
The prior art effectively simplifies model learning, training process using multistage, multistage method.Fine granularity of the present invention simultaneously
Convolutional neural networks training process only needs single cell image and its type label, and data set is collected and annotation process is easy.
Detailed description of the invention
Fig. 1 is broad flow diagram of the invention.
Specific embodiment
To enable the goal of the invention, feature, advantage of the invention patent more obvious and understandable, below in conjunction with this
Attached drawing in patent of invention is clearly and completely described the technical solution in the invention patent, it is clear that disclosed below
Embodiment be only a part of the embodiment of the present invention, and not all embodiment.Based on the invention patent, the common skill in this field
Art personnel all other embodiment obtained without making creative work belongs to the invention patent protection
Range.
The invention will be further described below in conjunction with the accompanying drawings:
It is as shown in Figure 1 it is a kind of cell category identification method is mixed based on fine granularity identification, include the following steps:
Fine granularity identification convolutional neural networks model and cell image database are pre-established, is wrapped in cell image database
Include miscegenation cell image and have been marked with the single cell image of cell category label, miscegenation cell image be include multiple types
The image of type cell, single cell image are the image for including single type cell；
S1, miscegenation cell image is collected, and miscegenation cell image is pre-processed；
S2, miscegenation cell image input fine granularity is identified in convolutional neural networks model, obtains cell category thermal map；
S3, thresholding is carried out to pretreated miscegenation cell image, removes noise and hole using morphological operation, obtains
To cell compartment bianry image, noise or hole refer to object or hole of the size less than 64 pixels；
S4, combination cell region bianry image and cell category thermal map, obtain cell category qualification result.
Establishing the step of fine granularity identifies convolutional neural networks model includes:
Construction identifies convolutional Neural including the fine granularity of five convolution blocks, the outer lamination of a bilinearity, a full articulamentum
Network model；
Single cell image is collected, and pretreatment and data amplification are carried out to single cell image；
Will by pretreatment and data amplification single cell image input fine granularity identify convolutional neural networks model into
Row training.
Above-mentioned preprocessing process includes: that background illumination normalization, brightness normalization and contrast are promoted, specifically: according to
The size of cell in image selects size to be greater than the Gaussian convolution core size (W of cell size (such as 64x64)kernel,Hkernel),
Cell image and Gaussian kernel are subjected to convolution later, obtain the background illumination luminance picture of cell image:
Wherein, G (x, y) is dimensional Gaussian convolution kernel, and σ is the standard deviation of Gaussian Profile, IsrcFor initial cell image, Ibg
For background illumination intensity image,For convolution operation.
Later, initial cell image is subtracted into background illumination intensity, adds background illumination mean value later, obtains background illumination
Cell image after homogenization:
Wherein, Ibg_norm(x, y) is the cell image after background illumination homogenization,For background illumination mean value.
Finally, carrying out gray scale normalization and contrast promotion.Input picture surrounding is expanded using most recent value first,
The mean value and standard deviation of calculating input image gray value later calculates the gray value after pixel gray level normalization:
Wherein, Iin(x,y),Iout(x, y) is respectively the gray value for inputting, exporting image slices vegetarian refreshments,For input figure
Picture gray average and standard deviation,For the output gray value of image mean value and standard deviation of setting.
Above-mentioned data amplification procedure includes that translation, scaling, rotation and color channel deviate, specifically: in order to original
Beginning image zooms in and out the scaling that coefficient is respectively { 0.9,1.0,1.1 }, and is the thin of original image by image tagged after scaling
Born of the same parents' type label.
To image obtained in the previous step, sequence carries out the rotation that rotation angle is { -90,0,90 } respectively, and will contracting
Put the label that rear image tagged is original image.
To image obtained in the previous step, carrying out coefficient respectively in order to its gray value of image is { -10,0,10 }
Color channel offset, i.e., add deviation ratio for the brightness value in each channel of original image respectively.And by the image mark after offset
It is denoted as original image label.
It is expanded and is operated by above-mentioned data, data set quantity can be promoted 3x3x3=27 times.
Before image input fine granularity identification convolutional neural networks model, need image cropping to be multiple images block；By
The multiple images block that miscegenation cell image is formed obtains multiple cell kinds after input fine granularity identification convolutional neural networks model
Class label forms cell category thermal map by this multiple cell category label in conjunction with miscegenation cell image.Especially by sliding window
Image cropping is multiple images block by the mode of mouth；Multiple cell categories are label converting to be indicated with pixel value, are mapped to corresponding
Image block center, then multiple images block carries out rearranging combination further according to former miscegenation cell image, with formed
Final cell category thermal map.
Before miscegenation cell image is cut to multiple images block by way of sliding window, in miscegenation cell image
Distinguishing packed height up and down is Hwin/ 2 and gray value be 0 block of pixels, its left and right fill respectively width be Wwin/ 2 and
The block of pixels that gray value is 0 is (W using size when followed by sliding windowwin,Hwin) sliding window.Sliding window
Mouth is larger, then recognition effect is poor, and recognition time is short, if sliding window is smaller, recognition effect is good, but recognition time is long.?
After multiple images block input fine granularity identification convolutional neural networks model, multiple cell category labels have been obtained, these are thin
Born of the same parents' type label is as (W at each image block center of miscegenation cell imagecnt_win,Hcnt_win) in the range of pixel it is thin
Born of the same parents' type label, then each image block is combined into former miscegenation cell image to get cell category thermal map.Wherein have 1≤
Wcnt_win≤Wwin,1≤Hcnt_win≤Hwin。(Wwin,Hwin) it is the wide and high of sliding window, (Wcnt_win,Hcnt_win) can basis
Actual demand carries out different selections, is defaulted as (1,1).As (Wcnt_win,Hcnt_win) value it is larger when, the used time is shorter, but identify effect
Fruit is poor；As (Wcnt_win,Hcnt_win) value it is smaller when, the recognition effect of picture is preferable, but the used time is longer.
The step S4 specifically: for the connected region in cell compartment bianry image, count pixel in connected region
The corresponding cell category label of point will be corresponding with the most cell category label of pixel quantity and own as in connected region
The cell category label of pixel, obtains cell category qualification result.Cell category label corresponding to above-mentioned pixel, essence
On be to be mapped to cell category is label converting on pixel for pixel value.
The above, the above embodiments are merely illustrative of the technical solutions of the present invention, although rather than its limitations referring to before
Stating embodiment, invention is explained in detail, those skilled in the art should understand that it still can be to aforementioned
Technical solution documented by each embodiment is modified or equivalent replacement of some of the technical features, and these are repaired
Change or replaces, the spirit and scope for the invention patent technical solution that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of mix cell category identification method based on fine granularity identification, which comprises the steps of:
Fine granularity identification convolutional neural networks model and cell image database are pre-established, includes in cell image database
Miscegenation cell image, miscegenation cell image are the image for including multiple types cell；
S1, miscegenation cell image is collected；
S2, miscegenation cell image input fine granularity is identified in convolutional neural networks model, obtains cell category thermal map；
S3, thresholding is carried out to miscegenation cell image, obtains cell compartment bianry image；
S4, combination cell region bianry image and cell category thermal map, obtain cell category qualification result.
2. it is according to claim 1 it is a kind of based on fine granularity identification mix cell category identification method, which is characterized in that
Cell image database further includes the single cell image for having been marked with cell category label, single cell image be include single
The image of cell type；Pre-establishing the step of fine granularity identifies convolutional neural networks model includes:
It constructs fine granularity and identifies convolutional neural networks model；
Collect single cell image；
Convolutional neural networks model, which is trained, to be identified to fine granularity by single cell image.
3. it is according to claim 2 it is a kind of based on fine granularity identification mix cell category identification method, which is characterized in that
Before training, data amplification is carried out to single cell image.
4. it is according to claim 3 it is a kind of based on fine granularity identification mix cell category identification method, which is characterized in that
Data amplification procedure includes: translation, scaling, rotation and color channel offset.
5. it is according to any one of claims 1 to 4 it is a kind of based on fine granularity identification mix cell category identification method,
It is characterized in that, the image in cell image database is taking out before use, being pre-processed.
6. it is according to claim 5 it is a kind of based on fine granularity identification mix cell category identification method, which is characterized in that
Preprocessing process includes: that background illumination normalization, brightness normalization and contrast are promoted.
7. it is according to any one of claims 1 to 4 it is a kind of based on fine granularity identification mix cell category identification method,
It is characterized in that, before image input fine granularity identification convolutional neural networks model, needs image cropping to be multiple images block；By
The multiple images block that miscegenation cell image is formed obtains multiple cell kinds after input fine granularity identification convolutional neural networks model
Class label forms cell category thermal map by this multiple cell category label in conjunction with miscegenation cell image.
8. it is according to any one of claims 1 to 4 it is a kind of based on fine granularity identification mix cell category identification method,
It is characterized in that, after thresholding, it is also necessary to using morphological operation removal noise and hole, can just obtain cell compartment binary map
Picture.
9. it is according to any one of claims 1 to 4 it is a kind of based on fine granularity identification mix cell category identification method,
It is characterized in that, the step S4 specifically: for the connected region in cell compartment bianry image, count pixel in connected region
The corresponding cell category label of point will be corresponding with the most cell category label of pixel quantity and own as in connected region
The cell category label of pixel, obtains cell category qualification result.
10. it is according to any one of claims 1 to 4 it is a kind of based on fine granularity identification mix cell category identification method,
It is characterized in that, fine granularity identification convolutional neural networks model connects entirely including five convolution blocks, the outer lamination of a bilinearity, one
Connect layer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810608329.4A CN109117703B (en) | 2018-06-13 | 2018-06-13 | A fine-grained identification-based method for identification of promiscuous cell types |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810608329.4A CN109117703B (en) | 2018-06-13 | 2018-06-13 | A fine-grained identification-based method for identification of promiscuous cell types |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109117703A true CN109117703A (en) | 2019-01-01 |
| CN109117703B CN109117703B (en) | 2022-03-22 |
Family
ID=64822925
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810608329.4A Active CN109117703B (en) | 2018-06-13 | 2018-06-13 | A fine-grained identification-based method for identification of promiscuous cell types |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109117703B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109886321A (en) * | 2019-01-31 | 2019-06-14 | 南京大学 | A kind of image characteristic extracting method and device for icing image fine grit classification |
| CN110675368A (en) * | 2019-08-31 | 2020-01-10 | 中山大学 | Cell image semantic segmentation method integrating image segmentation and classification |
| CN111291692A (en) * | 2020-02-17 | 2020-06-16 | 咪咕文化科技有限公司 | Video scene recognition method, device, electronic device and storage medium |
| CN112560999A (en) * | 2021-02-18 | 2021-03-26 | 成都睿沿科技有限公司 | Target detection model training method and device, electronic equipment and storage medium |
| CN112816480A (en) * | 2021-02-01 | 2021-05-18 | 奎泰斯特(上海)科技有限公司 | Water quality enzyme substrate identification method |
| CN113516022A (en) * | 2021-04-23 | 2021-10-19 | 黑龙江机智通智能科技有限公司 | Fine-grained classification system for cervical cells |
| CN115700821A (en) * | 2022-11-24 | 2023-02-07 | 广东美赛尔细胞生物科技有限公司 | Cell identification method and system based on image processing |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110289109A1 (en) * | 2010-05-21 | 2011-11-24 | Sony Corporation | Information processing apparatus, information processing method, and program |
| CN102737232A (en) * | 2012-06-01 | 2012-10-17 | 天津大学 | Cleavage cell recognition method |
| CN106127255A (en) * | 2016-06-29 | 2016-11-16 | 深圳先进技术研究院 | The sorting technique of a kind of cancer numeral pathological cells image and system |
| CN106650796A (en) * | 2016-12-06 | 2017-05-10 | 国家纳米科学中心 | Artificial intelligence based cell fluorescence image classification method and system |
| CN106795558A (en) * | 2014-05-30 | 2017-05-31 | 维里纳塔健康公司 | Detection fetus Asia chromosomal aneuploidy and copy number variation |
| US20170154202A1 (en) * | 2013-09-16 | 2017-06-01 | EyeVerify Inc. | Feature extraction and matching for biometric authentication |
| CN106991673A (en) * | 2017-05-18 | 2017-07-28 | 深思考人工智能机器人科技（北京）有限公司 | A kind of cervical cell image rapid classification recognition methods of interpretation and system |
| CN106991417A (en) * | 2017-04-25 | 2017-07-28 | 华南理工大学 | A kind of visual projection's interactive system and exchange method based on pattern-recognition |
| CN107133569A (en) * | 2017-04-06 | 2017-09-05 | 同济大学 | The many granularity mask methods of monitor video based on extensive Multi-label learning |
| US20170337676A1 (en) * | 2016-05-23 | 2017-11-23 | General Electric Company | Textural analysis of diffused disease in the lung |
| CN107563444A (en) * | 2017-09-05 | 2018-01-09 | 浙江大学 | A kind of zero sample image sorting technique and system |
| WO2018065027A1 (en) * | 2016-10-03 | 2018-04-12 | Total E&P Uk Limited | Modelling geological faults |
| US20180124437A1 (en) * | 2016-10-31 | 2018-05-03 | Twenty Billion Neurons GmbH | System and method for video data collection |
| CN108010021A (en) * | 2017-11-30 | 2018-05-08 | 上海联影医疗科技有限公司 | A kind of magic magiscan and method |
| CN108021788A (en) * | 2017-12-06 | 2018-05-11 | 深圳市新合生物医疗科技有限公司 | The method and apparatus of deep sequencing data extraction biomarker based on cell free DNA |
-
2018
- 2018-06-13 CN CN201810608329.4A patent/CN109117703B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110289109A1 (en) * | 2010-05-21 | 2011-11-24 | Sony Corporation | Information processing apparatus, information processing method, and program |
| CN102737232A (en) * | 2012-06-01 | 2012-10-17 | 天津大学 | Cleavage cell recognition method |
| US20170154202A1 (en) * | 2013-09-16 | 2017-06-01 | EyeVerify Inc. | Feature extraction and matching for biometric authentication |
| CN106795558A (en) * | 2014-05-30 | 2017-05-31 | 维里纳塔健康公司 | Detection fetus Asia chromosomal aneuploidy and copy number variation |
| US20170337676A1 (en) * | 2016-05-23 | 2017-11-23 | General Electric Company | Textural analysis of diffused disease in the lung |
| CN106127255A (en) * | 2016-06-29 | 2016-11-16 | 深圳先进技术研究院 | The sorting technique of a kind of cancer numeral pathological cells image and system |
| WO2018065027A1 (en) * | 2016-10-03 | 2018-04-12 | Total E&P Uk Limited | Modelling geological faults |
| US20180124437A1 (en) * | 2016-10-31 | 2018-05-03 | Twenty Billion Neurons GmbH | System and method for video data collection |
| CN106650796A (en) * | 2016-12-06 | 2017-05-10 | 国家纳米科学中心 | Artificial intelligence based cell fluorescence image classification method and system |
| CN107133569A (en) * | 2017-04-06 | 2017-09-05 | 同济大学 | The many granularity mask methods of monitor video based on extensive Multi-label learning |
| CN106991417A (en) * | 2017-04-25 | 2017-07-28 | 华南理工大学 | A kind of visual projection's interactive system and exchange method based on pattern-recognition |
| CN106991673A (en) * | 2017-05-18 | 2017-07-28 | 深思考人工智能机器人科技（北京）有限公司 | A kind of cervical cell image rapid classification recognition methods of interpretation and system |
| CN107563444A (en) * | 2017-09-05 | 2018-01-09 | 浙江大学 | A kind of zero sample image sorting technique and system |
| CN108010021A (en) * | 2017-11-30 | 2018-05-08 | 上海联影医疗科技有限公司 | A kind of magic magiscan and method |
| CN108021788A (en) * | 2017-12-06 | 2018-05-11 | 深圳市新合生物医疗科技有限公司 | The method and apparatus of deep sequencing data extraction biomarker based on cell free DNA |
Non-Patent Citations (3)
| Title |
|---|
| T. LIN等: ""Bilinear CNN Models for Fine-Grained Visual Recognition"", 《2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)》 * |
| XIANGTENG HE 等: ""Fast Fine-grained Image Classification via Weakly Supervised Discriminative Localization"", 《COMPUTER VISION AND PATTERN RECOGNITION》 * |
| 倪磊 等: ""铜绿假单胞菌蹭行运动单细胞分析方法的建立及应用"", 《生物工程学报》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109886321B (en) * | 2019-01-31 | 2021-02-12 | 南京大学 | Image feature extraction method and device for fine-grained classification of icing image |
| CN109886321A (en) * | 2019-01-31 | 2019-06-14 | 南京大学 | A kind of image characteristic extracting method and device for icing image fine grit classification |
| CN110675368B (en) * | 2019-08-31 | 2023-04-07 | 中山大学 | Cell image semantic segmentation method integrating image segmentation and classification |
| CN110675368A (en) * | 2019-08-31 | 2020-01-10 | 中山大学 | Cell image semantic segmentation method integrating image segmentation and classification |
| CN111291692A (en) * | 2020-02-17 | 2020-06-16 | 咪咕文化科技有限公司 | Video scene recognition method, device, electronic device and storage medium |
| CN111291692B (en) * | 2020-02-17 | 2023-10-20 | 咪咕文化科技有限公司 | Video scene recognition method, device, electronic equipment and storage medium |
| CN112816480A (en) * | 2021-02-01 | 2021-05-18 | 奎泰斯特(上海)科技有限公司 | Water quality enzyme substrate identification method |
| CN112560999B (en) * | 2021-02-18 | 2021-06-04 | 成都睿沿科技有限公司 | Target detection model training method and device, electronic equipment and storage medium |
| CN112560999A (en) * | 2021-02-18 | 2021-03-26 | 成都睿沿科技有限公司 | Target detection model training method and device, electronic equipment and storage medium |
| CN113516022B (en) * | 2021-04-23 | 2023-01-10 | 黑龙江机智通智能科技有限公司 | Fine-grained classification system for cervical cells |
| CN113516022A (en) * | 2021-04-23 | 2021-10-19 | 黑龙江机智通智能科技有限公司 | Fine-grained classification system for cervical cells |
| CN115700821A (en) * | 2022-11-24 | 2023-02-07 | 广东美赛尔细胞生物科技有限公司 | Cell identification method and system based on image processing |
| CN115700821B (en) * | 2022-11-24 | 2023-06-06 | 广东美赛尔细胞生物科技有限公司 | Cell identification method and system based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109117703B (en) | 2022-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109117703A (en) | A fine-grained identification-based method for identification of promiscuous cell types | |
| CN108562589B (en) | Method for detecting surface defects of magnetic circuit material | |
| CN110428432B (en) | Deep neural network algorithm for automatically segmenting colon gland image | |
| CN107665492B (en) | A deep network-based tissue segmentation method for colorectal panoramic digital pathological images | |
| CN108596038B (en) | Method for identifying red blood cells in excrement by combining morphological segmentation and neural network | |
| CN111582111B (en) | Cell component segmentation method based on semantic segmentation | |
| CN115546605A (en) | Training method and device based on image labeling and segmentation model | |
| CN111079620A (en) | Leukocyte image detection and identification model construction method based on transfer learning and application | |
| EP3848472A2 (en) | Methods and systems for automated counting and classifying microorganisms | |
| CN110517273A (en) | Cytology image partition method based on dynamic gradient threshold value | |
| CN108305253A (en) | A kind of pathology full slice diagnostic method based on more multiplying power deep learnings | |
| CN104573701B (en) | A kind of automatic testing method of Tassel of Corn | |
| CN117252813B (en) | Cervical liquid-based cell detection and identification method and system based on deep learning | |
| CN115909006A (en) | Mammary tissue image classification method and system based on convolution Transformer | |
| CN107256398A (en) | The milk cow individual discrimination method of feature based fusion | |
| CN110874835A (en) | Crop leaf disease resistance identification method, system, electronic device and storage medium | |
| CN115240041A (en) | Fracture extraction method of shale electron microscope scanning image based on deep learning segmentation network | |
| CN116705289A (en) | Cervical pathology diagnosis device based on semantic segmentation network | |
| WO2024259796A1 (en) | Cell discrimination model construction method and apparatus, electronic device, and storage medium | |
| CN116978543A (en) | Artificial intelligence-assisted bone marrow tumor pathological diagnosis device | |
| CN113506253A (en) | Sandstone pore detection method based on multilayer multi-core learning and region combination | |
| CN116664932A (en) | A Classification Method for Colorectal Cancer Pathological Tissue Images Based on Active Learning | |
| CN114092456A (en) | Cell fluorescence image discrimination method and system | |
| CN203970354U (en) | A kind of Tongue analytical system based on mobile terminal | |
| CN116468690B (en) | Subtype analysis system of invasive non-mucous lung adenocarcinoma based on deep learning |
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 |