CN109740483A - A kind of rice growing season detection method based on deep-neural-network - Google Patents
A kind of rice growing season detection method based on deep-neural-network Download PDFInfo
- Publication number
- CN109740483A CN109740483A CN201811602306.9A CN201811602306A CN109740483A CN 109740483 A CN109740483 A CN 109740483A CN 201811602306 A CN201811602306 A CN 201811602306A CN 109740483 A CN109740483 A CN 109740483A
- Authority
- CN
- China
- Prior art keywords
- rice
- image
- neural
- deep
- growing season
- 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
- 235000007164 Oryza sativa Nutrition 0.000 title claims abstract description 47
- 235000009566 rice Nutrition 0.000 title claims abstract description 47
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 41
- 238000001514 detection method Methods 0.000 title claims abstract description 17
- 240000007594 Oryza sativa Species 0.000 title 1
- 241000209094 Oryza Species 0.000 claims abstract description 46
- 238000013461 design Methods 0.000 claims abstract description 12
- 230000012010 growth Effects 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 10
- 239000000284 extract Substances 0.000 claims abstract description 6
- 238000013095 identification testing Methods 0.000 claims abstract description 4
- 238000005070 sampling Methods 0.000 claims description 18
- 238000001914 filtration Methods 0.000 claims description 11
- 230000035800 maturation Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 241000196324 Embryophyta Species 0.000 abstract description 7
- 238000003062 neural network model Methods 0.000 abstract description 3
- 238000012271 agricultural production Methods 0.000 abstract description 2
- 230000003698 anagen phase Effects 0.000 abstract description 2
- 230000007613 environmental effect Effects 0.000 abstract description 2
- 238000000034 method Methods 0.000 description 9
- 239000013598 vector Substances 0.000 description 9
- 238000012360 testing method Methods 0.000 description 7
- 201000010099 disease Diseases 0.000 description 5
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 230000004913 activation Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 229920003266 Leaf® Polymers 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 235000009754 Vitis X bourquina Nutrition 0.000 description 2
- 235000012333 Vitis X labruscana Nutrition 0.000 description 2
- 240000006365 Vitis vinifera Species 0.000 description 2
- 235000014787 Vitis vinifera Nutrition 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 244000068988 Glycine max Species 0.000 description 1
- 235000010469 Glycine max Nutrition 0.000 description 1
- 206010039509 Scab Diseases 0.000 description 1
- 241000607479 Yersinia pestis Species 0.000 description 1
- 240000008042 Zea mays Species 0.000 description 1
- 235000016383 Zea mays subsp huehuetenangensis Nutrition 0.000 description 1
- 235000002017 Zea mays subsp mays Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000013100 final test Methods 0.000 description 1
- 235000009973 maize Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to the image identification technical fields of agricultural production, and in particular to a kind of rice growing season detection method based on deep-neural-network, specific steps include S1: the image in acquisition rice each growth period；S2: the image of acquisition is pre-processed；S3: design is suitable for the neural network of rice growing season identification；S4: training neural network simultaneously extracts corresponding identification model；S5: identification test is carried out using trained identification model；S6: determine recognition result；S7: centrally stored；It is inputted without external other information, such as starts to plant date, environmental parameter etc., only can detect which stage rice growing season is in by a rice field image photograph using the present invention.The deep-neural-network model for being suitable for rice image detection is established, can accurately detect true correct growth phase in the rice similar in each category feature of rice.
Description
Technical field
The present invention relates to the image identification technical fields of agricultural production, and in particular to a kind of water based on deep-neural-network
Rice growth period detection method.
Background technique
Level of application of the computer vision technique in agricultural lags far behind industry, the computer vision application master of early stage
If identifying seed.Into in the 1980s, research object and application field gradually expand, image procossing is from simple vision
The visual information that simulation develops to substitution, explains people, and accelerate visual information acquisition etc..So far side respectively has been formd
The multiple technologies form such as visual simulation, micro image, macroscopic analysis, thermal imaging, internal image, machine vision is overweighted, and is respectively had
The application system of characteristic.China starts late to application study of the computer vision in agriculture project, mainly and concentrates on
Crop disease diagnosis aspect.Method therefor list is made slow progress in terms of identifying growth period, and still has larger difference compared with foreign countries
Away from also needing further to make efforts in terms of depth, range and practice.
With the continuous development of computer technology, about crops digital image recognition research also achieve it is good into
Exhibition.Such as document " plant pest recognition methods [J] the agricultural research of Tan Feng, Ma Xiaodan based on blade, 2013,31
(6): 41-43. " by calculating blade chromatic value, Multi-layer BP Neural Network model is established, realizes the detection identification of soybean leaves.
" Tian Youwen, Li Tianlai, Li Chenghua wait Method for recognition of grape disease [J] the agriculture project of based on support vector machines to document
Journal, 2015,23 (6): 175-180. " by the color and textural characteristics of extraction grape disease leaf, utilize support vector machines
(support vector machine, SVM) knows method for distinguishing and achieves effect more better than neural network recognization.Document
《Zhang S W,Shang Y J,Wang L.Plant disease recognition based on plant leaf
Image [J] .Journal of Animal&Plant Sciences, 2017,25 (3): 42-45. " it is also that spot is divided it
It extracts the color, shape and textural characteristics of scab again afterwards, is then classified by K arest neighbors (K-nearest neighbor, KNN)
Algorithm identifies 5 kinds of maize leafs.
Document above is to by extracting plant, and specific image features combination conventional sorting methods are to characteristic areas such as diseases
Not biggish image is detected.Although achieving certain recognition effect, this kind of for such as rice includes not move
Plant, transplant, turning green, tiller, jointing, booting, heading, milking maturity, multiple similar features such as maturation growth period recognition effect
And it is bad, particularly for needing accuracy to reach the identification requirement of couple of days, conventional method is unable to satisfy at all.Not only such as
This, based on traditional characteristic know method for distinguishing, it is also necessary to additional parameter, e.g. which day start plantation, carry out auxiliary sentence
Disconnected, this made work amount undoubtedly aggravated, there is no embody " intelligence ".But also it is influenced by picture quality very big.It is exposing
Degree, contrast influence the effect of identification very much in the case where either thering are other objects to block.And same one piece of rice field, pass through
Different angle goes to shoot.If going to identify using traditional manual features, since feature is fixed, cause generalization ability not strong, most
The result of identification mistake can be all caused afterwards.
Summary of the invention
To solve the above-mentioned problems, the present invention provides a kind of rice growing season detection side based on deep-neural-network
Method, specific technical solution are as follows:
Detection method includes the following steps for a kind of rice growing season based on deep-neural-network:
S1: the image in acquisition rice each growth period builds image data base；
S2: the image of acquisition is pre-processed, image is disturbed；
S3: design is suitable for the neural network of rice growing season identification；
S4: training neural network simultaneously extracts corresponding identification model；
S5: identification test is carried out using trained identification model；
S6: determine recognition result, accurate rate is counted, the convergency value of accurate rate is found out, whether determines the convergency value
The deep-neural-network structure of design is adjusted again if convergency value is less than preset value more than or equal to preset value；
S7: centrally stored, the image of storage identification mistake will when the amount of images of identification mistake reaches given threshold
All the picture of identification mistake, which is re-fed into neural network, is trained.
Preferably, in the step S1 by rice field fixed point camera and sensor collect never transplanting,
Transplant, turn green, tiller, jointing, booting, heading, milking maturity, maturation each growth period rice image, and according to corresponding life
It is arranged for a long time, completes the construction of image data base.
Preferably, image is disturbed in the step S2 specifically: using OpenCV to collected rice image
The disturbance of associated image parameter is carried out, the associated image parameter includes brightness, contrast, angle.
Preferably, the neural network designed in the step S3 is 7 layers of neural network, including C1 convolutional layer, S2 down-sampling
Layer, C3 convolutional layer, S4 down-sampling layer, the full linking layer of F5, the full linking layer of F6, output layer.
Preferably, the C1 convolutional layer uses side length for the square filtering device of 3-8 pixel, and quantity is 6, step-length s=
1, fill padding=0.
Preferably, the square filtering device that side length is 2-4 pixel is respectively adopted in the S2 down-sampling layer and S4 down-sampling layer,
Step-length s=2 fills padding=0.
Preferably, the C3 convolutional layer uses side length for the square filtering device of 3-8 pixel, and quantity is 16, step-length s=
1, fill padding=0.
The invention has the benefit that
1, it is inputted without external other information, such as starts to plant date, environmental parameter etc., can only led to using the present invention
Crossing a rice field image photograph can detect which stage rice growing season is in.
2, the deep-neural-network model for being suitable for rice image detection is established, it can the rice similar in each category feature of rice
In, accurately detect true correct growth phase.Compared with conventional method, no matter carrying out angle rotation or scaling etc. to picture
Operation, also or in different angle shoots same rice field, will not impact to recognition result.
3, in the case where image has shade or blocks, an accurately identification can be still made to rice image.
Detailed description of the invention
Fig. 1 is flow diagram of the invention；
Fig. 2 is the correspondence diagram of convolutional neural networks connection and matrix convolution；
Fig. 3 is the connection relationship diagram of a convolution node；
Fig. 4 is the structural schematic diagram of down-sampling layer.
Specific embodiment
In order to better understand the present invention, the present invention will be further explained below with reference to the attached drawings and specific examples:
As shown in Figure 1, detection method includes the following steps for a kind of rice growing season based on deep-neural-network:
S1: the image in acquisition rice each growth period builds image data base；Pass through the camera of the fixed point in rice field
And sensor collect never transplant, transplant, turning green, tiller, jointing, booting, heading, milking maturity, maturation each growth period
Rice image, and arranged according to corresponding growth period, complete the construction of image data base.
S2: the image of acquisition is pre-processed, image is disturbed；Original image is cut into the piecemeal of fixed size
Image, using OpenCV to collected rice image carry out associated image parameter disturbance, associated image parameter include brightness,
Contrast, angle.The size of block image is 100*100~300*300 pixel.For example, by the figure of original image 3456*2304 pixel
Block image as being cut into muti-piece 216*192 pixel, program input parameter Y=16, X=12, and wherein X is picture average transverse
The line number cut, Y are the average line number longitudinally cut of picture.The wherein relationship of original image and block image size are as follows: 3456/16=216,
2304/12=192.
Image is rotated, the angle of rotation is random number, is arranged brightness, the contrast of picture, the brightness of setting, right
It is random number than degree, can so enhances the generalization ability of the neural network recognization of foundation.
S3: design is suitable for the neural network of rice growing season identification, and the neural network of design is 7 layers of neural network, packet
Include C1 convolutional layer, S2 down-sampling layer, C3 convolutional layer, S4 down-sampling layer, the full linking layer of F5, the full linking layer of F6, output layer.
(1) C1 convolutional layer uses the square filtering device quantity of 5*5 pixel for 6, step-length s=1, fills padding=
0.That is: it is made of 6 characteristic pattern Feature Map, each neuron is connected with the neighborhood of 5*5 in input in characteristic pattern.Filtering
Device slides a pixel (stride=1) every time, and a characteristic spectrum uses the same filter.It, can be with by convolution algorithm
Enhance original signal feature, and reduces noise.
Convolutional neural networks connection is with the corresponding relationship of matrix convolution as shown in Fig. 2, the connection relationship of a convolution node
As shown in figure 3, the value of each upper layer node is added these products and an offset parameter to obtain one multiplied by the parameter in connection
It is a and, this and input activation primitive, the output of activation primitive is the value of next node layer.
The calculating of layer matrix width is exported after convolution:
Outlength=(inlength-fileterlength+2*padding)/stridelength+1；
Wherein, Outlength is the width for exporting layer matrix；Inlength is the width for inputting layer matrix；Padding is
Mend 0 circle number (inessential)；Stridelength is step-length, i.e., filter calculates primary result every several steps；
Fileterlength is the side length of filter.
(2) S2 down-sampling layer uses the square filtering device of 2*2 pixel, and step-length s=2 fills padding=0.That is: special
Each unit in sign figure is connected with the 2*2 neighborhood of characteristic pattern corresponding in C1, there is the characteristic pattern of 6 14*14.S2 down-sampling
Layer only has one group of hyper parameter f and s, the parameter for not needing to learn.S2 down-sampling layer utilizes the principle of image local correlation, right
Image carries out sub-sample, it is possible to reduce data processing amount retains useful information simultaneously, reduces the mistake of network training parameter and model
Fitting degree.
Down-sampling layer is using the input domain of 2x2 pixel, i.e., upper one layer of 4 nodes are as 1 node of next layer
Input, and input domain is not overlapped, i.e., filter slides 2 pixels every time, and the structure of down-sampling node is as shown in figure 4, under each
It is averaged (average pond) after 4 input nodes summation of sampling node and obtains a mean value, mean value is multiplied by a network weight
Parameter w adds input of the offset parameter bias as activation primitive, and the output of activation primitive is the value of next node layer.
Down-sampling layer is also average pond layer, i.e., is only averaging to characteristic point in neighborhood, specific as follows:
hmFor m-th of characteristic point average pond as a result, NmFor the characteristic point sum in m-th of feature vertex neighborhood, aiFor with
Ith feature point in neighborhood centered on m-th of characteristic point.
(3) C3 convolutional layer uses the square filtering device of 5*5 pixel, and quantity is 16, step-length s=1, fills padding
=0.That is: it is made of 16 characteristic pattern Feature Map, each neuron is connected with the neighborhood of 5*5 in input in characteristic pattern.
(4) the square filtering device of S4 down-sampling layer 2*2 pixel, step-length s=2 fill padding=0.That is: characteristic pattern
In each unit be connected with the 2*2 neighborhood of characteristic pattern corresponding in C3, have the characteristic pattern of 16 5*5.It does not need to learn
Parameter.
(5) the full linking layer of F5 has 120 units.It is connected entirely between each unit and S4 layers of all 400 units.
Such as classical neural network, the F5 layers of dot product calculated between input vector and weight vectors, along with a biasing.
(6) the full linking layer of F6 has 84 units.It is connected entirely between each unit and F5 layers of all 120 units.
The F6 layers of dot product calculated between input vector and weight vectors, along with a biasing.
(7) output layer is made of European radial basis function (Euclidean Radial Basis Function) unit, often
One unit of class each has 84 inputs.In other words, each output RBF unit calculates between input vector and parameter vector
Euclidean distance.Input it is remoter from parameter vector, RBF output it is bigger.
European radial basis function is specific as follows:
R=| | x-xi||；
Wherein, φ (r) is European radial distance, and r is the distance between two o'clock.X is origin, xiFor target point.
S4: training neural network simultaneously extracts corresponding identification model；Pretreated picture is input to designed mind
Be trained through network, and extract respectively iteration difference number 50 times, 100 times, 500 training patterns are for testing.
S5: identification test is carried out using trained identification model；Prepare the rice picture unrelated with training sample to carry out
Test, and counted.
S6: determining recognition result, for statistical analysis to accurate rate, finds out the convergency value of accurate rate, works as validation
When acc no longer increases, maximum value is exactly convergency value.Determine whether the convergency value is more than or equal to preset value, if convergency value is less than
Preset value is then again adjusted the deep-neural-network structure of design.Such as in the present embodiment, accurate rate preset value is
90%, if convergency value when accurate rate is restrained needs again to adjust the deep-neural-network structure of design less than 98%
It is whole.
S7: centrally stored, the image of storage identification mistake will when the amount of images of identification mistake reaches given threshold
All the picture of identification mistake, which is re-fed into, is trained to improve the accuracy of identification of the neural network of design in neural network.
Following table 1 is the convergent (when the number of iterations is set as 100) of rice picture recognition accuracy rate, the mind of design
Through network constantly training test, each iteration of training dataset is all with 64 pictures for one group, totally 12800 picture.Test number
According to each iteration is collected equally with 64 pictures for one group, totally 1280 picture, finally maintains picture recognition accuracy rate
93.3388% or so.Table 2 is to be tested using the neural network of design as a result, can be seen that the present invention mentions from test result
The feasibility of the algorithm of confession.Wherein:
SjIt is j-th of value of upper one layer of output vector S, expression is probability that this sample belongs to j-th of classification.yjBefore
There is a summation symbol in face, and the range of j is also 1 to classification number T, in this experiment due to output be never transplant, transplant, turn green,
Tiller, tillering regularity, jointing, booting, heading, milking maturity, mature totally 10 stages, therefore T takes 10.Therefore y be a 1*T to
Amount, there be T value in the inside, and only 1 value is 1, other T-1 value is all 0.The value of the corresponding position of true tag is 1,
He is 0.
SjIt is j-th of value of upper one layer of output vector S, expression is probability that this sample belongs to j-th of classification.yjBefore
There is a summation symbol in face, and the range of j is also 1 to classification number T, in this experiment due to output be never transplant, transplant, turn green,
Tiller, tillering regularity, jointing, booting, heading, milking maturity, mature totally 10 stages, therefore T takes 10.Therefore y be a 1*T to
There is T value in amount, the inside, and only 1 value is 1, other T-1 value is all 0.That value of the corresponding position of true tag is 1,
He is 0.
1 neural metwork training result of table
2 neural network final testing result of table
| Testing time | Test chart the piece number | Identify positive exact figures | Wrong identification number | Accuracy rate |
| For the first time | 100 | 94 | 6 | 94% |
| Second | 100 | 98 | 2 | 98% |
| For the third time | 100 | 95 | 5 | 95% |
| 4th time | 100 | 94 | 6 | 94% |
The present invention is not limited to above-described specific embodiment, and the foregoing is merely preferable case study on implementation of the invention
, it is not intended to limit the invention, any modification done within the spirit and principles of the present invention and changes equivalent replacement
Into etc., it should all be included in the protection scope of the present invention.
Claims (7)
1. a kind of rice growing season detection method based on deep-neural-network, it is characterised in that: the following steps are included:
S1: the image in acquisition rice each growth period builds image data base；
S2: the image of acquisition is pre-processed, image is disturbed；
S3: design is suitable for the neural network of rice growing season identification；
S4: training neural network simultaneously extracts corresponding identification model；
S5: identification test is carried out using trained identification model；
S6: determine recognition result, accurate rate is counted, the convergency value of accurate rate is found out, determines whether the convergency value is greater than
The deep-neural-network structure of design is adjusted again if convergency value is less than preset value equal to preset value；
S7: centrally stored, the image of storage identification mistake will be whole when the amount of images of identification mistake reaches given threshold
The picture of identification mistake, which is re-fed into neural network, to be trained.
2. a kind of rice growing season detection method based on deep-neural-network according to claim 1, it is characterised in that:
In the step S1 by rice field camera and the sensor collection of fixed point never transplant, transplant, turn green, tiller,
Jointing, booting, heading, milking maturity, maturation each growth period rice image, and arranged according to corresponding growth period, it is complete
At the construction of image data base.
3. a kind of rice growing season detection method based on deep-neural-network according to claim 1, it is characterised in that:
Image is disturbed in the step S2 specifically: associated image parameter is carried out to collected rice image using OpenCV
Disturbance, the associated image parameter includes brightness, contrast, angle.
4. a kind of rice growing season detection method based on deep-neural-network according to claim 1, it is characterised in that:
The neural network designed in the step S3 is under 7 layers of neural network, including C1 convolutional layer, S2 down-sampling layer, C3 convolutional layer, S4
The full linking layer of sample level, F5, the full linking layer of F6, output layer.
5. a kind of rice growing season detection method based on deep-neural-network according to claim 4, it is characterised in that:
The C1 convolutional layer uses side length for the square filtering device of 3-8 pixel, and quantity is 6, and padding=0 is filled in step-length s=1.
6. a kind of rice growing season detection method based on deep-neural-network according to claim 4, it is characterised in that:
The square filtering device that side length is 2-4 pixel, step-length s=2, filling is respectively adopted in the S2 down-sampling layer and S4 down-sampling layer
padding=0。
7. a kind of rice growing season detection method based on deep-neural-network according to claim 4, it is characterised in that:
The C3 convolutional layer uses side length for the square filtering device of 3-8 pixel, and quantity is 16, and padding=0 is filled in step-length s=1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811602306.9A CN109740483A (en) | 2018-12-26 | 2018-12-26 | A kind of rice growing season detection method based on deep-neural-network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811602306.9A CN109740483A (en) | 2018-12-26 | 2018-12-26 | A kind of rice growing season detection method based on deep-neural-network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109740483A true CN109740483A (en) | 2019-05-10 |
Family
ID=66359945
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811602306.9A Pending CN109740483A (en) | 2018-12-26 | 2018-12-26 | A kind of rice growing season detection method based on deep-neural-network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109740483A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110427922A (en) * | 2019-09-03 | 2019-11-08 | 陈�峰 | One kind is based on machine vision and convolutional neural networks pest and disease damage identifying system and method |
| CN110705347A (en) * | 2019-08-22 | 2020-01-17 | 仲恺农业工程学院 | Neural network-based method and system for identifying growth period of flowering cabbage and storage medium |
| CN110751019A (en) * | 2019-09-03 | 2020-02-04 | 武汉珈和科技有限公司 | High-resolution image crop automatic extraction method and device based on deep learning |
| CN111160250A (en) * | 2019-12-30 | 2020-05-15 | 安徽易刚信息技术有限公司 | Blueberry growing period detection method and device based on artificial neural network |
| CN111479128A (en) * | 2020-04-01 | 2020-07-31 | 黑龙江省农业科学院农业遥感与信息研究所 | System and method for popularizing agricultural planting technology by using Internet |
| CN111964723A (en) * | 2020-08-18 | 2020-11-20 | 合肥金果缘视觉科技有限公司 | Peanut short bud detecting system based on artificial intelligence |
| CN112907464A (en) * | 2021-02-01 | 2021-06-04 | 涂可致 | Underwater thermal disturbance image restoration method |
| CN114821182A (en) * | 2022-05-05 | 2022-07-29 | 安徽农业大学 | Rice growth stage image recognition method |
| CN114818909A (en) * | 2022-04-22 | 2022-07-29 | 北大荒信息有限公司 | Weed detection method and device based on crop growth characteristics |
| CN115131670A (en) * | 2022-09-02 | 2022-09-30 | 广州艾米生态人工智能农业有限公司 | Intelligent auditing method, system, device and equipment for rice pictures |
| CN115797771A (en) * | 2022-12-07 | 2023-03-14 | 北大荒信息有限公司 | Method and device for judging leaf age of rice in cold region, computer and storage medium |
| WO2024093653A1 (en) * | 2022-11-03 | 2024-05-10 | 中国科学院分子植物科学卓越创新中心 | Method for constructing prediction model for heading period of rice, and prediction method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103514441A (en) * | 2013-09-21 | 2014-01-15 | 南京信息工程大学 | Facial feature point locating tracking method based on mobile platform |
| US20140112543A1 (en) * | 2012-10-18 | 2014-04-24 | Fujitsu Limited | Image processing device and image processing method |
| CN104199027A (en) * | 2014-08-29 | 2014-12-10 | 中国科学院遥感与数字地球研究所 | Method for realizing large-area near real-time monitoring on phenological period of rice based on compactly polarimetric radar |
| CN105184226A (en) * | 2015-08-11 | 2015-12-23 | 北京新晨阳光科技有限公司 | Digital identification method, digital identification device, neural network training method and neural network training device |
| CN108764257A (en) * | 2018-05-23 | 2018-11-06 | 郑州金惠计算机系统工程有限公司 | A kind of pointer instrument recognition methods of various visual angles |
-
2018
- 2018-12-26 CN CN201811602306.9A patent/CN109740483A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140112543A1 (en) * | 2012-10-18 | 2014-04-24 | Fujitsu Limited | Image processing device and image processing method |
| CN103514441A (en) * | 2013-09-21 | 2014-01-15 | 南京信息工程大学 | Facial feature point locating tracking method based on mobile platform |
| CN104199027A (en) * | 2014-08-29 | 2014-12-10 | 中国科学院遥感与数字地球研究所 | Method for realizing large-area near real-time monitoring on phenological period of rice based on compactly polarimetric radar |
| CN105184226A (en) * | 2015-08-11 | 2015-12-23 | 北京新晨阳光科技有限公司 | Digital identification method, digital identification device, neural network training method and neural network training device |
| CN108764257A (en) * | 2018-05-23 | 2018-11-06 | 郑州金惠计算机系统工程有限公司 | A kind of pointer instrument recognition methods of various visual angles |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705347A (en) * | 2019-08-22 | 2020-01-17 | 仲恺农业工程学院 | Neural network-based method and system for identifying growth period of flowering cabbage and storage medium |
| CN110427922A (en) * | 2019-09-03 | 2019-11-08 | 陈�峰 | One kind is based on machine vision and convolutional neural networks pest and disease damage identifying system and method |
| CN110751019A (en) * | 2019-09-03 | 2020-02-04 | 武汉珈和科技有限公司 | High-resolution image crop automatic extraction method and device based on deep learning |
| CN111160250A (en) * | 2019-12-30 | 2020-05-15 | 安徽易刚信息技术有限公司 | Blueberry growing period detection method and device based on artificial neural network |
| CN111479128A (en) * | 2020-04-01 | 2020-07-31 | 黑龙江省农业科学院农业遥感与信息研究所 | System and method for popularizing agricultural planting technology by using Internet |
| CN111964723A (en) * | 2020-08-18 | 2020-11-20 | 合肥金果缘视觉科技有限公司 | Peanut short bud detecting system based on artificial intelligence |
| CN112907464A (en) * | 2021-02-01 | 2021-06-04 | 涂可致 | Underwater thermal disturbance image restoration method |
| CN114818909A (en) * | 2022-04-22 | 2022-07-29 | 北大荒信息有限公司 | Weed detection method and device based on crop growth characteristics |
| CN114818909B (en) * | 2022-04-22 | 2023-09-15 | 北大荒信息有限公司 | Weed detection method and device based on crop growth characteristics |
| CN114821182A (en) * | 2022-05-05 | 2022-07-29 | 安徽农业大学 | Rice growth stage image recognition method |
| CN115131670A (en) * | 2022-09-02 | 2022-09-30 | 广州艾米生态人工智能农业有限公司 | Intelligent auditing method, system, device and equipment for rice pictures |
| CN115131670B (en) * | 2022-09-02 | 2022-12-20 | 广州艾米生态人工智能农业有限公司 | Intelligent auditing method, system, device and equipment for rice pictures |
| WO2024093653A1 (en) * | 2022-11-03 | 2024-05-10 | 中国科学院分子植物科学卓越创新中心 | Method for constructing prediction model for heading period of rice, and prediction method |
| CN115797771A (en) * | 2022-12-07 | 2023-03-14 | 北大荒信息有限公司 | Method and device for judging leaf age of rice in cold region, computer and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109740483A (en) | A kind of rice growing season detection method based on deep-neural-network | |
| Jasim et al. | Plant leaf diseases detection and classification using image processing and deep learning techniques | |
| CN111986099B (en) | Tillage monitoring method and system based on convolutional neural network with residual error correction fused | |
| Zhang et al. | Automated robust crop-row detection in maize fields based on position clustering algorithm and shortest path method | |
| Kumar et al. | Resnet-based approach for detection and classification of plant leaf diseases | |
| CN111340141A (en) | Crop seedling and weed detection method and system based on deep learning | |
| CN107507170A (en) | A kind of airfield runway crack detection method based on multi-scale image information fusion | |
| CN109325495A (en) | A kind of crop image segmentation system and method based on deep neural network modeling | |
| CN113657326A (en) | Weed detection method based on multi-scale fusion module and feature enhancement | |
| Bukhari et al. | Assessing the impact of segmentation on wheat stripe rust disease classification using computer vision and deep learning | |
| CN109859101A (en) | The recognition methods of corps canopy thermal infrared images and system | |
| CN111291818B (en) | Non-uniform class sample equalization method for cloud mask | |
| Zhang et al. | Robust image segmentation method for cotton leaf under natural conditions based on immune algorithm and PCNN algorithm | |
| CN116977960A (en) | Rice seedling row detection method based on example segmentation | |
| CN114841961A (en) | Wheat scab detection method based on image enhancement and improvement of YOLOv5 | |
| CN117743975A (en) | Hillside cultivated land soil environment improvement method | |
| Balakrishna et al. | Tomato Leaf Disease Detection Using Deep Learning: A CNN Approach | |
| CN114782455B (en) | Cotton row center line image extraction method for agricultural machine embedded equipment | |
| Musyoka | Automatic delineation of small holder agricultural field boundaries using fully convolutional networks | |
| CN113673340B (en) | Pest type image identification method and system | |
| CN113989253A (en) | Farmland target object information acquisition method and device | |
| CN114140428A (en) | Method and system for detecting and identifying larch caterpillars based on YOLOv5 | |
| Deng et al. | A paddy field segmentation method combining attention mechanism and adaptive feature fusion | |
| Satoto et al. | Rice disease classification based on leaf damage using deep learning | |
| CN116052141B (en) | Crop growth period identification method, device, equipment and 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: 20190510 |
|
| RJ01 | Rejection of invention patent application after publication |