CN108764199B - Automatic identification method and system for invasive plant mikania micrantha - Google Patents
Automatic identification method and system for invasive plant mikania micrantha Download PDFInfo
- Publication number
- CN108764199B CN108764199B CN201810572277.XA CN201810572277A CN108764199B CN 108764199 B CN108764199 B CN 108764199B CN 201810572277 A CN201810572277 A CN 201810572277A CN 108764199 B CN108764199 B CN 108764199B
- Authority
- CN
- China
- Prior art keywords
- layer
- neural network
- training
- convolutional neural
- deep convolutional
- 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.)
- Active
Links
- 241001440840 Mikania micrantha Species 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 65
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 59
- 241000196324 Embryophyta Species 0.000 claims abstract description 35
- 238000012360 testing method Methods 0.000 claims abstract description 16
- 230000006870 function Effects 0.000 claims description 28
- 238000011176 pooling Methods 0.000 claims description 18
- 241001464837 Viridiplantae Species 0.000 claims description 14
- 238000002372 labelling Methods 0.000 claims description 3
- 230000002265 prevention Effects 0.000 abstract 1
- 230000004913 activation Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000977 initiatory effect Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the technical field of intelligent prevention and control of exotic invasive plants, and particularly relates to an automatic identification method and system of an invasive plant mikania micrantha. The method comprises the steps of constructing a deep convolutional neural network, training and testing the deep convolutional neural network by using field target invasive plants, and identifying the field invasive plants with identification, and has the characteristics of high identification speed, high accuracy and strong complex environment interference resistance.
Description
Technical Field
The invention belongs to the technical field of intelligent image identification, and particularly relates to an automatic identification method and system for invasive plant mikania micrantha.
Background
The field invasive plants seriously destroy the ecological environment of forests and farmlands, causing huge economic loss. The digital image automatic identification technology is utilized to accurately detect the field invasive plants in real time, further prevent the further diffusion of the invasive plants and have important significance in protecting the biodiversity.
Because the colors of a plurality of field invasive plants such as mikania micrantha and auxiliary main plants under visible light are similar, the visual identification degree is low, the field environment is complex and changeable, and the interference of unknown background factors is added, the real-time accurate identification of the field invasive plants is difficult to realize by the common image identification method. At present, the research on the automatic identification of the field invasive plants is few, the large-scale field invasive plants are effectively identified mainly based on high-altitude or satellite remote sensing images, but the large-scale field invasive plants in medium and small scales are often ignored as error points, and the overall identification accuracy is limited. Therefore, with the increasing severity of the damage of the invasive plants such as mikania micrantha and the defect of the automatic identification technology of the invasive plants, the development of a high-precision identification method for quickly and accurately identifying the invasive plants of the field targets is urgently needed.
Disclosure of Invention
In order to solve the problem that the invasive plant mikania micrantha lacks a high-precision and rapid identification method in the prior art, the invention provides an automatic identification method and system for the invasive plant mikania micrantha, and the automatic identification method and system have the characteristics of high precision and rapid identification speed.
The invention aims to provide an automatic identification method of mikania micrantha.
It is a further object of the present invention to provide an automatic mikania micrantha identification system.
The method for automatically identifying the invasive plant mikania micrantha according to the specific embodiment of the invention comprises the following steps:
constructing a deep convolutional neural network, wherein the deep convolutional neural network structure comprises an input layer, an intermediate layer, a full connection layer, a classifier and an output layer;
training a deep convolutional neural network by using the colorful high-definition image of the field target invasive plant of the training set to generate a deep convolutional neural network for identification;
verifying the accuracy of the deep convolutional neural network generated by training by using the color high-definition images of the field target invasive plants in the test set; when the accuracy is smaller than the preset threshold, adjusting the intermediate layer and the training parameters of the deep convolutional neural network structure, retesting the accuracy of the deep convolutional neural network, and repeating the steps until the accuracy is larger than or equal to the preset threshold to finish training;
and (3) carrying out type identification on the color high-definition images of the field target invasive plants to be identified by using the trained deep convolutional neural network, labeling the identification results and splicing.
According to the automatic identification method of the invasive plant mikania micrantha, which is provided by the embodiment of the invention, the color high-definition image of the field target plant and the color high-definition image of the field target invasive plant to be identified are respectively cut into a plurality of square input units.
According to the automatic identification method of the invasive plant mikania micrantha, training parameters are set to be 0.1-1 of learning rate attenuation factor, 10-100 of learning rate attenuation period, 0.0001-0.1 of initial learning rate, 10-1000 of maximum training cycle times and 10-200 of randomly extracted sample numbers, wherein the optimal parameters are set to be 0.8 of learning rate attenuation factor, 100 of learning rate attenuation period, 0.0005 of initial learning rate, 50 of maximum training cycle times and 100 of randomly extracted sample numbers.
According to the automatic identification method of the invasive plant mikania micrantha, the preset threshold is set to be 90% -100%, wherein the preferred preset threshold is 90%.
According to the automatic identification method of the invasive plant mikania micrantha of the embodiment of the invention, the number ratio of the training set to the testing set is 4: 1.
according to the automatic identification method of the invasive plant mikania micrantha, the classification number of the full connection layer is 3, and each classification number corresponds to the invasive target plant, the green plant background and the background except the green plant respectively; alternatively, the classification number of the full-link layer may be 2, and the classification number corresponds to the invasive plant mikania micrantha target and all backgrounds except the target.
An automatic identification system for the invasive plant mikania micrantha according to an embodiment of the present invention, the system comprising,
the model module is used for constructing a deep convolutional neural network;
the training module is used for training and generating a deep convolutional neural network for identification and verifying the accuracy of the deep convolutional neural network generated by training; if the recognition accuracy of the deep convolutional neural network generated by training is greater than or equal to a preset threshold, finishing the training, and if the recognition accuracy of the deep convolutional neural network generated by training is smaller than the preset threshold, reminding a user to adjust the intermediate layer of the deep convolutional neural network structure and the training parameters, and retraining the deep convolutional neural network;
and the identification module is used for identifying the type of the colorful high-definition image of the field invasive plant to be identified, marking the identification result and splicing.
According to the automatic identification system of the invasive plant mikania micrantha, the system further comprises an image acquisition module, and the image acquisition module is used for acquiring a color high-definition image of the field target invasive plant to be identified.
According to the automatic identification system of the invasive plant mikania micrantha, the system further comprises a data module, wherein the data module is used for dividing the color high-definition images of the field target invasive plant into a training set and a testing set, and cutting the color high-definition images into a plurality of square input units.
The invention has the beneficial effects that:
1. the trained deep convolutional neural network has high speed, high accuracy and strong complex environment interference resistance for identifying the target invasive plant;
2. the training process of the deep convolutional neural network is simple, and only the intermediate layer and the training parameters need to be adjusted;
3. the automatic target identification efficiency is high, and low-altitude color high-definition images of massive invasive plants can be identified.
Drawings
FIG. 1 is a schematic flow diagram of the method for automatic identification of the invasive plant Mikania micrantha of the present invention;
FIG. 2 is a schematic diagram of the structure of an automatic identification system for invasive plant Mikania micrantha according to the present invention;
FIG. 3 is a schematic diagram of a deep convolutional neural network structure according to the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Example 1
As shown in fig. 1, a schematic flow chart of the method for automatically identifying an invasive plant mikania micrantha according to the present invention includes the following steps:
s01, constructing the deep convolutional neural network:
the constructed deep convolutional neural network structure comprises an input layer, an intermediate layer, a full connection layer, a classifier and an output layer.
The intermediate layer structure can refer to intermediate layer structures of deep convolutional neural networks such as AlexNet, GoogleLeNet and ResNet, and can also be constructed by a basic structure that a convolutional layer is followed by an activation function and then followed by a pooling layer, wherein the activation function can be a sigmoid function with good binary classification, a tanh function with characteristic effect continuously enlarged in the training process, and a ReLU function with good gradient disappearance problem in the training process.
Wherein, the classification number of the full connecting layer is 3, which respectively corresponds to the invasive plant mikania micrantha target, the green plant background and the background except the green plant; the classification number of the full-connection layer can also be 2, and the classification number respectively corresponds to the invasive plant mikania micrantha target and all backgrounds except the target. The classifier can adopt softmax with better multi-classification, and also can adopt SVM with better two-classification.
And the output layer outputs and splices the picture results.
As shown in fig. 3, the deep convolutional neural network structure is:
an input layer: data, input is an RGB3 channel image of size 224 × 224;
a second layer: convolution2dLayer, filter function size 7 × 7, step size 2, no padding;
and a third layer: the activation function reluleyer;
a fourth layer: a pooling layer maxmachining 2dLayer, a filter function size of 3 multiplied by 3, a step length of 2 and no filling;
and a fifth layer: a regularization Layer Response-regularization Layer;
a sixth layer: a pooling layer maxmachining 2dLayer, a filter function size of 3 multiplied by 3, a step length of 2 and no filling;
seventh to twentieth: the initiation v1 structure of google lenet;
twenty-first to thirty-fourth layers: the initiation v1 structure of google lenet;
thirty-fifth layer: a pooling layer maxmachining 2dLayer, a filter function size of 3 multiplied by 3, a step length of 2 and no filling;
thirty-sixth to forty-ninth layers: the initiation v1 structure of google lenet;
a fifty-th layer: a pooling layer maxmachining 2dLayer with a function size of 3 x 3 and a step size of 2;
fifty-th layer: a pooling layer maxmachining 2dLayer with a function size of 7 × 7 and a step size of 1;
fifty-second layer: full connectivity layer, 3 channels;
a fifty-third layer: a sorting layer softmaxLayer;
an output layer: (iii) a classificationLayer.
The 7 x 7 large filter is used, so that the algorithm is more sensitive to the regional target, and the mikania micrantha target can be more accurately identified based on regional characteristics because the mikania micrantha is generally flaky. The use of the concept structure of google lenet can speed up the computation speed.
S02, dividing a certain amount of color high-definition images of the wild invasive plant mikania micrantha in different environments in different periods into a training set and a testing set according to a certain proportion, cutting the images of the training set and the testing set into input units of square small blocks according to a fixed size, wherein the size can be generally divided by 2, the sizes of the input units are 32,64,96,128, 224,384 and 512, the more the number of layers of the deep convolutional neural network is, the larger the selected fixed cutting size is, and all the square small blocks are marked according to the plant mikania micrantha target, the green plant background and the background except the green plant.
The colorful high-definition images of the wild invasive plant mikania micrantha used in the invention under different periods and different environments can be colorful high-definition images of the wild invasive plant mikania micrantha collected by the unmanned aerial vehicle in low-altitude flight, and the number of the high-definition images is 10 ten thousand or more.
For the collected wild invasive plant mikania micrantha color high-definition images, 80% of pictures can be used as a training set, and the rest 20% of pictures can be used as a testing set.
S03, inputting the square small blocks of the training set into the deep convolutional neural network, setting training parameters to train and generate the deep convolutional neural network for identification, inputting the square small blocks of the test set into the deep convolutional neural network generated by training, and verifying the accuracy of the deep convolutional neural network generated by training;
the setting of the training parameters generally includes: the learning rate attenuation factor is 0.8, the learning rate attenuation period is 100, the initial learning rate is 0.0005, the maximum number of training cycles is 50, and the number of randomly-extracted samples is 100; the preset threshold value in step S03 is 90%.
S04, judging the accuracy and the size of a preset threshold;
s05, if the accuracy is more than or equal to a preset threshold, ending the training;
s06, if the accuracy is smaller than a preset threshold value, adjusting the intermediate layer of the deep convolutional neural network structure and the training parameters, and repeating the steps S04-S06; the specific adjusting method is to increase or decrease the convolutional layer, the activation function reluLayer and the pooling layer maxParling 2dLayer, adjust the size of the filter function of the convolutional layer, the size and the step length of the pooling layer, and increase or decrease the size of the training parameters.
S07, cutting the field invasive plant mikania micrantha color high-definition image to be recognized into square small blocks according to the fixed size, adopting the depth convolution neural network generated by pre-training to recognize the types of the square small blocks, marking recognition results and splicing.
The invention relates to an automatic identification method of invasive plants, wherein colored high-definition images of wild invasive plants mikania micrantha to be identified can be acquired by unmanned aerial vehicle low-altitude flight at any time, weather, place and terrain.
Cutting a wild invasive plant mikania micrantha color high-definition image into 154 square small blocks according to the size of 224 × 224 pixels, inputting the square small blocks into a deep convolution neural network generated by pre-training, outputting and splicing an identification result, manually verifying that the identification accuracy is about 91.56%, and amplifying and displaying the wrongly identified square small blocks.
The automatic identification method for the invasive plant, disclosed by the invention, wherein the size of the input image can be selected according to the constructed deep convolutional neural network structure, the size can be generally divided by 2, the common sizes are 32,64,96,128, 224,384, 512 and the like, and the larger the number of layers of the deep convolutional neural network is, the larger the selected fixed cropping size is.
Example 2
As shown in fig. 2, fig. 2 is a schematic structural diagram of an automatic identification system for an invasive plant mikania micrantha according to the present invention, wherein the identification system comprises:
a model module 102, configured to construct the deep convolutional neural network, where the deep convolutional neural network structure includes an input layer, an intermediate layer, a fully-connected layer, a classifier, and an output layer.
The classification number of the full connection layer is 3, and the classification number corresponds to an invasive plant mikania micrantha target, a green plant background and a background except the green plant respectively; or the classification number of the full connection layer is 2, and the classification number corresponds to the invasive plant mikania micrantha target and all backgrounds except the target.
The data module 103 is used for dividing a certain amount of colorful high-definition images of the wild invasive plant mikania micrantha in different environments in different periods into a training set and a testing set according to a certain proportion, cutting the images of the training set and the testing set into square small blocks according to the fixed size, and labeling all the square small blocks according to an invasive plant mikania micrantha target, a green plant background and a background except the green plant.
The training module 104 is configured to input the square small blocks of the training set into the deep convolutional neural network, set training parameters to train and generate the deep convolutional neural network for recognition, input the square small blocks of the test set into the deep convolutional neural network generated by training, verify the accuracy of the deep convolutional neural network generated by training, finish training if the recognition accuracy of the deep convolutional neural network generated by training is greater than or equal to a preset threshold, and remind a user to adjust the intermediate layer of the deep convolutional neural network structure and the training parameters and retrain turn train the deep convolutional neural network if the recognition accuracy of the deep convolutional neural network generated by training is less than the preset threshold.
The identification module 101 is used for cutting the field invasive plant mikania micrantha color high-definition image to be identified into square small blocks according to a fixed size, and the deep convolutional neural network generated by pre-training is adopted to identify the types of the square small blocks, label the identification result and splice the square small blocks.
The small blocks cut into squares are generally divisible by 2, the common sizes are 32,64,96,128, 224,384, 512 and the like, and the larger the number of layers of the deep convolutional neural network, the larger the fixed cut size is selected.
Claims (2)
1. The automatic identification method of the invasive plant mikania micrantha is characterized by comprising the following steps:
constructing a deep convolutional neural network, wherein the structure of the deep convolutional neural network is as follows:
an input layer: the input is an RGB3 channel image of size 224 x 224,
a second layer: convolutional layer, filter function size 7 x 7, step size 2, no padding,
and a third layer: the function is activated in such a way that,
a fourth layer: pooling layer, filter function size 3 x 3, step size 2, no padding,
and a fifth layer: the layer of regularization is a layer of regularization,
a sixth layer: pooling layer, filter function size 3 x 3, step size 2, no padding,
seventh to twentieth: the acceptance v1 structure of google lenet,
twenty-first to thirty-fourth layers: the acceptance v1 structure of google lenet,
thirty-fifth layer: pooling layer, filter function size 3 x 3, step size 2, no padding,
thirty-sixth to forty-ninth layers: the acceptance v1 structure of google lenet,
a fifty-th layer: pooling layer, function size 3 x 3, step size 2,
fifty-th layer: pooling layer, function size 7 x 7, step size 1,
fifty-second layer: a full connection layer, the classification number of the full connection layer is set to be 3, each classification number respectively corresponds to the invasive target plant, the green plant background and the background except the green plant,
a fifty-third layer: the softmaxLayer has the characteristics of high temperature,
an output layer: a classification layer;
dividing the color high-definition image of a field target plant mikania micrantha into a training set and a testing set, wherein the quantity ratio of the training set to the testing set is 4: training the deep convolutional neural network by using color high-definition images of a training set to generate the deep convolutional neural network for identification, wherein the color high-definition images of a field target plant mikania micrantha are cut into a plurality of square input units;
verifying the accuracy of the deep convolutional neural network generated by training by using the color high-definition images of the field target invasive plant mikania micrantha in the test set, wherein the color high-definition images of the field target invasive plant to be identified are cut into a plurality of square input units;
when the accuracy is smaller than a preset threshold, adjusting an intermediate layer and training parameters of the deep convolutional neural network structure, testing the accuracy again, repeating the step until the accuracy is larger than or equal to the preset threshold, and finishing training, wherein the training parameters are set to be a learning rate attenuation factor of 0.8, a learning rate attenuation period of 100, an initial learning rate of 0.0005, the maximum number of training cycles of 50, the number of randomly extracted samples of 100, and the preset threshold is 90%;
and performing type identification on the input color high-definition image of the field target invasive plant to be identified by using the trained deep convolutional neural network, labeling the identification result and splicing.
2. An automatic identification system for invasive plant mikania micrantha, characterized in that the system comprises,
a model module for constructing a deep convolutional neural network,
the deep convolutional neural network structure is as follows:
an input layer: the input is an RGB3 channel image of size 224 x 224,
a second layer: convolutional layer, filter function size 7 x 7, step size 2, no padding,
and a third layer: the function is activated in such a way that,
a fourth layer: pooling layer, filter function size 3 x 3, step size 2, no padding,
and a fifth layer: the layer of regularization is a layer of regularization,
a sixth layer: pooling layer, filter function size 3 x 3, step size 2, no padding,
seventh to twentieth: the acceptance v1 structure of google lenet,
twenty-first to thirty-fourth layers: the acceptance v1 structure of google lenet,
thirty-fifth layer: pooling layer, filter function size 3 x 3, step size 2, no padding,
thirty-sixth to forty-ninth layers: the acceptance v1 structure of google lenet,
a fifty-th layer: pooling layer, function size 3 x 3, step size 2,
fifty-th layer: pooling layer, function size 7 x 7, step size 1,
fifty-second layer: a full connection layer, the classification number of the full connection layer is set to be 3, each classification number respectively corresponds to the invasive target plant, the green plant background and the background except the green plant,
a fifty-third layer: the softmaxLayer has the characteristics of high temperature,
an output layer: a classification layer;
the training module is used for training and generating a deep convolutional neural network for recognition, verifying the accuracy of the deep convolutional neural network generated by training, finishing the training if the recognition accuracy of the deep convolutional neural network generated by training is greater than or equal to a preset threshold value, and reminding a user to adjust the intermediate layer of the deep convolutional neural network structure and the training parameters and retraining the deep convolutional neural network if the recognition accuracy of the deep convolutional neural network generated by training is smaller than the preset threshold value;
the identification module is used for carrying out type identification on the color high-definition images of the field invasive plants to be identified, marking identification results and splicing;
the image acquisition module is used for acquiring a color high-definition image of the field target invasive plant to be identified;
and the data module is used for dividing the color high-definition images of the field target invasive plants into a training set and a test set and cutting the color high-definition images into a plurality of square input units.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810572277.XA CN108764199B (en) | 2018-06-06 | 2018-06-06 | Automatic identification method and system for invasive plant mikania micrantha |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810572277.XA CN108764199B (en) | 2018-06-06 | 2018-06-06 | Automatic identification method and system for invasive plant mikania micrantha |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108764199A CN108764199A (en) | 2018-11-06 |
| CN108764199B true CN108764199B (en) | 2022-03-25 |
Family
ID=63999128
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810572277.XA Active CN108764199B (en) | 2018-06-06 | 2018-06-06 | Automatic identification method and system for invasive plant mikania micrantha |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108764199B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188735A (en) * | 2019-06-10 | 2019-08-30 | 中国农业科学院深圳农业基因组研究所 | A kind of instruction plant recognition methods based on EO-1 hyperion |
| CN111339953B (en) * | 2020-02-27 | 2022-11-11 | 广西大学 | Clustering analysis-based mikania micrantha monitoring method |
| CN111339954B (en) * | 2020-02-27 | 2022-08-09 | 广西大学 | Mikania micrantha monitoring method based on image recognition |
| CN112016484B (en) * | 2020-08-31 | 2024-04-05 | 深圳市赛为智能股份有限公司 | Plant invasion evaluation method, plant invasion evaluation device, computer equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107392130A (en) * | 2017-07-13 | 2017-11-24 | 西安电子科技大学 | Classification of Multispectral Images method based on threshold adaptive and convolutional neural networks |
| WO2018075674A1 (en) * | 2016-10-22 | 2018-04-26 | Burden Keith Charles | Automated pruning or harvesting system for complex morphology foliage |
-
2018
- 2018-06-06 CN CN201810572277.XA patent/CN108764199B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018075674A1 (en) * | 2016-10-22 | 2018-04-26 | Burden Keith Charles | Automated pruning or harvesting system for complex morphology foliage |
| CN107392130A (en) * | 2017-07-13 | 2017-11-24 | 西安电子科技大学 | Classification of Multispectral Images method based on threshold adaptive and convolutional neural networks |
Non-Patent Citations (2)
| Title |
|---|
| 《基于高光谱遥感数据的入侵植物监测》;万华伟;《农业工程学报》;20101231;第59-64页 * |
| 《面向植被识别的无人机图像处理关键技术研究》;刘笃晋;《中国博士学位论文全文数据库》;20180115;第3、75-84页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108764199A (en) | 2018-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108764199B (en) | Automatic identification method and system for invasive plant mikania micrantha | |
| CN110298266B (en) | Deep neural network target detection method based on multiscale receptive field feature fusion | |
| CN110163187B (en) | F-RCNN-based remote traffic sign detection and identification method | |
| CN112434672B (en) | Marine human body target detection method based on improved YOLOv3 | |
| CN105740894B (en) | Semantic annotation method for hyperspectral remote sensing image | |
| CN112380921A (en) | Road detection method based on Internet of vehicles | |
| CN106991666B (en) | A kind of disease geo-radar image recognition methods suitable for more size pictorial informations | |
| CN111461190A (en) | Deep convolutional neural network-based non-equilibrium ship classification method | |
| CN109978032B (en) | Bridge crack detection method based on space pyramid cavity convolution network | |
| CN111898651A (en) | Tree detection method based on Tiny Yolov3 algorithm | |
| CN108875620B (en) | Invasive plant monitoring method and system | |
| CN109508756B (en) | Foundation cloud classification method based on multi-cue multi-mode fusion depth network | |
| CN113780132A (en) | Lane line detection method based on convolutional neural network | |
| CN111652835A (en) | Method for detecting insulator loss of power transmission line based on deep learning and clustering | |
| CN112364721A (en) | Road surface foreign matter detection method | |
| CN111291818B (en) | Non-uniform class sample equalization method for cloud mask | |
| CN107886065A (en) | A kind of Serial No. recognition methods of mixing script | |
| CN111815526B (en) | Rain image rainstrip removing method and system based on image filtering and CNN | |
| CN112395953A (en) | Road surface foreign matter detection system | |
| CN105160285A (en) | Method and system for recognizing human body tumble automatically based on stereoscopic vision | |
| CN108154199B (en) | High-precision rapid single-class target detection method based on deep learning | |
| CN112804522B (en) | Method and device for detecting abnormal conditions of camera | |
| CN113516652A (en) | Battery surface defect and adhesive detection method, device, medium and electronic equipment | |
| CN112288702A (en) | Road image detection method based on Internet of vehicles | |
| CN112364864A (en) | License plate recognition method and device, electronic equipment and storage 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |