CN112464766B - Automatic farmland land identification method and system - Google Patents
Automatic farmland land identification method and system Download PDFInfo
- Publication number
- CN112464766B CN112464766B CN202011287841.7A CN202011287841A CN112464766B CN 112464766 B CN112464766 B CN 112464766B CN 202011287841 A CN202011287841 A CN 202011287841A CN 112464766 B CN112464766 B CN 112464766B
- Authority
- CN
- China
- Prior art keywords
- farmland
- land
- image
- network model
- image sample
- 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
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000012549 training Methods 0.000 claims abstract description 48
- 239000002689 soil Substances 0.000 claims abstract description 23
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 claims description 19
- 238000003860 storage Methods 0.000 claims description 15
- 238000005520 cutting process Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000001131 transforming effect Effects 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 17
- 238000012360 testing method Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 6
- 238000012795 verification Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 4
- 241000196324 Embryophyta Species 0.000 description 3
- 238000013459 approach Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000013526 transfer learning Methods 0.000 description 3
- 230000002238 attenuated effect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 235000008733 Citrus aurantifolia Nutrition 0.000 description 1
- 235000011941 Tilia x europaea Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000004568 cement Substances 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000004571 lime Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (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 embodiment of the invention provides an automatic farmland head recognition method and system, comprising the following steps: acquiring farmland ground head images; inputting farmland land images into a land identification network model, and obtaining farmland operation environment types corresponding to the farmland land images according to the output result of the land identification network model; the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment types comprise crop-free farmland, crop-bearing farmland, green vegetation heads, yellow vegetation heads, bare soil heads and artificial facility heads. According to the farmland ground head automatic identification method and system provided by the embodiment of the invention, the farmland ground head image under the agricultural machinery operation scene is identified based on deep learning, the operation environment of the current agricultural machinery operation can be accurately obtained, the real-time performance is good, and a feasible technical solution can be provided for autonomous sensing of the ground head environment of the intelligent agricultural machinery.
Description
Technical Field
The invention relates to the field of agricultural machinery, in particular to an automatic farmland land identification method and device.
Background
The field operation process of the agricultural machine can be understood as follows: agricultural machinery moves regularly within a specific range for a specific target. Although the farmland plots are different from the agricultural machine units in form, each agricultural machine unit must perform a plot turning operation, so that the plot turning area of the farmland is the boundary of the agricultural machine operation range. The agricultural machinery automatic navigation technology can obviously reduce overlapping and omission among the ridges of the reciprocating operation, lighten the labor intensity of drivers, and the agricultural machinery automatic navigation system which is currently put into practical use is still in a semi-automatic stage and still needs manual operation at the turning position of the land.
One approach to achieving automatic navigation of agricultural machinery in the prior art is based on digital maps of the farmland. And drawing farmland plots by a mapping or remote sensing image mode, wherein plot boundaries are abstracted into linear attributes in a digital map of the farmland plots. In fact, however, the land for providing the agricultural machine with the turning is an area, so the precision of the digital map of the farmland at present cannot meet the requirement of automatic driving of the agricultural machine.
Another approach in the prior art for achieving automatic navigation of agricultural machinery is based on image recognition technology. Because the key step of image recognition is image feature extraction, the traditional image feature extraction mode uses an artificially designed extractor, and the local features such as color, texture, shape and the like are extracted through professional knowledge and a complex parameter adjusting process. In addition, the actual farmland land types are complex and various, and the farmland land boundaries of the planted crops can be adjacent farmland land boundaries or non-farmland areas (such as ridges, lime roads, ditches or bare soil and the like), so that the artificial feature extraction mode adopting the traditional image recognition has various limitations and cannot be well suitable for the characteristics of complex and varied farmland land environments under natural conditions.
Disclosure of Invention
In order to overcome the defect that the farmland soil head environment is complex and changeable under natural conditions, which is in the prior art, the embodiment of the invention provides the farmland soil head automatic identification method and device, which can acquire and automatically identify farmland soil head images under the farm machinery operation scene so as to meet the application requirements of intelligent agricultural machinery soil head environment sensing.
In a first aspect, an embodiment of the present invention provides a method for automatically identifying a farmland, which mainly includes: acquiring farmland ground head images; inputting farmland land images into a land identification network model, and obtaining farmland operation environment types corresponding to the farmland land images according to the output result of the land identification network model; the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment types comprise crop-free farmland, crop-bearing farmland, green vegetation heads, yellow vegetation heads, bare soil heads and artificial facility heads.
Optionally, after the farmland head image is acquired, the method further comprises: and cutting the farmland ground head image into an aspect ratio 1 by taking the short side of the farmland ground head image as the side length: 1; scaling the first image based on a nearest neighbor interpolation method to obtain a second image; pixels of the second image are identifiable pixels of the terrain identification network model.
Optionally, the ground identification network model is constructed based on MobileNet V 2.
Optionally, the network structure of the ground identification network model includes an input layer, a first convolution module, inverted Residual Block modules, a second convolution module, a global averaging pooling layer, a Dropout layer, a 1×1 convolution layer and a Softmax layer which are sequentially connected; the first convolution module and the second convolution module have the same structure and comprise a convolution layer, a batch normalization layer and Relu activation layers which are sequentially connected; the Inverted Residual Block modules comprise 7 modules Inverted Residual Block which are connected in sequence; shortcut is used between convolutions of 1 in each Inverted Residual Block.
Optionally, before inputting the farmland ground head image into the ground head recognition network model, the method further comprises: acquiring a plurality of farmland land image samples of each farmland operation environment type; taking the image top of each farmland ground head image sample as a cutting starting point, and cutting the farmland ground head image samples according to a preset size proportion in a central symmetry mode to obtain multi-size image samples; turning over each farmland ground head image sample according to a preset angle to obtain a multi-angle image sample; randomly transforming the brightness, contrast and chromaticity of each farmland head image sample to obtain a multi-color image sample; constructing a farmland ground head image sample set consisting of the farmland ground head image sample, the multi-size image sample, the multi-angle image sample and the multi-color image sample; each image sample in the farmland land image sample set corresponds to a farmland operation environment type label; and pre-training the land identification network model by using the farmland land image sample set.
Optionally, after acquiring the plurality of farmland soil head image samples of each farmland operation environment category, preprocessing each image sample in the farmland soil head image sample set is further included, including: clipping processing and scaling processing so that the pixel of each processed image sample is an identifiable pixel of the ground identification network model.
Optionally, before pre-training the ground truth recognition network model by using the farmland ground truth image sample set, the method further comprises: and pre-training the ground identification network model by taking the ImageNet as a source data set.
In a second aspect, an embodiment of the present invention further provides an automatic farmland head recognition device, mainly including a sensor module and a microprocessor module, where:
The sensor module is mainly used for acquiring farmland ground head images; the microprocessor module is loaded with a land head recognition network model and is mainly used for receiving farmland land head images so as to output farmland operation environment types corresponding to the farmland land head images.
The land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment comprises crop-free farmland, crop-bearing farmland, green vegetation land, yellow vegetation land, bare soil land and artificial facility land.
In a third aspect, an embodiment of the present invention further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, where the steps of the method for automatically identifying a farmland head according to any one of the above-mentioned methods are implemented when the processor executes the program.
In a fourth aspect, embodiments of the present invention also provide a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of a method for automatically identifying a farmland heading as described in any of the above.
According to the farmland ground head automatic identification method and device provided by the embodiment of the invention, the farmland ground head image under the agricultural machinery operation scene is identified based on deep learning, the operation environment of the current agricultural machinery operation can be accurately obtained, the real-time performance is good, and a feasible technical solution can be provided for autonomous sensing of the ground head environment of the intelligent agricultural machinery.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of an automatic farmland head recognition method provided by an embodiment of the invention;
FIG. 2 is a schematic diagram of classification of farmland soil environment according to an embodiment of the present invention;
Fig. 3 is a schematic diagram of a MobileNetV network structure according to an embodiment of the present invention;
FIG. 4 is a schematic structural view of an automatic farmland head recognition device according to an embodiment of the present invention;
FIG. 5 is a schematic flow chart of another automatic farmland head recognition method according to an embodiment of the present invention;
FIG. 6 is a visual confusion matrix diagram of automatic farmland head recognition results provided by the embodiment of the invention;
FIG. 7 is a schematic structural diagram of another automatic farmland head recognition device according to an embodiment of the present invention;
fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Fig. 1 is a flow chart of an automatic farmland head recognition method according to an embodiment of the present invention, as shown in fig. 1, including, but not limited to, the following steps:
step S1: acquiring farmland ground head images;
Step S2: and inputting the farmland land image into a land identification network model, and obtaining the farmland operation environment category corresponding to the farmland land image according to the output result of the land identification network model.
The land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels.
The farmland operation environment comprises crop-free farmland, crop-bearing farmland, green vegetation land, yellow vegetation land, bare soil land and artificial facility land.
The farmland ground head image can be photographed in real time through a high-resolution camera arranged on the operation agricultural machine.
In order to better implement the farmland ground environment automatic identification application, the camera is preferably designed to be a three-proofing industrial-grade firm shell with dustproof, shockproof and waterproof functions, so that the use requirement of the agricultural machine vehicle-mounted terminal is met. In addition, the camera is suggested to be fixed above the cab of the agricultural machine and mounted at a suitable depression angle with the ground so as to take an image of the ground environment of the farmland suitable for model recognition.
Fig. 2 is a schematic diagram of classification of farmland environment provided by the embodiment of the present invention, as shown in fig. 2, the embodiment of the present invention uses dry farmland operation of an agricultural machine as an application background, and provides a method for determining the classification of farmland operation environment:
Firstly, classifying the dry farmland operation environment scenes, and particularly dividing the dry farmland operation environment scenes into two major categories of farmlands and land crops. For farmland environments, it is divided into two cases of no crop coverage and crop coverage. For the ground environment, the ground environment is used as a transition zone of farmland, and can be specifically divided into two cases of adjacent farmland and adjacent non-farmland according to adjacent environments. For adjacent farmland type land, the three conditions of immature green crop coverage, mature yellow crop coverage and bare soil without crop coverage can be specifically classified according to the difference of ground coverage. Because the adjacent non-farmland land is more complex, the land can be divided into three conditions of vegetation coverage (weeds, shrubs and trees), ridge (bare soil) and artificial facilities according to the structure of landscape elements.
Through the above analysis of classification of farmland plots, classification studies can be performed from the viewpoint of image recognition, although the ground cover of some plot environments is different, for example, green crops in adjacent farmland plots and green vegetation in adjacent non-farmland plots. According to the characteristics of colors, textures and the like of farmland ground head images, 6 farmland operation environment categories are divided, and automatic image recognition research is carried out, wherein the method comprises the following steps of: ① No crop farmland image; ② Crop farmland images are available; ③ The green vegetation land image comprises two kinds of land adjacent to green crop coverage and shrubs, weeds and other green vegetation coverage; ④ Yellow vegetation land images, including mature crop coverage and yellow vegetation coverage of weeds, bushes and the like; ⑤ The bare soil land image comprises two land adjacent to the farmland white land and ridge; ⑥ And (3) artificial facility ground head images such as cement roads, stone slab roads, ditches, fences and the like.
After determining the farmland operation environment category, the embodiment of the invention provides a land identification network model constructed based on deep learning, which is used for learning and analyzing the image characteristics such as the color, the texture and the like of any farmland land image input in real time so as to output the farmland operation environment category corresponding to any farmland land image. It should be noted that, in the embodiment of the present invention, the network model architecture of the ground identification network model is not specifically limited.
Optionally, after the land identification network model is built, the land identification network model can be pre-trained by using a farmland land image sample with a farmland operation environment type label, and super parameters of the land identification network model are adjusted according to the training result of each time until the training result converges.
According to the farmland ground head automatic identification method provided by the embodiment of the invention, the farmland ground head image under the agricultural machinery operation scene is identified based on deep learning, the operation environment of the current agricultural machinery operation can be accurately obtained, the real-time performance is good, and a feasible technical solution can be provided for autonomous sensing of the ground head environment of the intelligent agricultural machinery.
Based on the foregoing embodiment, as an alternative embodiment, after acquiring the farmland head image, the method further includes: and cutting the farmland ground head image into an aspect ratio 1 by taking the short side of the farmland ground head image as the side length: 1; scaling the first image based on a nearest neighbor interpolation method to obtain a second image; pixels of the second image are identifiable pixels of the terrain identification network model.
Generally, since the size of the farmland ground head image collected in real time is generally larger than the input size of the ground head recognition network model, the farmland ground head image is directly input into the ground head recognition network model, which may cause excessive error of the recognition result and even failure in recognition. Therefore, in the farmland ground head automatic identification method provided by the embodiment of the invention, after the farmland ground head image is acquired, the size of the farmland ground head image needs to be adjusted on the basis of not losing the image characteristics.
Since the original image of the farmland head image acquired in real time is generally different in length and width, if the length and width are directly according to the aspect ratio 1:1, scaling can lead to deformation of the image content of farmland ground head images, thereby affecting the classification recognition effect. Therefore, in the embodiment of the present invention, the short side of the original image of the farmland ground head image is taken as the side length, and is cut into the aspect ratio 1: an image of 1 (hereinafter, collectively referred to as a first image).
Further, a nearest neighbor interpolation method is adopted, and each first image is scaled under the condition that image information is not lost as much as possible. For example, 224 x 224 pixels (i.e., identifiable pixels of the terrain identification network model) may be scaled to meet the pixel input requirements of the terrain identification network model.
The nearest neighbor interpolation (Nearest Neighbor interpolation, NNI) refers to setting the pixel value of each point of the second image as the nearest point in the first image. That is, the first image is regarded as a pixel point diagram a composed of innumerable pixels, then a pixel point diagram b with the same size but sparse pixels is covered on the first image, the gray value of each pixel point on the diagram b is searched for the gray value of the nearest pixel from the diagram a, and is assigned to the point on the diagram b, and after all the points on the diagram b are assigned, the first image can be reduced to the required multiple, and the pixel point diagram b corresponds to the second image.
According to the farmland ground head automatic identification method provided by the embodiment of the invention, the farmland ground head image is preprocessed before the farmland ground head image is analyzed by utilizing the pre-constructed ground head identification network model, so that the accuracy of model identification can be effectively improved.
Based on the foregoing embodiment, as an alternative embodiment, the location identification network model is constructed based on MobileNet V 2.
Because the ground automatic identification model needs to be operated on a vehicle-mounted computer of the operation agricultural machinery, the operation and storage performances of the model are lower than those of a desktop computer with the same price. Because the current popular deep learning model consumes huge computer resources and is not suitable for being directly deployed on vehicle-mounted equipment, the ground identification network model needs to be compressed.
The farmland ground automatic identification method provided by the embodiment of the invention uses MobileNet V 2 with a compact model structure design, the MobileNet V 2 is a lightweight deep neural network provided for embedded equipment such as mobile phones, and the aim of reducing model parameters can be fulfilled on the basis of not losing too much prediction precision by using a large number of depth separable convolutions to replace the traditional convolutions, so that the operation requirement of a vehicle-mounted computer can be met.
Based on the foregoing embodiments, as an optional embodiment, the network structure of the ground identification network model includes an input layer, a first convolution module, inverted Residual Block modules, a second convolution module, a global average pooling layer, a Dropout layer, a1×1 convolution layer, and a Softmax layer that are sequentially connected; the first convolution module and the second convolution module have the same structure and comprise a convolution layer, a batch normalization layer and Relu activation layers which are sequentially connected; the Inverted Residual Block modules comprise 7 modules Inverted Residual Block which are connected in sequence; shortcut is used between the convolutions with a sliding step of 1 in each of said Inverted Residual Block.
Fig. 3 is a schematic diagram of a network structure of MobileNet V 2 according to an embodiment of the present invention, as shown in fig. 3, an input farmland top image (provided that its pixels are 224×224) first passes through a convolution module (Conv-Block, hereinafter referred to as a first convolution module), which includes a convolution layer (Convolution), a batch normalization layer (BatchNorm), and a Relu activation layer. Then, a Inverted Residual Block module consisting of 7 Inverted Residual Block is used to perform dimension reduction processing to reduce the number of parameters, wherein the convolution layer with the sliding step length of 2 uses no shortcut, and the convolution layer with the sliding step length of 1 uses shortcut. After being output from Inverted Residual Block, the data is passed through a convolution module, and then a global average pooling layer (Global Average Pooling) is used for replacing a full connection layer, so that the parameter number is further reduced. Processing is then continued with the Dropout, 1 x 1convolution layer (1 x 1 Convolution) that prevents overfitting, and finally output by Softmax layer classification.
At the heart of MobileNet V 2 is a victory-disentangled convolution (Separable Convolution), among other things, which can effectively reduce the number of parameters at the expense of smaller performance.
The operation steps of each Inverted Residual Bloc are as follows: the input low-dimensional image features are amplified to high dimensions, then convolution operation is carried out in a high-dimensional convolution mode, and then linear convolution is used for mapping the low-dimensional image features into a low-dimensional space.
The embodiment of the invention introduces the thought of shortcut into the ground identification network model, and aims to solve the problem that gradient divergence in the ground identification network model is difficult to train by adding (weighted) shortcut between two convolution layers with the sliding step length of 1.
According to the farmland ground head automatic identification method provided by the embodiment of the invention, the lightweight model MobileNet V is selected as the prediction model, so that the accuracy is high, the identification speed is high, the memory occupation amount is high, the farmland ground head automatic identification method is supported to be used on a vehicle-mounted computer of an operation agricultural machine, and the ground head environment sensing application requirement is met. The farmland ground head images are identified based on the migration learning mode, so that a model with good generalization capability can be trained under the condition that the sample number of the farmland ground head images is limited, and the farmland ground head identification rate under the complex background is greatly improved.
Based on the foregoing embodiment, as an alternative embodiment, before inputting the farmland head image into the head recognition network model, the method may further include:
Acquiring a plurality of farmland land image samples of each farmland operation environment type;
Taking the image top of each farmland ground head image sample as a cutting starting point, and cutting the farmland ground head image samples according to a preset size proportion in a central symmetry mode to obtain multi-size image samples;
Turning over each farmland ground head image sample according to a preset angle to obtain a multi-angle image sample;
Randomly transforming the brightness, contrast and chromaticity of each farmland head image sample to obtain a multi-color image sample;
constructing a farmland ground head image sample set consisting of the farmland ground head image sample, the multi-size image sample, the multi-angle image sample and the multi-color image sample; each image sample in the farmland land image sample set corresponds to a farmland operation environment type label;
and pre-training the land identification network model by using the farmland land image sample set.
Further, after obtaining a plurality of farmland head image samples of each farmland operation environment category, preprocessing is further performed on each image sample in the farmland head image sample set, including: clipping processing and scaling processing so that the pixel of each processed image sample is an identifiable pixel of the ground identification network model.
The embodiment of the invention needs to pretrain the ground head recognition network model before analyzing the input farmland ground head image by using the ground head recognition network model, and the whole pretraining process is described in specific embodiments below.
First, collecting farmland land image samples for each farmland operation environment category, and constructing a training sample set by 6000 images in total. Wherein, each farmland operation environment category is 1000 respectively. Alternatively, farmland soil head image samples for each farmland operation environment category may be collected while constructing a verification sample set and a test sample set. The ratio of the number of images in the training sample set, the validation sample set, and the test sample set may be set to 4:1:1. the verification sample set is used for verifying the accuracy of the trained land identification network model, and total 1500 images are obtained, and each farmland operation environment category is 250; the test sample set is used for testing the practical application performance of the ground head recognition network model after training, and total 1500 images are obtained, and each farmland operation environment category is 250.
Further, each farmland ground head image sample is preprocessed in two aspects of image size cutting-scaling and image enhancement respectively, so that the obtained image samples can meet the input requirements of a ground head recognition network model, the image characteristic information of the image samples can be prevented from being lost as much as possible, and training set data can be effectively expanded and enhanced.
On the one hand, since the size of the collected farmland ground head image samples is generally larger than the input size of the ground head recognition network model, each farmland ground head image sample needs to be scaled to unify the image size.
Since the original image of the farmland head image sample is different in length and width, if the length and width of the original image are directly according to the length-width ratio 1:1, scaling can cause distortion of image content and affect classification recognition effect. Therefore, in the embodiment of the present invention, the short side of the original image is taken as the side length, and clipping is performed according to the preset size ratio (for example, clipping into an image with an aspect ratio of 1:1).
On the other hand, the cropped image is scaled to identifiable pixels of the ground truth recognition network model by employing nearest neighbor interpolation. And if the identifiable pixels of the ground identification network model are 224×224 pixels, correspondingly scaling the cropped image into 224×224 pixels so as to meet the input requirement of the network and ensure that the image information is not lost as much as possible.
By adopting the image size clipping-scaling method, one farmland ground head image sample can be clipped into two samples, and the problem of unbalanced data caused by insufficient quantity of specific categories can be solved.
Furthermore, a multi-scale clipping-scaling method can be adopted to expand the training sample set, for example, the original image is clipped and scaled to different scales, so that the farmland boundary features can be more obviously and comprehensively displayed. When the agricultural machinery runs close to the ground head area, useful information of the ground head firstly appears from the top end, therefore, when an image is cut, the image top end of each farmland ground head image sample is used as a cutting starting point, three scales of 60%, 70% and 80% of an original image are respectively cut in a mode of bilateral center symmetry, and then the image is scaled to the same size as the original image, so that 3 multi-size image samples corresponding to the original farmland ground head image sample can be obtained.
Furthermore, each farmland ground head image sample can be turned over according to a preset angle, and if each farmland ground head image sample is turned over horizontally once, a multi-angle image sample corresponding to the farmland ground head image sample can be added.
Further, each multi-size image sample may be flipped to obtain a multi-angle image sample corresponding to each multi-size image sample.
Further, the brightness, contrast and chromaticity of each multi-size image sample can be randomly transformed to eliminate the interference of illumination on the image, so as to obtain a plurality of multi-color image samples.
And finally, forming a farmland ground head image sample set by all the multi-size image samples, the multi-angle image samples, the multi-color image samples and original images of the farmland ground head image samples.
Further, in the embodiment of the invention, label labeling can be performed on each image sample in the farmland head image sample set based on TensorFlow deep learning frames. The training sample set, the verification sample set and the test sample set respectively comprise image samples corresponding to the listed 6 farmland operation environment categories. Under each image sample folder, image sample data are stored, the image sample data are converted into TFRecord data format files by using TensorFlow frames, a complete farmland land image sample set to be labeled is constructed, and a data foundation is laid for model training.
Further, in the farmland ground head automatic identification method provided by the embodiment of the invention, in the process of pre-training the ground head identification network model, the attenuated learning rate is used for training the network, and the value of the learning rate is attenuated, so that the parameter updating speed can be improved in the early stage of training, the network can be ensured not to have too large fluctuation in the later stage of training, and the optimal solution is easier to approach.
Further, in the process of pre-training the ground head recognition network model, a Momentum learning optimization algorithm is used, and the algorithm introduces Momentum of historical gradient information to accelerate (Stochastic GRADIENT DESCENT, SGD), so that the problem of large swing of the updating amplitude of the SGD optimization algorithm can be solved, and convergence to an optimal solution can be accelerated.
Further, in the process of pre-training the ground head recognition network model, in order to reduce the overfitting during training, L2 regularization is added.
Further, in pre-training the ground identification network model, exponential sliding averages are also used to enhance the generalization ability of the model.
According to the farmland ground head automatic identification method provided by the embodiment of the invention, the farmland ground head image sample is subjected to multi-scale expansion, the image is turned over, and the random transformation of the brightness, the contrast and the chromaticity of the image is realized, so that the farmland ground head image sample set is enhanced, the interference of illumination on the image is eliminated, and the reliability and the accuracy of model identification are improved.
Based on the foregoing embodiment, as an optional embodiment, before the pre-training the land feature identification network model by using the farmland land feature image sample set, the method further includes: and pre-training the ground identification network model by taking the ImageNet as a source data set.
Wherein ImageNet is a large visual database for visual object recognition software research. The transfer learning is a method for transferring knowledge in one domain (namely, source domain) to another domain (namely, target domain), so that the target domain can obtain better learning effect. According to the embodiment of the invention, a transfer learning method is adopted, the ImageNet is used as a source data set, and a source model MobileNet V 2 of the constructed ground identification network model is pre-trained to obtain a pre-trained model, wherein the pre-trained model has a certain image identification capability. And then, the constructed farmland ground head image sample set is used for retraining the ground head recognition network model again, and the model can complete the recognition task of the farmland ground head image more quickly by fine adjustment of model parameters.
According to the farmland ground head automatic identification method provided by the embodiment of the invention, the farmland ground head images are identified based on the migration learning mode, the source model MobileNet V 2 is pre-trained in advance, and then the ground head identification network model is pre-trained again by utilizing the farmland ground head image sample set, so that the model can be trained into a model with good generalization capability under the condition of limited ground head image sample size, and the farmland ground head identification rate under the complex background is greatly improved.
The embodiment of the invention provides an automatic farmland head recognition device, which is shown in fig. 4 and comprises, but is not limited to, a sensor module 1 and a microprocessor module 2. The sensor module 1 is mainly used for acquiring farmland ground head images; the microprocessor module is loaded with a land identification network model and is used for receiving the farmland land image and outputting farmland operation environment types corresponding to the farmland land image; the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment comprises crop-free farmland, crop-bearing farmland, green vegetation land, yellow vegetation land, bare soil land and artificial facility land.
Fig. 5 is a schematic flow chart of another farmland automatic recognition method provided by the embodiment of the present invention, and as shown in fig. 5, the whole flow of the farmland automatic recognition device provided by the embodiment of the present invention when working may include three parts of contents of data preparation, model training and model deployment.
In the data preparation stage, the method mainly comprises the steps of collecting farmland ground head images so as to obtain farmland ground head image sample sets containing a certain number of various farmland operation environment types, and carrying out image data enhancement processing on the collected farmland ground head image sample sets by adopting means of cutting-scaling, image enhancement and the like so as to provide more sufficient training sample sets, verification sample sets and test sample sets.
In the model training stage, mobileNet V 2 with compact model structural design is selected to construct a ground head recognition network model, then a transfer learning method is adopted, after the pre-training of MobileNet V 2 is completed by using ImageNet as a source data set, the constructed farmland ground head image sample set is used for pre-training the ground head recognition network model.
Further, after the model is pre-trained by using each farmland ground head image sample, the pre-trained ground head recognition network model can be verified by using a verification sample set, and super parameters of the model are properly adjusted according to a verification result so as to realize optimization of the model until the pre-training result converges, and the pre-training process of the whole model is completed, so that the optimal ground head recognition network model is obtained.
In the model deployment stage, the microprocessor module loaded with the land identification network model can be deployed in the agricultural machine vehicle-mounted device so as to realize the identification of the land environment of the farmland by utilizing the agricultural machine vehicle-mounted device.
It should be noted that, in order to accurately verify the actual classification accuracy of MobileNet V 2 networks, the embodiment of the invention uses a test sample set to test the network generalization capability of MobileNet V 2 after pre-training, and adopts a confusion matrix and an F1-score to evaluate the model identification accuracy.
In the image recognition accuracy evaluation, the accuracy, recall, and F1-score can be further calculated by the confusion matrix. FIG. 6 is a visual confusion matrix diagram of automatic farmland head recognition results provided by the embodiment of the invention; as shown in fig. 6, each column in the matrix represents a prediction category, and the total number of each column represents the number of data predicted as that category; each row represents the true home class of data, and the total number of data for each row represents the number of data instances for that class.
The Precision refers to the proportion of the number of correctly predicted positive examples of the model to the number of fully predicted positive examples. The Recall (Recall) is the proportion of the number of cases in which the model is correctly predicted to be positive to the number of all actual positive cases. F1—score can be regarded as a harmonic mean of model accuracy and recall, with a maximum of 1 and a minimum of 0, and its calculation formula is:
Table1 shows the accuracy, recall and F1-score results of identifying the land images of 6 types of farm work environment categories using the farm land automatic identification device provided by the embodiment of the invention.
TABLE 1
As shown in Table 1, mobileNet V 2 shows that the model of the ground identification network constructed based on MobileNet V 2 can accurately identify each type of sample when the Precision average value of the result of analyzing the test sample set is 0.97 and the average value of recall is 0.97; the average value of the F1-score is 0.97, which shows that the land identification network model can accurately identify 6 types of farmland land under the natural environment, has good robustness and robustness, and can meet the requirements of practical application.
Fig. 7 is a schematic diagram of another farmland automatic recognition device according to an embodiment of the present invention, where, as shown in fig. 7, the device uses a microprocessor module as a core, integrates functional modules such as an input module, an output module, a storage module, and a sensor module, and constructs an embedded terminal device suitable for application in a vehicle-mounted scene of an operation farm machine, so as to implement automatic recognition of farmland environments.
The micro-processing module is a core for controlling the whole device and is also a key module for operating the ground identification network model. The input module can be selected from a keyboard, a touch screen and other devices, and is mainly used for interaction of human-computer interfaces so as to realize configuration and operation of farmland head recognition application. The output module can be an LCD display screen, and is used for displaying real-time images in front of the running of the operation agricultural machinery and outputting the ground type identification result. The storage module can be a hard disk or a flash memory card, and is used for storing the model and the acquired ground environment image. The sensor module is used for acquiring images, can select a high-resolution camera, shoots image data in front of running of the operation agricultural machinery in real time, and inputs the image data to the microprocessor module through the interface.
Specifically, the whole process of the farmland head automatic recognition device recognizing the farmland head image can be as follows:
The training model at Tensorflow is saved as an h5 format and then converted to a tflite format model. And loading a ground head recognition network model through a Tensorflow Lite tool, configuring image prediction information while loading the model, acquiring an input layer and an input layer of a network, preprocessing and predicting farmland images acquired by a camera in real time, displaying a maximum probability prediction label, and realizing automatic recognition of farmland ground head images.
According to the farmland ground head automatic identification device provided by the embodiment of the invention, based on the embedded hardware technology modularization integration thought, based on the embedded main board, the function modules such as a microprocessor, an input, an output, a storage and an image acquisition sensor are integrated, the vehicle-mounted embedded terminal equipment is constructed, the vehicle-mounted embedded terminal equipment is arranged on an agricultural machinery, farmland images in front of driving are acquired in real time, and the automatic identification of the ground head environment is realized through a deployed farmland ground head image identification model.
It should be noted that, when the automatic farmland head recognition device provided in the embodiment of the present invention is specifically implemented, the automatic farmland head recognition device may be implemented based on the automatic farmland head recognition method described in any one of the above embodiments, which is not described in detail in this embodiment.
Fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, as shown in fig. 8, the electronic device may include: processor 810, communication interface (communication interface) 820, memory 830, and communication bus (bus) 840, wherein processor 810, communication interface 820, memory 830 accomplish communication with each other through communication bus 840. The processor 810 may call logic instructions in the memory 830 to perform the farmland heading automatic identification method, which basically comprises: acquiring farmland ground head images; inputting farmland land images into a land identification network model, and obtaining farmland operation environment types corresponding to the farmland land images according to the output result of the land identification network model; the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment types comprise crop-free farmland, crop-bearing farmland, green vegetation heads, yellow vegetation heads, bare soil heads and artificial facility heads.
Further, the logic instructions in the memory 830 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-only Memory (ROM), a random-access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Further, the logic instructions in the memory 830 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, an embodiment of the present invention further provides a computer program product, where the computer program product includes a computer program stored on a non-transitory computer readable storage medium, where the computer program includes program instructions, when the program instructions are executed by a computer, for executing the method for automatically identifying farmland head provided by the above method embodiments, where the method mainly includes: acquiring farmland ground head images; inputting farmland land images into a land identification network model, and obtaining farmland operation environment types corresponding to the farmland land images according to the output result of the land identification network model; the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment types comprise crop-free farmland, crop-bearing farmland, green vegetation heads, yellow vegetation heads, bare soil heads and artificial facility heads.
In still another aspect, an embodiment of the present invention further provides a non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor is implemented to perform the method for automatically identifying a farmland head provided in the above embodiments, and mainly includes: acquiring farmland ground head images; inputting farmland land images into a land identification network model, and obtaining farmland operation environment types corresponding to the farmland land images according to the output result of the land identification network model; the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels; the farmland operation environment types comprise crop-free farmland, crop-bearing farmland, green vegetation heads, yellow vegetation heads, bare soil heads and artificial facility heads.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (7)
1. An automatic farmland head recognition method is characterized by comprising the following steps:
Acquiring farmland ground head images;
Inputting the farmland land image into a land identification network model, and obtaining farmland operation environment types corresponding to the farmland land image according to the output result of the land identification network model;
the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels;
the farmland operation environment comprises crop-free farmland, crop-bearing farmland, green vegetation land, yellow vegetation land, bare soil land and artificial facility land;
The ground identification network model is constructed based on MobileNet V 2;
the network structure of the ground head recognition network model comprises an input layer, a first convolution module, a Inverted Residual Block module, a second convolution module, a global average pooling layer, a Dropout layer, a1 multiplied by 1 convolution layer and a Softmax layer which are connected in sequence;
the first convolution module and the second convolution module have the same structure and comprise a convolution layer, a batch normalization layer and Relu activation layers which are sequentially connected;
the Inverted Residual Block modules comprise 7 modules Inverted Residual Block which are connected in sequence;
Using shortcut between convolutions with a sliding step size of 1 in each of said Inverted Residual Block;
before inputting the farmland ground head image into the ground head recognition network model, the method further comprises the following steps:
Acquiring a plurality of farmland land image samples of each farmland operation environment type;
Taking the image top of each farmland ground head image sample as a cutting starting point, and cutting the farmland ground head image samples according to a preset size proportion in a central symmetry mode to obtain multi-size image samples;
Turning over each farmland ground head image sample according to a preset angle to obtain a multi-angle image sample;
Randomly transforming the brightness, contrast and chromaticity of each farmland head image sample to obtain a multi-color image sample;
constructing a farmland ground head image sample set consisting of the farmland ground head image sample, the multi-size image sample, the multi-angle image sample and the multi-color image sample; each image sample in the farmland land image sample set corresponds to a farmland operation environment type label;
and pre-training the land identification network model by using the farmland land image sample set.
2. The method for automatically identifying a farmland head according to claim 1, further comprising, after acquiring the farmland head image:
and cutting the farmland ground head image into an aspect ratio 1 by taking the short side of the farmland ground head image as the side length: 1;
Scaling the first image based on a nearest neighbor interpolation method to obtain a second image; pixels of the second image are identifiable pixels of the terrain identification network model.
3. The method of claim 1, further comprising, after obtaining a plurality of farmland footprint image samples for each type of farmland environment, pre-processing each image sample in the farmland footprint image sample set, comprising: clipping processing and scaling processing so that the pixel of each processed image sample is an identifiable pixel of the ground identification network model.
4. The method of claim 1, further comprising, prior to pre-training the terrain identification network model using the set of farmland terrain image samples:
And pre-training the ground identification network model by taking the ImageNet as a source data set.
5. An automatic farmland soil head recognition device, which is characterized by comprising: a sensor module and a microprocessor module;
The sensor module is used for acquiring farmland ground head images;
the microprocessor module is loaded with a land identification network model and is used for receiving the farmland land image and outputting farmland operation environment types corresponding to the farmland land image;
the land identification network model is obtained after training according to farmland land image samples with farmland operation environment type labels;
the farmland operation environment comprises crop-free farmland, crop-bearing farmland, green vegetation land, yellow vegetation land, bare soil land and artificial facility land;
The ground identification network model is constructed based on MobileNet V 2;
the network structure of the ground head recognition network model comprises an input layer, a first convolution module, a Inverted Residual Block module, a second convolution module, a global average pooling layer, a Dropout layer, a1 multiplied by 1 convolution layer and a Softmax layer which are connected in sequence;
the first convolution module and the second convolution module have the same structure and comprise a convolution layer, a batch normalization layer and Relu activation layers which are sequentially connected;
the Inverted Residual Block modules comprise 7 modules Inverted Residual Block which are connected in sequence;
Using shortcut between convolutions with a sliding step size of 1 in each of said Inverted Residual Block;
the farmland ground head automatic recognition device further comprises a model training module, wherein the model training module is used for:
Acquiring a plurality of farmland land image samples of each farmland operation environment type;
Taking the image top of each farmland ground head image sample as a cutting starting point, and cutting the farmland ground head image samples according to a preset size proportion in a central symmetry mode to obtain multi-size image samples;
Turning over each farmland ground head image sample according to a preset angle to obtain a multi-angle image sample;
Randomly transforming the brightness, contrast and chromaticity of each farmland head image sample to obtain a multi-color image sample;
constructing a farmland ground head image sample set consisting of the farmland ground head image sample, the multi-size image sample, the multi-angle image sample and the multi-color image sample; each image sample in the farmland land image sample set corresponds to a farmland operation environment type label;
and pre-training the land identification network model by using the farmland land image sample set.
6. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor performs the steps of the method for automatically identifying the head of a farmland according to any of claims 1 to 4 when the program is executed by the processor.
7. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, carries out the steps of the farmland head automatic identification method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011287841.7A CN112464766B (en) | 2020-11-17 | 2020-11-17 | Automatic farmland land identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011287841.7A CN112464766B (en) | 2020-11-17 | 2020-11-17 | Automatic farmland land identification method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112464766A CN112464766A (en) | 2021-03-09 |
| CN112464766B true CN112464766B (en) | 2024-07-02 |
Family
ID=74837380
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011287841.7A Active CN112464766B (en) | 2020-11-17 | 2020-11-17 | Automatic farmland land identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112464766B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113052369B (en) * | 2021-03-15 | 2024-05-10 | 北京农业智能装备技术研究中心 | Intelligent agricultural machinery operation management method and system |
| CN113469116A (en) * | 2021-07-20 | 2021-10-01 | 河南科技大学 | Face expression recognition method combining LBP (local binary pattern) features and lightweight neural network |
| CN113778110B (en) * | 2021-11-11 | 2022-02-15 | 山东中天宇信信息技术有限公司 | Intelligent agricultural machine control method and system based on machine learning |
| CN114067158B (en) * | 2021-11-17 | 2022-09-23 | 江苏天汇空间信息研究院有限公司 | Farmland use state monitoring system and method applying multi-source remote sensing data |
| CN114485612B (en) * | 2021-12-29 | 2024-04-26 | 广州极飞科技股份有限公司 | Route generation method and device, unmanned operation vehicle, electronic equipment and storage medium |
| CN114972807B (en) * | 2022-05-17 | 2023-03-28 | 北京百度网讯科技有限公司 | Method and device for determining image recognition accuracy, electronic equipment and medium |
| CN118071843B (en) * | 2023-11-20 | 2024-08-27 | 南京佳格耕耘科技有限公司 | Farmland mapping system based on remote sensing data |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102167038A (en) * | 2010-12-03 | 2011-08-31 | 北京农业信息技术研究中心 | Method and device for generating all-region-covering optimal working path for farmland plot |
| CN108776803A (en) * | 2018-04-20 | 2018-11-09 | 中国农业大学 | The method and system of weeds in a kind of removal farmland |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10664702B2 (en) * | 2016-12-30 | 2020-05-26 | International Business Machines Corporation | Method and system for crop recognition and boundary delineation |
| MX2021006830A (en) * | 2018-12-10 | 2021-07-02 | Climate Llc | Mapping field anomalies using digital images and machine learning models. |
| CN110825118B (en) * | 2019-12-23 | 2022-05-27 | 河南大学 | Multi-unmanned aerial vehicle cooperative farmland spraying method based on deep learning algorithm |
| CN111259898B (en) * | 2020-01-08 | 2023-03-24 | 西安电子科技大学 | Crop segmentation method based on unmanned aerial vehicle aerial image |
| CN111724340A (en) * | 2020-05-09 | 2020-09-29 | 北京农业智能装备技术研究中心 | Farmland margin line visual detection method and system |
| CN111814597A (en) * | 2020-06-20 | 2020-10-23 | 南通大学 | Urban function partitioning method coupling multi-label classification network and YOLO |
-
2020
- 2020-11-17 CN CN202011287841.7A patent/CN112464766B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102167038A (en) * | 2010-12-03 | 2011-08-31 | 北京农业信息技术研究中心 | Method and device for generating all-region-covering optimal working path for farmland plot |
| CN108776803A (en) * | 2018-04-20 | 2018-11-09 | 中国农业大学 | The method and system of weeds in a kind of removal farmland |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112464766A (en) | 2021-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112464766B (en) | Automatic farmland land identification method and system | |
| CN110148120B (en) | Intelligent disease identification method and system based on CNN and transfer learning | |
| CN108717524B (en) | Gesture recognition system based on double-camera mobile phone and artificial intelligence system | |
| CN111696101A (en) | Light-weight solanaceae disease identification method based on SE-Inception | |
| CN109344699A (en) | Winter jujube disease recognition method based on depth of seam division convolutional neural networks | |
| CN113159300A (en) | Image detection neural network model, training method thereof and image detection method | |
| CN114187450A (en) | Remote sensing image semantic segmentation method based on deep learning | |
| CN112419333B (en) | Remote sensing image self-adaptive feature selection segmentation method and system | |
| CN113657326A (en) | Weed detection method based on multi-scale fusion module and feature enhancement | |
| CN111062441A (en) | Scene classification method and device based on self-supervision mechanism and regional suggestion network | |
| CN111178438A (en) | ResNet 101-based weather type identification method | |
| CN113033321A (en) | Training method of target pedestrian attribute identification model and pedestrian attribute identification method | |
| CN111046793A (en) | Tomato disease identification method based on deep convolutional neural network | |
| CN111340019A (en) | Grain bin pest detection method based on Faster R-CNN | |
| CN114972208A (en) | YOLOv 4-based lightweight wheat scab detection method | |
| CN113435254A (en) | Sentinel second image-based farmland deep learning extraction method | |
| CN112861666A (en) | Chicken flock counting method based on deep learning and application | |
| CN117456358A (en) | Method for detecting plant diseases and insect pests based on YOLOv5 neural network | |
| CN116612386A (en) | Pepper disease and pest identification method and system based on hierarchical detection double-task model | |
| CN117197553A (en) | Fish abnormal state detection method and system based on improved Yolov5s | |
| CN112966698A (en) | Freshwater fish image real-time identification method based on lightweight convolutional network | |
| CN113011506A (en) | Texture image classification method based on depth re-fractal spectrum network | |
| CN116543325A (en) | Unmanned aerial vehicle image-based crop artificial intelligent automatic identification method and system | |
| CN116721385A (en) | Machine learning-based RGB camera data cyanobacteria bloom monitoring method | |
| CN115424243A (en) | Parking stall number identification method, equipment and medium based on yolov5-shufflenetv2 |
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 |