CN109859167A - The appraisal procedure and device of cucumber downy mildew severity - Google Patents

Abstract

The embodiment of the present invention provides the appraisal procedure and device of a kind of cucumber downy mildew severity, belongs to technical field of crop cultivation.This method comprises: obtaining collected cucumber leaves image to be assessed under natural environment；Using stochastic gradient descent method, and the neuralnetwork estimating model based on downy mildew severity, the cucumber downy mildew severity of cucumber leaves image is assessed.Method provided in an embodiment of the present invention, by using stochastic gradient descent method, and the neuralnetwork estimating model based on downy mildew severity, the cucumber downy mildew severity of cucumber leaves image is assessed.Due to that can estimate degree of disease automatically, to high degree of automation and recognition efficiency is high, manual intervention bring subjective impact can be effectively reduced, reduce the application cost and complexity of diagnosis process, the accuracy and real-time of disease screening can be effectively improved, also the correlative study for cucumber disease diagnosis provides reliable and accurate data basis.

Description

Method and device for evaluating the severity of cucumber downy mildew

technical field

The embodiments of the present invention relate to the technical field of crop cultivation, and in particular, to a method and device for evaluating the severity of downy mildew on cucumbers.

Background technique

Greenhouse cucumbers cause diseases due to various reasons during the planting process, resulting in reduced yield and quality. Downy mildew is one of the more common diseases in greenhouse cucumbers. The accurate diagnosis of the disease is divided into two aspects, one is the identification of the disease type, and the other is the estimation of the disease degree. Accurate acquisition of disease severity is a prerequisite for growers to scientifically prevent and control diseases, and is of great significance for reducing pesticide usage and improving economic benefits. The traditional disease degree estimation method mainly relies on the experience of growers, which is not only time-consuming and labor-intensive, but also highly subjective. In the related art, computer vision is usually used for disease diagnosis of greenhouse cucumbers. Specifically, the RGB images of cucumber leaves are used to obtain the color, texture and shape features of the disease spots, and a diagnosis model is established through the machine learning method, so as to realize the diagnosis of the disease. These methods can only identify the type of disease, but cannot assess the severity of the disease. Therefore, an effective method for evaluating the severity of downy mildew in cucumber is urgently needed.

SUMMARY OF THE INVENTION

In order to solve the above problems, the embodiments of the present invention provide a method and device for assessing the severity of downy mildew of cucumber by situational perception of greenhouse crops, which overcomes the above problems or at least partially solves the above problems.

According to a first aspect of the embodiments of the present invention, a method for evaluating the severity of cucumber downy mildew is provided, comprising:

Obtaining images of cucumber leaves to be evaluated collected in the natural environment;

The stochastic gradient descent method was used to evaluate the downy mildew severity of cucumber leaf images based on the neural network estimation model of downy mildew severity.

In the method provided by the embodiment of the present invention, by acquiring images of cucumber leaves to be evaluated collected in the natural environment, using the stochastic gradient descent method, and based on the neural network estimation model of downy mildew severity, the images of cucumber leaves of cucumber downy mildew Disease severity was assessed. Since the degree of disease can be automatically estimated, the degree of automation is high and the recognition efficiency is high, which can effectively reduce the subjective impact caused by manual intervention, reduce the application cost and complexity of the diagnosis process, and effectively improve the accuracy and real-time performance of disease diagnosis. It provides a reliable and accurate data basis for the related research on cucumber disease diagnosis.

According to a second aspect of the embodiments of the present invention, a device for evaluating the severity of cucumber downy mildew is provided, comprising:

an acquisition module, used to acquire the cucumber leaf images to be evaluated collected in the natural environment;

The evaluation module is used to evaluate the downy mildew severity of cucumber leaf images based on the stochastic gradient descent method and the neural network estimation model of downy mildew severity.

According to a third aspect of the embodiments of the present invention, an electronic device is provided, including:

at least one processor; and

at least one memory communicatively coupled to the processor, wherein:

The memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the method for evaluating the severity of downy mildew on cucumber provided by any one of the possible implementations of the first aspect.

According to a fourth aspect of the present invention, a non-transitory computer-readable storage medium is provided, the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause a computer to execute any of the various possible implementations of the first aspect. A possible implementation provides a method for assessing the severity of downy mildew in cucumbers.

It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory and are not limiting of embodiments of the present invention.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

1 is a schematic flowchart of a method for evaluating the severity of downy mildew on cucumbers provided in the embodiment of the present invention;

2 is a schematic structural diagram of a neural network estimation model for the severity of downy mildew provided by an embodiment of the present invention;

3 is a schematic structural diagram of a device for evaluating the severity of downy mildew on cucumber according to an embodiment of the present invention;

FIG. 4 is a block diagram of an electronic device according to an embodiment of the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in 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. Obviously, the described embodiments These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Greenhouse cucumbers cause diseases due to various reasons during the planting process, resulting in reduced yield and quality. Downy mildew is one of the more common diseases in greenhouse cucumbers. The accurate diagnosis of the disease is divided into two aspects, one is the identification of the disease type, and the other is the estimation of the disease degree. Accurate acquisition of disease severity is a prerequisite for growers to scientifically prevent and control diseases, and is of great significance for reducing pesticide usage and improving economic benefits. The traditional disease degree estimation method mainly relies on the experience of growers, which is not only time-consuming and labor-intensive, but also highly subjective. In the related art, computer vision is usually used for disease diagnosis of greenhouse cucumbers. Specifically, the RGB images of cucumber leaves are used to obtain the color, texture and shape features of the disease spots, and a diagnosis model is established through the machine learning method, so as to realize the diagnosis of the disease. These methods can only identify the type of disease, but cannot assess the severity of the disease. Therefore, an effective method for evaluating the severity of downy mildew in cucumber is urgently needed.

A small number of studies use shallow machine learning methods to build estimation models for the degree of disease. Although there are certain effects, it is necessary to perform early disease spot segmentation and artificially set image features. Due to the influence of complex background and illumination in the natural environment, the segmentation is difficult. The accuracy is difficult to guarantee, which makes it difficult to expand the use of these methods in the natural environment. Convolutional neural network is an unsupervised learning method with self-learning ability, which is considered to be one of the most effective ways of image recognition at present. Convolutional neural networks have been widely used in agriculture, and their obvious advantages in image recognition provide us with an idea. Therefore, research on the quantitative estimation method of downy mildew severity of greenhouse cucumber based on convolutional neural network can provide support for accurate diagnosis of cucumber disease.

Based on the above description, in view of the existing problems, the embodiment of the present invention provides a method for evaluating the severity of downy mildew of cucumber. Referring to Figure 1, the method includes:

101. Acquire an image of a cucumber leaf to be evaluated collected in a natural environment.

102. Using stochastic gradient descent method, and based on the neural network estimation model of downy mildew severity, evaluate the cucumber downy mildew severity of cucumber leaf images.

In the method provided by the embodiment of the present invention, by acquiring images of cucumber leaves to be evaluated collected in the natural environment, using the stochastic gradient descent method, and based on the neural network estimation model of downy mildew severity, the images of cucumber leaves of cucumber downy mildew Disease severity was assessed. Since the degree of disease can be automatically estimated, the degree of automation is high and the recognition efficiency is high, which can effectively reduce the subjective impact caused by manual intervention, reduce the application cost and complexity of the diagnosis process, and effectively improve the accuracy and real-time performance of disease diagnosis. It provides a reliable and accurate data basis for the related research on cucumber disease diagnosis.

Based on the content of the above embodiment, as an optional embodiment, before evaluating the cucumber downy mildew severity of the cucumber leaf image based on the neural network estimation model of downy mildew severity, the method further includes: The collected sample images of cucumber downy mildew leaves are preprocessed, and the original data set is constructed based on the preprocessed sample images; based on the sample images in the original data set and the disease degree value of the sample images, the original neural network model is performed After training, the neural network estimation model of downy mildew severity is obtained. The original neural network model may be a convolutional neural network model, which is not specifically limited in this embodiment of the present invention.

Based on the content of the foregoing embodiment, as an optional embodiment, regarding the method of preprocessing the sample images of cucumber downy mildew diseased leaves collected in the natural environment, the embodiment of the present invention does not specifically limit this, including but not limited to Limited to: remove sample images whose resolution is lower than the preset threshold from the collected sample images; remove the background pattern of each sample image, and adjust each sample image to a preset size.

Specifically, after acquiring the sample images collected in the greenhouse environment, images with lower quality, such as images with lower resolution or definition, may be removed first. In addition, normalization processing may also be performed on the sample image, that is, the sample image is adjusted to the same color space and the same size, such as 128×128 pixels, which is not specifically limited in this embodiment of the present invention. It should be noted that the larger the size of the sample label image, the higher the computational cost. In addition, the background patterns of the remaining sample images can also be removed to reduce irrelevant information in the images.

Based on the content of the foregoing embodiment, as an optional embodiment, before the original neural network model is trained based on the original data set to obtain the neural network estimation model for the severity of downy mildew, the method further includes: according to a preset processing method , to process the sample images in the original data set to expand the original data set, the preset processing method includes at least any one of the following three methods, the following three methods are color jitter, horizontal flip and vertical flip respectively.

Specifically, color dithering, flipping in horizontal and vertical directions can be performed on the sample image, and data enhancement can be performed by rotating at 90°, 180°, and 270°, so as to expand the sample. It should be noted that the main purpose of expanding the sample images here is to improve the estimation effect of the subsequent network model. During the training process of the network model, the sample images obtained before and expanded in this step can be used as the training set for training. The training set can be divided into two parts according to the function, namely the training set and the test set. When dividing the data set, it should be divided according to an appropriate ratio to ensure that the data volume of each type of data set is relatively balanced.

Based on the content of the above embodiment, as an optional embodiment, the neural network estimation model of downy mildew severity includes an input layer, 5 convolutional layers, 4 pooling layers, 4 batch normalization layers, 2 full connection layer and output layer. Certainly, a dropout layer may also be included, and the dropout layer may be placed before the two fully connected layers, which is not specifically limited in this embodiment of the present invention.

Based on the content of the above embodiment, as an optional embodiment, the input layer is connected to the first convolutional layer and the first batch normalization layer, and the first convolutional layer and the first batch normalization layer are connected to the first convolutional layer and the first batch normalization layer. The pooling layer is connected. The first pooling layer is connected to the second convolutional layer and the second batch normalization layer. The second convolutional layer and the second batch normalization layer are connected to the second pooling layer. Two pooling layers are connected to the third convolutional layer and the third batch normalization layer, the third convolutional layer and the third batch normalization layer are connected to the third pooling layer, and the third pooling layer is connected to The fourth convolutional layer and the fourth batch normalization layer are connected, the fourth convolutional layer and the fourth batch normalization layer are connected with the fourth pooling layer, and the fourth pooling layer is connected with the fifth convolutional layer The fifth convolutional layer is connected to two fully connected layers and the output layer in turn. For the connection relationship between the layers, reference may be made to FIG. 2 .

Specifically, the input size of the model is 128 × 128 pixels, the size of the convolution kernel in the convolution layer is 5 × 5, and the number of convolution kernels in the convolution layer can be 32, 64, 128, 256 and 512, respectively. After each convolution operation, the network will effectively extract the features in the image and generate the corresponding number of feature maps. The pooling layer uses a 2×2 convolution kernel for average pooling to achieve downsampling of feature maps. After 4 pooling layers, the weight parameters in the network structure can be greatly reduced and the computational cost can be reduced. The last convolutional layer is followed by 2 fully connected layers, which vectorize all feature maps and represent the features of the entire image as 1D vectors. A dropout layer is added before the fully connected layer, and the neural network unit is temporarily discarded from the network according to a certain probability, thereby preventing overfitting and improving the accuracy of model recognition. Finally there is a regressionLayer regression layer.

Among them, the size of the output feature map of the convolution operation can be expressed by the following formula:

W ⁱ⁺¹ =(W ⁱ -F+2P)/S+1

In the above formula, ^Wi represents the input image size, F represents the size of the convolution kernel, and P and S represent padding pixels and stride, respectively. For a convolution operation, its input-output relationship can be expressed by the following formula:

In the above formula, l represents the layer index, i represents the input feature map index, k represents the output feature map index, Indicates that the ith feature map on the l-1th layer is used as input. represents the output of the kth feature map on the lth layer. In addition, W represents the convolution weight tensor, b represents the bias parameter, and f( ) represents the activation function.

The pooling layer reduces the size of the received results. For details, please refer to the following formula:

where down( ) is the downsampling function, F is the downsampling filter size, and S is the downsampling step size.

Based on the contents of the foregoing embodiments, the embodiments of the present invention further provide a device for evaluating the severity of downy mildew on cucumber, which is used to perform the method for evaluating the severity of downy mildew on cucumber provided in the above method embodiments. Referring to FIG. 3, the apparatus includes: an acquisition module 301 and an evaluation module 302; wherein,

The acquisition module 301 is used for acquiring the cucumber leaf image to be evaluated collected in the natural environment;

The evaluation module 302 is configured to use the stochastic gradient descent method and based on the neural network estimation model of downy mildew severity to evaluate the cucumber downy mildew severity of the cucumber leaf image.

Based on the content of the foregoing embodiment, as an optional embodiment, the apparatus further includes:

The preprocessing module is used to preprocess the sample images of the cucumber downy mildew leaves collected in the natural environment, and construct the original data set based on the preprocessed sample images;

The training module is used to train the original neural network model based on the sample images in the original data set and the disease degree values of the sample images, and obtain the neural network estimation model of downy mildew severity.

Based on the content of the foregoing embodiment, as an optional embodiment, the preprocessing module is configured to remove sample images whose resolution is lower than a preset threshold from the collected sample images; remove the background pattern of each sample image, and Resize each sample image to a preset size.

Based on the content of the foregoing embodiment, as an optional embodiment, the apparatus further includes:

The expansion module is used to process the sample images in the original data set according to the preset processing method to expand the original data set. The preset processing method includes at least any one of the following three methods, and the following three methods are respectively For color dithering, flipping horizontally and flipping vertically.

Based on the content of the above embodiment, as an optional embodiment, the neural network estimation model of downy mildew severity includes an input layer, 5 convolutional layers, 4 pooling layers, 4 batch normalization layers, 2 full connection layer and output layer.

Based on the content of the above embodiment, as an optional embodiment, the input layer is connected to the first convolutional layer and the first batch normalization layer, and the first convolutional layer and the first batch normalization layer are connected to the first convolutional layer and the first batch normalization layer. The pooling layer is connected. The first pooling layer is connected to the second convolutional layer and the second batch normalization layer. The second convolutional layer and the second batch normalization layer are connected to the second pooling layer. Two pooling layers are connected to the third convolutional layer and the third batch normalization layer, the third convolutional layer and the third batch normalization layer are connected to the third pooling layer, and the third pooling layer is connected to The fourth convolutional layer and the fourth batch normalization layer are connected, the fourth convolutional layer and the fourth batch normalization layer are connected with the fourth pooling layer, and the fourth pooling layer is connected with the fifth convolutional layer The fifth convolutional layer is connected to two fully connected layers and the output layer in turn.

The device provided by the embodiment of the present invention obtains images of cucumber leaves to be evaluated collected in the natural environment, adopts the stochastic gradient descent method, and is based on the neural network estimation model of downy mildew severity. Disease severity was assessed. Since the degree of disease can be automatically estimated, the degree of automation is high and the recognition efficiency is high, which can effectively reduce the subjective impact caused by manual intervention, reduce the application cost and complexity of the diagnosis process, and effectively improve the accuracy and real-time performance of disease diagnosis. It provides a reliable and accurate data basis for the related research on cucumber disease diagnosis.

FIG. 4 illustrates a schematic diagram of the physical structure of an electronic device. As shown in FIG. 4 , the electronic device may include: a processor (processor) 410, a communication interface (Communications Interface) 420, a memory (memory) 430, and a communication bus 440, The processor 410 , the communication interface 420 , and the memory 430 communicate with each other through the communication bus 440 . The processor 410 can call the logic instructions in the memory 430 to execute the following methods: obtain the cucumber leaf images to be evaluated collected in the natural environment; adopt the stochastic gradient descent method, and based on the neural network estimation model of downy mildew severity, Cucumber downy mildew severity was assessed for cucumber leaf images. It should be noted that, in actual implementation, the electronic device may be in the form of a PC or a tablet computer. The PC or the tablet computer can collect data and have a decision-making control function, which is not specifically limited in this embodiment of the present invention.

In addition, the above-mentioned logic instructions in the memory 430 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product in essence, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, an electronic device, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

Embodiments of the present invention further provide a non-transitory computer-readable storage medium on which a computer program is stored, and the computer program is implemented when executed by a processor to execute the methods provided by the foregoing embodiments, for example, including: The collected cucumber leaf images to be evaluated; the stochastic gradient descent method is used to evaluate the cucumber downy mildew severity of the cucumber leaf images based on the neural network estimation model of downy mildew severity.

The device embodiments described above are only illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed over multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

From the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on this understanding, the above-mentioned technical solutions can be embodied in the form of software products in essence or the parts that make contributions to the prior art, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic A disc, an optical disc, etc., includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be The technical solutions described in the foregoing embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (9)

1. a method for assessing the severity of cucumber downy mildew, is characterized in that, comprises: Obtaining images of cucumber leaves to be evaluated collected in the natural environment; Using the stochastic gradient descent method and based on the neural network estimation model of downy mildew severity, the cucumber downy mildew severity of the cucumber leaf image was evaluated. 2. The method according to claim 1, wherein the neural network estimation model based on downy mildew severity, before evaluating the cucumber downy mildew severity of the cucumber leaf image, further comprises: The sample images of cucumber downy mildew leaves collected in the natural environment were preprocessed, and the original data set was constructed based on the preprocessed sample images; Based on the sample images in the original data set and the disease degree values of the sample images, the original neural network model is trained to obtain the neural network estimation model of the downy mildew severity. 3. The method according to claim 2, wherein the sample images of the cucumber downy mildew leaves collected under the natural environment are preprocessed, comprising: Eliminate sample images whose resolution is lower than the preset threshold from the collected sample images; The background pattern of each sample image is removed, and each sample image is adjusted to a preset size. 4. The method according to claim 2, wherein, before the training of the original neural network model based on the original data set to obtain the neural network estimation model for the severity of downy mildew, the method further comprises: According to a preset processing method, the sample images in the original data set are processed to expand the original data set, and the preset processing method includes at least any one of the following three methods, the following three methods The modes are color dithering, horizontal flipping and vertical flipping. 5. The method according to claim 1, wherein the neural network estimation model for the severity of downy mildew comprises an input layer, 5 convolution layers, 4 pooling layers, 4 batch normalization layers, 2 A fully connected layer and an output layer. 6. The method of claim 5, wherein the input layer is connected to a first convolutional layer and a first batch normalization layer, the first convolutional layer and the first batch A normalization layer is connected to a first pooling layer, which is connected to a second convolutional layer and a second batch normalization layer, the second convolutional layer and the second batch The normalization layer is connected to the second pooling layer, which is connected to the third convolutional layer and the third batch normalization layer, the third convolutional layer and the third batch The normalization layer is connected to the third pooling layer, which is connected to the fourth convolutional layer and the fourth batch normalization layer, which is connected to the fourth convolutional layer and the fourth batch normalization layer. The normalization layer is connected to the fourth pooling layer, the fourth pooling layer is connected to the fifth convolutional layer, and the fifth convolutional layer is sequentially connected to the two fully connected layers and the output layer connect. 7. an assessment device for the severity of downy mildew of cucumber, is characterized in that, comprises: an acquisition module, used to acquire the cucumber leaf images to be evaluated collected in the natural environment; The evaluation module is configured to use the stochastic gradient descent method and based on the neural network estimation model of downy mildew severity to evaluate the cucumber downy mildew severity of the cucumber leaf image. 8. An electronic device, characterized in that, comprising: at least one processor; and at least one memory communicatively coupled to the processor, wherein: The memory stores program instructions executable by the processor, and the processor invokes the program instructions to be able to perform the method as claimed in any one of claims 1 to 6 . 9 . A non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, the computer instructions cause the computer to execute any one of claims 1 to 6 . Methods.