Drought grade discrimination method based on crop curling degree

Abstract

The invention provides a crop curl degree-based drought grade discrimination method, which comprises the following steps: acquiring images of different drought grades of crops in different growth stages, preprocessing the images, and training a Mask R-CNN model as input to obtain a leaf curl degree-based drought grade discrimination model of the crops in different growth stages. The crop image can be collected in real time and preprocessed, the drought grade of the crop can be discriminated by using the drought grade discrimination model of the crop in different growth stages based on the leaf curl degree to obtain the drought grade of the crop, the change of the drought grade of the crop can be timely and accurately discriminated in a large range, the rescue opportunity is prevented from being missed, and the survival rate of the crop is improved. And the situation that the crop growth is severely influenced is avoided.

Description

Technical Field

[0001] The present application belongs to the technical field of crop drought level determination, and specifically relates to a drought level determination method based on the degree of crop curling. Background Art

[0002] Crop crops often suffer from varying degrees of drought at various growth stages, such as the jointing stage, the tasseling and silking stage, and the milky and mature stages. The drought levels are usually divided into no drought, light drought, moderate drought, and severe drought. Accurately identifying the drought level at each growth stage of crops and taking corresponding rescue measures in a timely manner are of great significance for ensuring the normal growth of crops.

[0003] At present, most people rely on manual inspections to judge the drought level at various growth stages of crops. This method is extremely dependent on manual experience, consumes a lot of manpower, and is prone to omissions. It is difficult to ensure that changes in drought levels at various growth stages of crops can be judged in a timely manner over a large area, which may lead to missed rescue opportunities and serious impacts on crop growth.

[0004] This application is proposed in view of the above-mentioned technical defects.

[0005] It should be noted that the disclosure of the above background technology content is only used to assist in understanding the inventive concept and technical solution of the present invention, and it does not necessarily belong to the prior art of this patent application. In the absence of clear evidence that the above content has been disclosed on the filing date of this application, the above background technology should not be used to evaluate the novelty and creativity of the present application. Summary of the invention

[0006] The purpose of the present application is to provide a method for determining drought levels based on the curling degree of crops, so as to overcome or alleviate at least one of the known technical deficiencies.

[0007] The technical solution of this application is:

[0008] A method for determining drought levels based on crop curling degree, comprising:

[0009] Step 1: Collect images of crops at different growth stages experiencing different drought levels;

[0010] Step 2: Preprocess the images of crops at different growth stages suffering from different drought levels;

[0011] Step 3: Using images of crops at different growth stages experiencing different drought levels as input, the Mask R-CNN model is trained to obtain a drought level discrimination model for crops at different growth stages based on leaf curling.

[0012] Step 4: Collect crop images in real time, perform preprocessing, and use a drought level discrimination model for crops at different growth stages based on leaf curling to discriminate drought levels and obtain the drought level of the crops.

[0013] Optionally, in the above-mentioned method for determining drought level based on crop curling degree, step 2 is specifically as follows:

[0014] The images of crops at different growth stages suffering from different drought levels were compressed into a size of 1920×1280, and the curled leaves of the crops were annotated using the image annotation tool LabelMe software. After each image was annotated, a JSON format file was generated, which contained annotation information such as the image's location coordinates, category label, and image size. The open source computer vision library OpenCV was used to annotate the annotated images and draw the coordinate image of each curled leaf.

[0015] Optionally, in the above-mentioned method for distinguishing drought levels based on the degree of crop curling, in step three, 80% of the images of crops suffering from different drought levels at different growth stages are divided into a training set and 20% are divided into a test set to train the Mask R-CNN model.

[0016] Optionally, in the above-mentioned drought level discrimination method based on crop curling degree, in the Mask R-CNN model, a convolution kernel is used to automatically extract image features, and the convolution kernel slides on the image to generate a feature map.

[0017] Optionally, in the above-mentioned drought level discrimination method based on crop curling degree, in the Mask R-CNN model, RoI Align uses a bilinear interpolation method to achieve pixel-to-pixel alignment, instead of RoIPooling to perform pooling operations on feature maps.

[0018] Optionally, in the above-mentioned drought level discrimination method based on the degree of crop curling, in the Mask R-CNN model, the drought level is graded by calculating the curling confidence of the curled leaves, the maximum, minimum, average, variance and standard deviation indicators of the border coordinates and the length and width of the rectangular border to obtain the drought level of the crop.

[0019] Optionally, in the above-mentioned drought level discrimination method based on crop curling degree, the mean average precision MAP of model detection, the precision Precision of the model, the recall rate Recall of the model, and the mean average precision F1 are used. score , perform model evaluation and parameter tuning.

[0020] Optionally, in the above-mentioned method for determining drought level based on crop curling degree:

[0021]

[0022]

[0023]

[0024]

[0025] Among them, TP is a positive example with correct classification; FP is a positive example with incorrect classification; FN and TN are negative examples with incorrect classification and positive examples with correct classification, respectively; N is the number of image samples.

[0026] This application has at least the following beneficial technical effects:

[0027] A drought level discrimination method based on crop curling degree is provided. By collecting images of crops suffering from different drought levels at different growth stages, preprocessing them, and using them as input to train a Mask R-CNN model, a drought level discrimination model for crops at different growth stages based on leaf curling degree is obtained. Then, crop images can be collected in real time, preprocessed, and drought level discrimination model for crops at different growth stages based on leaf curling degree is used to discriminate drought levels, and the drought level of crops can be obtained. The change of crop drought level can be timely and accurately judged over a large range, so as to prevent the loss of rescue opportunities and avoid situations that seriously affect crop growth. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Schematic diagram of a method for distinguishing drought levels based on crop curling degree provided in an embodiment of the present application;

[0029] Figure 2 Schematic diagram of the Mask R-CNN model structure provided in the embodiment of the present application;

[0030] Figure 3 This is a schematic diagram of an embodiment of the present application, which uses images of crops suffering from different drought levels at different growth stages as input, trains the Mask R-CNN model, and obtains a drought level discrimination model for crops at different growth stages based on leaf curling.

[0031] In order to better illustrate this embodiment, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product. In addition, the drawings are only used for illustrative purposes and cannot be understood as limitations on this patent. DETAILED DESCRIPTION

[0032] In order to make the technical solution and advantages of the present application clearer, the technical solution of the present application will be further described in detail in detail and in detail in conjunction with the accompanying drawings. It can be understood that the specific embodiments described here are only partial embodiments of the present application, which are only used to explain the present application, not to limit the present application. It should be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, and other related parts can refer to the general design. In the absence of conflict, the embodiments in the present application and the technical features in the embodiments can be combined with each other to obtain new embodiments.

[0033] In addition, unless otherwise defined, the technical terms or scientific terms used in the description of this application shall be the common meanings understood by the general technicians in the field to which this application belongs. The words "upper", "lower", "left", "right", "center", "vertical", "horizontal", "inside", "outside" and other words indicating orientation used in the description of this application are only used to indicate the relative direction or positional relationship, and do not imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. When the absolute position of the described object changes, its relative positional relationship may also change accordingly, so it cannot be understood as a limitation on this application. The "first", "second", "third" and similar terms used in the description of this application are only used for descriptive purposes to distinguish different components, and cannot be understood as indicating or implying relative importance. The similar words "one", "one" or "the" used in the description of this application should not be understood as an absolute limitation on quantity, but should be understood as the existence of at least one. The similar words "including" or "comprising" used in the description of this application mean that the elements or objects appearing in front of the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0034] In addition, it should be noted that, unless otherwise clearly specified and limited, the words "installed", "connected", "connected" and similar terms used in the description of this application should be understood in a broad sense. For example, the connection can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, an indirect connection through an intermediate medium, or the internal connection of two components. Technical personnel in the field can understand their specific meanings in this application according to the specific circumstances.

[0035] The impact of different drought levels on crops is usually directly reflected in the curling degree of crop leaves. Based on this, the embodiment of the present application provides a method for distinguishing drought levels based on the curling degree of crops, such as Figure 1 shown.

[0036] Step 1: Collect images of crops at different growth stages experiencing different drought levels.

[0037] Freely rotating cameras or drones can be used to collect images of crops during the jointing stage, tasseling and silking stage, and milky and mature stages, and when the crops are experiencing no drought, light drought, moderate drought, or severe drought.

[0038] Step 2: Preprocess the images of crops at different growth stages experiencing different drought levels.

[0039] The images of crops at different growth stages suffering from different drought levels were compressed into a size of 1920×1280, and the curled leaves of the crops were annotated using the image annotation tool LabelMe software. After each image was annotated, a JSON format file was generated, which contained annotation information such as the image's location coordinates, category label, and image size. The open source computer vision library OpenCV was used to annotate the annotated images and draw the coordinate image of each curled leaf.

[0040] Step 3: Using images of crops at different growth stages experiencing different drought levels as input, the Mask R-CNN model is trained to obtain a drought level discrimination model for crops at different growth stages based on leaf curling, such as Figure 3 shown.

[0041] The Mask R-CNN model was trained by dividing 80% of the images of crops at different growth stages suffering from different drought levels into training sets and 20% into test sets.

[0042] The Mask R-CNN model structure is as follows: Figure 2 shown.

[0043] The Mask R-CNN model uses convolution kernels to automatically extract image features. The convolution kernels slide on the image to generate feature maps to reduce the spatial dimension of the input data.

[0044] In the Mask R-CNN model, RoI Align uses bilinear interpolation to achieve pixel-to-pixel alignment, instead of RoIPooling to perform pooling operations on feature maps. During instance segmentation, the feature maps can be better mapped to the original image.

[0045] In the Mask R-CNN model, the input includes the number of curled leaves, curl confidence and segmentation mask (border length and width) of the image. By calculating the curl confidence of the curled leaves, the maximum, minimum, mean, variance and standard deviation of the border coordinates and the length and width of the rectangular border, the drought level is graded according to the curl information of the leaves to obtain the drought level of the crop.

[0046] In the Mask R-CNN model, the mean average precision MAP, the model's precision Precision, the model's recall rate Recall, and the mean average precision F1 are used for model detectionscore , perform model evaluation and parameter tuning.

[0047]

[0048]

[0049]

[0050]

[0051] Among them, TP is a positive example with correct classification; FP is a positive example with incorrect classification; FN and TN are respectively a negative example with incorrect classification and a negative example with correct classification; N is the number of image samples;

[0052] Precision is the accuracy of the model; Recall is the recall rate of the model; F1 score It is the harmonic mean of the model accuracy and recall rate; MAP is the mean average precision of the model detection.

[0053] Step 4: Collect crop images in real time, perform preprocessing, and use a drought level discrimination model for crops at different growth stages based on leaf curling to discriminate drought levels and obtain the drought level of the crops.

[0054] Crop images can be collected every two hours using a freely rotating camera or drone.

[0055] The drought level discrimination method based on the curling degree of crops disclosed in the above-mentioned embodiment collects images of crops suffering from different drought levels at different growth stages, performs preprocessing, and uses the images as input to train the Mask R-CNN model to obtain a drought level discrimination model for crops at different growth stages based on the degree of leaf curling. Then, crop images can be collected in real time, preprocessed, and the drought level discrimination model for crops at different growth stages based on the degree of leaf curling can be used to discriminate the drought level, and the drought level of the crop can be obtained. This method can timely and accurately discriminate the changes in the drought level of crops over a large range, prevent the loss of rescue opportunities, and avoid situations that seriously affect crop growth.

[0056] So far, the technical solution of the present application has been described in conjunction with the preferred embodiments shown in the accompanying drawings. Those skilled in the art should understand that the scope of protection of the present application is obviously not limited to these specific embodiments. Without departing from the principles of the present application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the scope of protection of the present application.

Claims

1. A method for determining drought levels based on crop curling, characterized in that: include: Step 1: Collect images of crops at different growth stages experiencing different drought levels; Step 2: Preprocess the images of crops at different growth stages suffering from different drought levels; Step 3: Using images of crops at different growth stages experiencing different drought levels as input, the Mask R-CNN model is trained to obtain a drought level discrimination model for crops at different growth stages based on leaf curling. Step 4: Collect crop images in real time, perform preprocessing, and use a drought level discrimination model for crops at different growth stages based on leaf curling to discriminate drought levels and obtain the drought level of the crops.

2. The method for determining drought level based on crop curling degree according to claim 1, characterized in that: In step 2, specifically: The images of crops at different growth stages suffering from different drought levels were compressed into a size of 1920×1280, and the curled leaves of the crops were annotated using the image annotation tool LabelMe software. After each image was annotated, a JSON format file was generated, which contained annotation information such as the image's location coordinates, category label, and image size. The open source computer vision library OpenCV was used to annotate the annotated images and draw the coordinate image of each curled leaf.

3. The method for determining drought level based on crop curling degree according to claim 1, characterized in that: In step 3, 80% of the images of crops at different growth stages suffering from different drought levels are divided into training sets and 20% are divided into test sets to train the Mask R-CNN model.

4. The method for determining drought level based on crop curling degree according to claim 1, characterized in that: In the Mask R-CNN model, the convolution kernel is used to automatically extract the features of the image, and the convolution kernel slides on the image to generate a feature map.

5. The method for determining drought level based on crop curling degree according to claim 1, characterized in that: In the Mask R-CNN model, RoIAlign uses bilinear interpolation to achieve pixel-to-pixel alignment instead of RoIPooling to perform pooling operations on feature maps.

6. The method for determining drought level based on crop curling degree according to claim 1, characterized in that: In the Mask R-CNN model, the drought level is graded by calculating the curling confidence of the curled leaves, the maximum, minimum, mean, variance and standard deviation indicators of the border coordinates and the length and width of the rectangular border to obtain the drought level of the crop.

7. The method for determining drought level based on crop curling degree according to claim 1, characterized in that: Use the mean average precision MAP, model precision Precision, model recall Recall, and mean average precision F1 of the model detection score , perform model evaluation and parameter tuning.

8. The method for determining drought level based on crop curling degree according to claim 7, characterized in that: Among them, TP is a positive example with correct classification; FP is a positive example with incorrect classification; FN and TN are negative examples with incorrect classification and positive examples with correct classification, respectively; N is the number of image samples.

Images