Rapid and accurate lossless measurement method and system for body size index and color of earthworm

Abstract

The invention discloses a rapid and accurate nondestructive measurement method and system for body size indexes and colors of earthworms, and belongs to the technical field of biological growth monitoring. The method comprises the following steps: S1, acquiring real-time earthworm form data, and carrying out object labeling on the form data to obtain labeling box data containing earthworms; s2, performing contour extraction on the annotation frame data to obtain earthworm contour data in the annotation frame; and S3, calculating an earthworm body size index and a color label based on the earthworm contour data. The body size index comprises the length and the diameter of the earthworm. According to the invention, based on a machine vision and learning algorithm, image acquisition is carried out on earthworms, image annotation and contour extraction are carried out on the acquired images, and single earthworm contours are obtained. And carrying out length calculation on the earthworm contour based on the scale layer to obtain an earthworm body size index.

Description

Technical Field

[0001] The invention belongs to the technical field of biological growth monitoring, and in particular relates to a method and system for quickly and accurately non-destructively measuring the body size index and color of an earthworm. Background Art

[0002] Earthworms are terrestrial invertebrates of the class Oligochaeta of the phylum Annelida. They have a cylindrical, bilaterally symmetrical, and segmented body. They mainly feed on plant remains, organic debris, soil containing organic matter or fungi, and livestock and poultry feces.

[0003] Earthworms can grow and reproduce at temperatures between 10℃ and 30℃. Earthworms are hermaphrodites, but they need cross-fertilization to lay eggs. Fertilized eggs develop into young earthworms after 18-21 days in the egg cocoon. It takes about 4 months for young earthworms to mature from birth. After maturity, they lay eggs once a month, with 10-12 eggs each time. Good earthworm varieties can reproduce nearly a thousand times a year.

[0004] The body size of earthworms is the main indicator of their breeding performance and is closely related to the amount of food they eat. The quality of earthworm species determines their variety and body size. Accurate and non-destructive measurement of earthworm body size is a prerequisite and basis for obtaining data and conducting earthworm biological breeding. Traditional methods of measuring earthworm body length often require manual measurement of earthworms using a ruler. Since earthworms are invertebrates, they are always moving and cannot remain static. Because it is difficult to measure living earthworms, they are even killed with alcohol before measurement. The above methods have problems such as inaccurate measurement accuracy, labor-intensive, and the inability of earthworms to revive after measurement, which destroys the seed source. Summary of the invention

[0005] The present invention aims to solve the deficiencies of the prior art and proposes a rapid, accurate and non-destructive measurement method and system for the body size index and color of earthworms, which can measure the body size index of living earthworms without direct contact with the earthworms.

[0006] To achieve the above object, the present invention provides the following scheme: a method for rapid, accurate and non-destructive measurement of body size index and color of earthworms, comprising the following steps:

[0007] S1, obtaining real-time earthworm morphological data, and marking the morphological data as an object to obtain marked frame data containing the earthworm;

[0008] S2, extracting the outline of the marked frame data to obtain the earthworm outline data in the marked frame;

[0009] S3. Calculate the earthworm body size index and color label based on the earthworm outline data.

[0010] Further preferably, S1 comprises the following steps:

[0011] S11, obtaining real-time morphological data of earthworms, and preprocessing the morphological data to obtain processed data;

[0012] S12, extracting the minimum edge rectangle of the processed data to obtain the labeled frame data containing the earthworm.

[0013] Further preferably, the pretreatment method in S11 includes:

[0014] The morphological data is subjected to color space conversion, denoising and enhancement processing to obtain the processed image.

[0015] Further preferably, the method for obtaining the earthworm outline data in S2 includes:

[0016] Performing edge detection on the annotation frame data to obtain feature points; and obtaining a contour curve based on the feature points;

[0017] The characteristic points are clustered to obtain a clustering result, and the contour curve is solved based on the clustering result to obtain the earthworm contour data.

[0018] Further preferably, S3 comprises the following steps:

[0019] S31, adding a proportional grid layer to the earthworm contour data to obtain a contour map with grid data;

[0020] S32, performing mathematical calculations on the contour map with grid data to obtain the length and diameter of the earthworm;

[0021] S33, performing color recognition on a single earthworm in the earthworm outline data to obtain the color label.

[0022] The present invention also provides a rapid, accurate and non-destructive measurement system for the body size index and color of earthworms. The measurement system is used to implement the above-mentioned measurement method, and comprises: a data acquisition module, a contour extraction module and a calculation module;

[0023] The data acquisition module is used to acquire real-time earthworm morphological data, perform object annotation on the morphological data, and obtain annotation frame data containing earthworms;

[0024] The contour acquisition module is used to extract the contour of the marked frame data to obtain the earthworm contour data in the marked frame;

[0025] The calculation module is used to calculate the earthworm body size index based on the earthworm outline data.

[0026] Further preferably, the data acquisition module includes: a preprocessing unit and an extraction unit;

[0027] The preprocessing unit is used to obtain real-time morphological data of earthworms, and preprocess the morphological data to obtain processed data;

[0028] The extraction unit is used to perform minimum edge rectangle extraction on the processed data to obtain the marked frame data containing the earthworm.

[0029] Further preferably, the method for preprocessing by the preprocessing unit includes:

[0030] The morphological data is subjected to color space conversion, denoising and enhancement processing to obtain the processed image.

[0031] Further preferably, the method in which the contour acquisition module acquires the earthworm contour data comprises:

[0032] Performing edge detection on the annotation frame data to obtain feature points; and obtaining a contour curve based on the feature points;

[0033] The characteristic points are clustered to obtain a clustering result, and the contour curve is solved based on the clustering result to obtain the earthworm contour data.

[0034] Further preferably, the calculation module includes: a layer adding unit, a calculation unit and a recognition unit;

[0035] The layer adding unit is used to add a proportional grid layer to the earthworm outline data to obtain an outline map with grid data;

[0036] The calculation unit is used to perform mathematical calculations on the contour map with grid data to obtain the length and diameter of the earthworm;

[0037] The recognition unit is used to perform color recognition on a single earthworm in the earthworm outline data to obtain the color label.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The invention discloses a method and system for quickly and accurately measuring the body size index and color of earthworms. Based on machine vision and learning algorithms, the earthworms are imaged, annotated, and contour extracted to obtain the contour of a single earthworm. The length of the earthworm contour is calculated based on a scale layer to obtain the body size index and color label of the earthworm. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0041] Figure 1 A schematic flow chart of a method for rapid, accurate and non-destructive measurement of earthworm body size index and color provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] Embodiment 1:

[0045] like Figure 1 As shown, this embodiment provides a method for rapid, accurate and non-destructive measurement of earthworm body size index and color, comprising the following steps:

[0046] S1. Acquire real-time earthworm morphological data, perform object annotation on the morphological data, and obtain annotation frame data containing the earthworm.

[0047] Specifically, S1 includes the following steps:

[0048] S11, obtaining real-time morphological data of earthworms, preprocessing the morphological data, and obtaining processed data.

[0049] In this embodiment, the preprocessing method includes:

[0050] The morphological data is subjected to color space conversion, denoising and enhancement processing to obtain processed data.

[0051] The morphological data is converted from RGB color space to grayscale space to obtain a grayscale image, which can better highlight the characteristics of earthworms. Then Gaussian noise and salt and pepper noise are added to the grayscale image. As the dimension increases, Gaussian noise and salt and pepper noise gradually fade away. On the basis of eliminating noise, the edge of the grayscale image is maintained. Finally, the histogram transformation method is used to enhance the grayscale image to obtain the processed data.

[0052] S12, extracting the minimum edge rectangle of the processed data to obtain the labeled frame data containing the earthworm.

[0053] In this embodiment, a recognition network is used to extract the minimum edge rectangle of the processed data to obtain the labeled box data containing the earthworm.

[0054] Specifically, the recognition network is based on the YOLOv7 network, and the six conv convolutions in the ELAN block in the Backbone part of the detection network are concatenated with the CBS block. The GAM attention module is introduced after each concat operation in the Head.

[0055] Among them, the GAM attention module defines the input feature map as F1∈R C×H×W , then the expressions of the intermediate state variable F2 and the final output F3 are as follows:

[0056]

[0057] Where M C represents the channel, M S represents the spatial attention map, Represents element-wise computation of multiplication.

[0058] The input feature F1 finally outputs a feature F3 through the channel sub-attention unit and the spatial sub-attention unit. The 3D arrangement is used to save cross-dimensional information by activating the channel attention sub-unit, and a multi-layer MLP (multi-layer perceptron) is used to amplify the cross-dimensional channel spatial correlation. In the spatial attention sub-unit, two convolutions are used to merge spatial information in order to fully focus on spatial information.

[0059] Finally, the SPPCSPC module in the head part is replaced by the SPPFCSPC module. The SPPFCSPC module processes the three maximum pooling layers sequentially, which can improve the speed without changing the receptive field.

[0060] For the 32x downsampled feature map C5 output by Backbone, the number of channels is changed from 1024 to 512 after passing through the SPPFCSPC module. First, it is fused with the C4 and C3 layers in a top-down manner to obtain the P3, P4, and P5 layers; then it is fused with the P4 and P5 layers in a bottom-up manner. Finally, three detection layers of different sizes are obtained: 80×80, 40×40, and 20×20. Among them, 80×80 is used to detect small targets, 40×40 is used to detect small and medium targets, and 20×20 is used to detect medium targets.

[0061] S2. Extract the contour of the marked frame data to obtain the contour data of the earthworm in the marked frame.

[0062] First, edge detection is performed on the labeled box data, and the corner detector is used to search for feature points within the labeled box. The earthworm outline is a curve on a plane, so the earthworm outline curve is reconstructed based on the acquired feature points. The curve equation is:

[0063]

[0064] In the formula, α k is the learning parameter; is the basis function, and a Gaussian function can be used; x represents the vector composed of the pixels and gradients of the feature points.

[0065]

[0066] In the formula, c k Represents the center of each category in the feature point classification; σ k represents the within-class variance.

[0067] Methods for clustering feature points include:

[0068] For the feature point set T, a nearest neighbor list B1 is established, where the nearest neighbor means that in the clustering process, the feature point (vector) x1 has the shortest distance with the feature point (vector) x2 in the entire space. At the same time, another list B2 is established, in which other feature points that do not exist in B1 are placed.

[0069] Then randomly select a vector x1 from B2 and put it into B1, then search for the nearest neighbor vector x2 of x1 from B2 and put it into B1, and then put the nearest neighbor vector x3 of vector x2 into B1, until there are l vectors added to B1. At this time, B1 is represented as {x1, x2, x3,..., x l-1 , x l}.

[0070] Judge x l-1 With x l When the similarity value is less than the set threshold, it means that the clustering result is not ideal. Each vector in B1 is taken as a class and deleted from the feature point set T. When the similarity value is greater than the set threshold, the vector x l-1 With x l The vectors in B1 are grouped into one category and deleted from B1. They are put into B2. The vectors in B1 are searched again to determine x. l-2 Nearest neighbors in B2, until B1 is empty.

[0071] Randomly select a new vector from B2 and put it into B1. Repeat the above steps until the feature point set T is empty and the clustering is completed. Cluster the feature points and set F(x) = 1 to obtain the learning parameters. *Perform clustering, calculate F(x*), mark all pixel points with a value of 1, and approximate the contour curve data to obtain the contour data of the earthworm.

[0072] S3. Calculate the earthworm body size index based on the earthworm outline data.

[0073] A grid layer of a preset ratio is added to the obtained earthworm contour data to obtain a contour map with grid data. At the same time, pixel points of the earthworm contour data obtained above are selected to obtain a number of calculation points, and the length and diameter of the earthworm are calculated based on the preset ratio and the coordinates of the calculation points.

[0074] Afterwards, color recognition is performed on individual earthworms in the earthworm outline data to obtain color labels.

[0075] Specifically, first, for all pixels, the main colors are obtained by statistically analyzing the color histogram. Then, the obtained main colors are reduced in dimension to obtain color features that can reflect the whole. Then, a semi-supervised k-means clustering algorithm is used to determine the similarity between the color features after dimensionality reduction and the standard colorimetric card. The present embodiment uses the Euclidean distance. According to the size of the Euclidean distance, it is clustered into k categories. The color features of the standard colorimetric card are set as the initial cluster center, and the similarity between the color features of the earthworm outline data and the initial cluster center is calculated, and it is divided into clusters of the most similar cluster centers to obtain earthworm color labels.

[0076] In order to demonstrate the accuracy of the measurement method proposed in the present invention compared with the manual measurement method, several earthworms were quickly measured under natural elongation conditions, and then cultured until the influence of the measurement on the earthworms was eliminated to obtain their body length data. After that, the method proposed in the present invention and the manual measurement method were used to perform non-destructive testing (i.e., during the manual measurement process, the earthworms were not touched to affect them) to obtain measurement data; some of the measurement data are shown in Table 1.

[0077] Table 1

[0078]

[0079] It can be seen that for earthworms with a natural body length of 7 cm, due to the large changes in earthworm morphology during the non-destructive testing process in the manual measurement method, such as curling up, wriggling, etc., it is impossible to accurately measure, resulting in a large difference between the measured value of the earthworm and the natural body length data. However, the method of the present invention is used to extract the external contour of the earthworm and then calculate the body length, which is less affected by the earthworm morphology and can greatly reduce the error under non-destructive conditions. Through a large number of multiple measurements, it can be closer to the true value.

[0080] Embodiment 2:

[0081] This embodiment provides an application of the method for quickly and accurately measuring body size index proposed in the present invention - earthworm seed production.

[0082] By setting the same living environment and feeding conditions, a seed production experiment is conducted on different experimental groups, for example, 3 groups. An even number of earthworms with relatively good length and diameter are selected from the first generation as first generation seeds for reproduction to obtain the second generation of earthworms.

[0083] The second generation earthworms are also selected by the above method to reproduce the second generation seeds. Similarly, the third generation seeds are obtained. The above method can reproduce earthworm individuals with relatively high quality in length and diameter within the same period of time.

[0084] Embodiment three:

[0085] This embodiment provides a rapid, accurate and non-destructive measurement system for earthworm body size index and color, including: a data acquisition module, a contour extraction module and a calculation module.

[0086] The data acquisition module is used to acquire real-time earthworm morphological data, perform object annotation on the morphological data, and obtain annotation frame data containing earthworms.

[0087] Specifically, the data acquisition module includes: a preprocessing unit and an extraction unit.

[0088] The preprocessing unit is used to obtain the real-time morphological data of the earthworm, preprocess the morphological data, and obtain processed data.

[0089] In this embodiment, the preprocessing method includes:

[0090] The morphological data is subjected to color space conversion, denoising and enhancement processing to obtain processed data.

[0091] The morphological data is converted from RGB color space to grayscale space to obtain a grayscale image, which can better highlight the characteristics of earthworms. Then Gaussian noise and salt and pepper noise are added to the grayscale image. As the dimension increases, Gaussian noise and salt and pepper noise gradually fade away. On the basis of eliminating noise, the edge of the grayscale image is maintained. Finally, the histogram transformation method is used to enhance the grayscale image to obtain the processed data.

[0092] The extraction unit is used to extract the minimum edge rectangle of the processed data to obtain the labeled frame data containing the earthworm.

[0093] In this embodiment, a recognition network is used to extract the minimum edge rectangle of the processed data to obtain the labeled box data containing the earthworm.

[0094] Specifically, the recognition network is based on the YOLOv7 network, and the six conv convolutions in the ELAN block in the Backbone part of the detection network are concatenated with the CBS block. The GAM attention module is introduced after each concat operation in the Head.

[0095] Among them, the GAM attention module defines the input feature map as F1∈R C×H×W , then the expressions of the intermediate state variable F2 and the final output F3 are as follows:

[0096]

[0097] Where M C represents the channel, M S represents the spatial attention map, Represents element-wise computation of multiplication.

[0098] The input feature F1 finally outputs a feature F3 through the channel sub-attention unit and the spatial sub-attention unit. The 3D arrangement is used to save cross-dimensional information by activating the channel attention sub-unit, and a multi-layer MLP (multi-layer perceptron) is used to amplify the cross-dimensional channel spatial correlation. In the spatial attention sub-unit, two convolutions are used to merge spatial information in order to fully focus on spatial information.

[0099] Finally, the SPPCSPC module in the head part is replaced by the SPPFCSPC module. The SPPFCSPC module processes the three maximum pooling layers sequentially, which can improve the speed without changing the receptive field.

[0100] For the 32x downsampled feature map C5 output by Backbone, the number of channels is changed from 1024 to 512 after passing through the SPPFCSPC module. First, it is fused with the C4 and C3 layers in a top-down manner to obtain the P3, P4, and P5 layers; then it is fused with the P4 and P5 layers in a bottom-up manner. Finally, three detection layers of different sizes are obtained: 80×80, 40×40, and 20×20. Among them, 80×80 is used to detect small targets, 40×40 is used to detect small and medium targets, and 20×20 is used to detect medium targets.

[0101] The contour extraction module is used to extract the contour of the marked frame data to obtain the earthworm contour data in the marked frame.

[0102] First, edge detection is performed on the labeled box data, and the corner detector is used to search for feature points within the labeled box. The earthworm outline is a curve on a plane, so the earthworm outline curve is reconstructed based on the acquired feature points. The curve equation is:

[0103]

[0104] In the formula, α k is the learning parameter; is the basis function, and a Gaussian function can be used; x represents the vector composed of the pixels and gradients of the feature points.

[0105]

[0106] In the formula, c k Represents the center of each category in the feature point classification; σ k represents the within-class variance.

[0107] Methods for clustering feature points include:

[0108] For the feature point set T, a nearest neighbor list B1 is established, where the nearest neighbor means that in the clustering process, the feature point (vector) x1 has the shortest distance with the feature point (vector) x2 in the entire space. At the same time, another list B2 is established, in which other feature points that do not exist in B1 are placed.

[0109] Then randomly select a vector x1 from B2 and put it into B1, then search for the nearest neighbor vector x2 of x1 from B2 and put it into B1, and then put the nearest neighbor vector x3 of vector x2 into B1, until there are l vectors added to B1. At this time, B1 is represented as {x1, x2, x3,..., x l-1 , x l}.

[0110] Judge x l-1 With x l When the similarity value is less than the set threshold, it means that the clustering result is not ideal. Each vector in B1 is taken as a class and deleted from the feature point set T. When the similarity value is greater than the set threshold, the vector x l-1 With x l The vectors in B1 are grouped into one category and deleted from B1. They are put into B2. The vectors in B1 are searched again to determine x. l-2 Nearest neighbors in B2, until B1 is empty.

[0111] Randomly select a new vector from B2 and put it into B1. Repeat the above steps until the feature point set T is empty and the clustering is completed. Cluster the feature points and set F(x) = 1 to obtain the learning parameters. * Perform clustering, calculate F(x*), mark all pixel points with a value of 1, and approximate the contour curve data to obtain the contour data of the earthworm.

[0112] The calculation module is used to calculate the earthworm body size index based on the earthworm outline data. The calculation module includes a layer adding unit, a calculation unit and a recognition unit.

[0113] The layer adding unit adds a grid layer of a preset ratio to the obtained earthworm outline data to obtain an outline map with grid data; the calculation unit selects pixel points of the earthworm outline data obtained above to obtain a number of calculation points, and calculates the length and diameter of the earthworm based on the preset ratio and the coordinates of the calculation points. Afterwards, the recognition unit performs color recognition on a single earthworm in the earthworm outline data to obtain a color label.

[0114] Specifically, first, for all pixels, the main colors are obtained by statistically analyzing the color histogram. Then, the obtained main colors are reduced in dimension to obtain color features that can reflect the whole. Then, a semi-supervised k-means clustering algorithm is used to determine the similarity between the color features after dimensionality reduction and the standard colorimetric card. The present embodiment uses the Euclidean distance. According to the size of the Euclidean distance, it is clustered into k categories. The color features of the standard colorimetric card are set as the initial cluster center, and the similarity between the color features of the earthworm outline data and the initial cluster center is calculated, and it is divided into clusters of the most similar cluster centers to obtain earthworm color labels.

[0115] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A rapid, accurate and non-destructive method for measuring the body size index and color of earthworms, characterized in that: The following steps are involved: S1, obtaining real-time earthworm morphological data, and marking the morphological data as an object to obtain marked frame data containing the earthworm; S2, extracting the outline of the marked frame data to obtain the earthworm outline data in the marked frame; S3. Calculate the earthworm body size index and color label based on the earthworm outline data.

2. A rapid, accurate and non-destructive measurement method for earthworm body size index and color according to claim 1, characterized in that: S1 includes the following steps: S11, obtaining real-time morphological data of earthworms, and preprocessing the morphological data to obtain processed data; S12, extracting the minimum edge rectangle of the processed data to obtain the labeled frame data containing the earthworm.

3. A rapid, accurate and non-destructive measurement method for earthworm body size index and color according to claim 2, characterized in that: The pre-processing method in S11 includes: The morphological data is subjected to color space conversion, denoising and enhancement processing to obtain the processed image.

4. A rapid, accurate and non-destructive measurement method for earthworm body size index and color according to claim 1, characterized in that: The method for obtaining the earthworm outline data in S2 includes: Performing edge detection on the annotation frame data to obtain feature points; and obtaining a contour curve based on the feature points; The characteristic points are clustered to obtain a clustering result, and the contour curve is solved based on the clustering result to obtain the earthworm contour data.

5. The method for rapid, accurate and non-destructive measurement of earthworm body size index and color according to claim 1, characterized in that: S3 includes the following steps: S31, adding a proportional grid layer to the earthworm contour data to obtain a contour map with grid data; S32, performing mathematical calculations on the contour map with grid data to obtain the length and diameter of the earthworm; S33, performing color recognition on a single earthworm in the earthworm outline data to obtain the color label.

6. A fast, accurate and non-destructive measurement system for earthworm body size index and color, the measurement system being used to implement the measurement method according to any one of claims 1 to 5, characterized in that: include: Data acquisition module, contour extraction module and calculation module; The data acquisition module is used to acquire real-time earthworm morphological data, perform object annotation on the morphological data, and obtain annotation frame data containing earthworms; The contour acquisition module is used to extract the contour of the marked frame data to obtain the earthworm contour data in the marked frame; The calculation module is used to calculate the earthworm body size index and color label based on the earthworm outline data.

7. A rapid, accurate and non-destructive measurement system for earthworm body size index and color according to claim 6, characterized in that: The data acquisition module includes: a preprocessing unit and an extraction unit; The preprocessing unit is used to obtain real-time morphological data of earthworms, and preprocess the morphological data to obtain processed data; The extraction unit is used to perform minimum edge rectangle extraction on the processed data to obtain the marked frame data containing the earthworm.

8. A rapid, accurate and non-destructive measurement system for earthworm body size index and color according to claim 7, characterized in that: The method for the preprocessing unit to perform preprocessing includes: The morphological data is subjected to color space conversion, denoising and enhancement processing to obtain the processed image.

9. A rapid, accurate and non-destructive measurement system for earthworm body size index and color according to claim 6, characterized in that: The method for the contour acquisition module to acquire the earthworm contour data includes: Performing edge detection on the annotation frame data to obtain feature points; and obtaining a contour curve based on the feature points; The characteristic points are clustered to obtain a clustering result, and the contour curve is solved based on the clustering result to obtain the earthworm contour data.

10. A rapid, accurate and non-destructive measurement system for earthworm body size index and color according to claim 6, characterized in that: The calculation module includes: a layer adding unit, a calculation unit and a recognition unit; The layer adding unit is used to add a proportional grid layer to the earthworm outline data to obtain an outline map with grid data; The calculation unit is used to perform mathematical calculations on the contour map with grid data to obtain the length and diameter of the earthworm; The recognition unit is used to perform color recognition on a single earthworm in the earthworm outline data to obtain the color label.

Images