阅读说明：本技术 一种小麦红蜘蛛测报样本采集统计系统和方法 (Wheat red spider forecast sample collection statistical system and method ) 是由 李霞 沈晓强 吴临平 李皓 李卫伟 宋毅飞 赵丽华 李培谦 李静 张泂然 于 2021-08-03 设计创作，主要内容包括：本发明公开了一种小麦红蜘蛛测报样本采集统计系统和方法,包括样本采集装置和样本统计装置；样本采集装置包括采样箱体、箱盖组件、振动控制器；样本统计装置包括智能手机客户端和服务器端,智能手机客户端用于对样本采集装置内的红蜘蛛进行拍摄,并将采集的图像传输到具有图像识别、统计功能的服务器端,在服务器端进行图像预处理、图像识别和分类统计；采用新型小麦红蜘蛛测报样本采集统计系统,可实现麦田小麦红蜘蛛样本的快速采集、样本留取和准确计数,省时省力,提高样本采集和统计效率。(The invention discloses a system and a method for collecting and counting wheat red spider test and report samples, wherein the system comprises a sample collecting device and a sample counting device; the sample acquisition device comprises a sampling box body, a box cover assembly and a vibration controller; the sample statistical device comprises a smart phone client and a server, wherein the smart phone client is used for shooting the red spiders in the sample acquisition device, transmitting the acquired images to the server with image recognition and statistical functions, and performing image preprocessing, image recognition and classified statistics on the server; adopt novel wheat red spider to survey and report sample collection statistical system, can realize quick collection, the sample of wheat red spider sample in wheat field and keep somewhere and get and accurate count, labour saving and time saving improves sample collection and statistical efficiency.)

1. A wheat red spider test and report sample collecting and counting system is characterized by comprising a sample collecting device and a sample counting device; the sample collecting device comprises a sampling box body (1), a box cover assembly (2) and a vibration controller (3); the sample statistical device comprises a smart phone client and a server, wherein the smart phone client is used for shooting the red spiders in the sample acquisition device, transmitting the acquired images to the server with image recognition and statistical functions, and performing image preprocessing, image recognition and classified statistics on the server;

the sampling box body (1) comprises a box body (11), a rotating shaft seat (12) and a box cover rotating shaft inserting guide groove (13); the box body (11) is a cuboid box body lacking a top surface and one side surface adjacent to the top surface, the lacking one side surface is an open end of the sampling box body (1), the open end is aligned with the roots of the wheat seedlings when in use, the lacking top surface is used for accommodating a box cover (21) of a box cover component (2), and the lacking top surface is closed by the box cover (21) when in use; two side surfaces of two sides of the open end are provided with rotating shaft seats (12) near the top, the rotating shaft seats (12) are round holes, the diameter of each round hole is slightly larger than the outer diameter of the circle-cutting rotating shaft (22), and the circle-cutting rotating shaft (22) can freely rotate in the round hole; the box cover rotating shaft inserting guide groove (13) is a rectangular opening, the rectangular opening is communicated with the rotating shaft seat (12), the tangent circle rotating shaft (22) is inserted into the rotating shaft seat (12) through the rectangular opening, and the size of the rectangular opening is smaller than the outer diameter of the tangent circle rotating shaft (22).

2. The wheat red spider forecast sample collecting and counting system according to claim 1, characterized in that the box cover component (2) comprises a box cover (21), a tangential rotating shaft (22), a battery and controller box (23) integrated with the box cover, beating plate guide posts (24), guide post limit check rings (25) and beating plates (26), four beating plate guide posts (24) are fixed on the lower surface of the box cover (21), a vibrating element (28) is also fixed on the lower surface of the box cover (21) and is positioned at the central positions of the four beating plate guide posts (24), the beating plates (26) are fixed on the vibrating element (28), beating plate ribs (27) are arranged on the lower surface of the beating plates (26), and four round holes are formed in four corners of the beating plates (26) and penetrate through the four beating plate guide posts (24) and can freely vibrate on the four beating plate guide posts (24); guide post limit retainer rings (25) are fixed at the bottom ends of the four beating plate guide posts (24) and used for limiting the beating plates (26).

3. The wheat red spider forecast sample collection and statistics system according to claim 1, wherein a battery and controller box (23) is fixed with one side of a box cover (21) and a clapper board (26), a tangential rotating shaft (22) is arranged on the upper portion of the battery and controller box (23), and after the tangential rotating shaft (22) is inserted into a rotating shaft seat (12), when the box cover (21) and the clapper board (26) rotate to be parallel to the bottom surface of the box body (11) or perpendicular to the bottom surface of the box body (11), the battery and controller box (23) can seal the open end of the sampling box body (1).

4. The wheat red spider forecast sample collecting and counting system according to claim 1, wherein the cross section of the battery and controller box (23) is in a sector shape of 90 degrees, the open end is well sealed by using a peripheral arc, and the wheat plant is not damaged by the arc of the battery and controller box (23) in the rotating process.

5. A method for collecting samples according to any one of claims 1 to 4, comprising the steps of:

a1, rightly arranging the open end of the sampling box body (1) at the root of the wheat seedling (6) to be sampled;

a2, vertically placing the box cover assembly (2) with the vibrator, and then pushing the tangent circle rotating shaft (22) on the side surface of the box cover into the bottom of the sampling box body (1) along the guide groove (13) to enable the rotating shaft to reach the rotating shaft seat (12);

a3, closing the box cover assembly (2), wherein in the process of installing the box cover assembly (2), a battery box shell of the box cover assembly (2) pushes upright wheat seedlings to be pressed against the edge of the bottom of the sampling box (1), in the process of closing the box cover, the arc-shaped battery box shell ensures that the wheat seedlings (6) are approximately horizontal when the box cover is closed, and after the box cover is closed, a beating plate and beating plate ribs (27) at the lower part of the box cover assembly are in contact with the wheat seedlings (6);

a4, opening a switch of a vibration controller (3) to connect a vibration element (28), wherein the vibration element (28) drives a beating plate and beating plate ribs to rapidly vibrate up and down along a beating plate guide post (24) to beat wheat;

a5, beating the wheat seedlings by vibration, and enabling the wheat red spiders to fall into the bottom of the sampling box body (1);

a6, after beating, closing the power supply, taking down the box cover assembly (2) and completing sample collection.

6. The method for statistical identification of a sample collection system according to any one of claims 1-4, comprising the steps of:

b1, image acquisition: collecting red spiders in the field, and shooting the red spiders in a collection box by a smart phone client;

b2, image preprocessing and graphic transformation: the server receives an image to be processed transmitted by the client, background removal, graying removal, noise removal and image binarization processing are firstly carried out, then a sliding window is adopted for target positioning, and a square frame is used for marking out the target to be detected;

(1) inputting a picture, and removing the background image by adopting a Lazy snapping method to perform gray processing; carrying out image binarization after Gaussian filtering;

(2) extracting and generating about 2000 candidate regions for the binarized image by using a selective search algorithm;

(3) performing expansion processing on the edge of each candidate region by adding p to 16 pixels, and framing the candidate region after expansion; transform each candidate region size to 227x 227;

b3, image recognition:

adopting a tensoflow frame to construct a parallelized AlexNet convolutional neural network model, inputting each candidate region into an R-CNN convolutional neural network to obtain a 4096-dimensional characteristic vector, judging whether the candidate region is a spider, and leaving a frame of the spider for the next processing;

the R-CNN convolutional neural network model is provided with two input ports, input data are subjected to multilayer processing of convolutional layers conv, activation functions relu, pooling layers pool and batch normalization BN, output data are subjected to cascade merge by adopting convolutional layers conv with parallel structures and then are connected by using full connection layers fc, and forward propagation results logits are obtained;

adopting a transfer learning strategy by the R-CNN convolutional neural network, pre-training on ImageNet, fully learning shallow features by using the convolutional neural network, using an SGD algorithm during training, and taking an initial learning rate of 0.001 and a batch _ size of 128; then removing the last full-connection layer which is pre-trained, modifying the number of categories, carrying out large-scale training on a small-scale data set by a user, and finely adjusting the CNN when the user label data is deficient, so that the capability of identifying the object type is improved;

b4, importing the feature vector derived from the neural network into an SVM classifier to generate a class score;

b5, setting a IoU index threshold (>0.3), calculating a IoU index, removing overlapped frames by adopting non-maximum inhibition and taking the highest-score frame as a basis, and obtaining a candidate frame with the highest score in the class;

b6, classification statistics: the classification is to identify two spiders and then count the number of the spiders;

b7, outputting the result: and transmitting the classified statistical result back to the client through the server communication module, and storing and identifying the statistical result by the client.

Technical Field

The invention belongs to the technical field of wheat red spider forecasting, and particularly relates to a system and a method for collecting and counting wheat red spider forecasting samples directly in a wheat field.

Background

At present, the wheat red spider is measured and reported by adopting a manual investigation mode, so that the method is time-consuming, labor-consuming and inaccurate in counting, and has the following three main defects:

1. the count is inaccurate. At present, the number of red spiders on a wheat is manually investigated, the red spiders on a plant are shaken down on white paper or white plastic cloth or a white porcelain plate paved on the rhizosphere of the wheat to obtain a sample, the red spiders which are disturbed move around rapidly on an object for obtaining the sample, and the red spiders are small in individuals, difficult to count by visual observation and often incapable of counting accurately.

2. The workload is large, and the working efficiency is low. According to the technical specification requirement of wheat red spider prediction of Ministry of agriculture, the wheat is surveyed for 1 time every 5 days from the green turning to the heading stage, the survey field pieces relate to wheat fields with different ecological environments, 2-3 pieces of each field need to be surveyed, each field adopts 5 diagonal points for sampling, the survey time is limited to 8-10 hours or 16-18 hours of the day, the time is short, the task is heavy, the workload is large, and the working efficiency is low.

3. Sample retention is difficult. Due to the large number of collected samples and the strong mobility of live mites, great difficulty is brought to sample retention.

The present invention solves all the disadvantages of available method.

Disclosure of Invention

The invention aims to solve the technical problem of providing a wheat red spider forecast sample collecting and counting system and method aiming at the defects of the prior art.

The technical scheme of the invention is as follows:

a wheat red spider test and report sample collecting and counting system comprises a sample collecting device and a sample counting device; the sample collecting device comprises a sampling box body (1), a box cover assembly (2) and a vibration controller (3); the sample statistical device comprises a smart phone client and a server, wherein the smart phone client is used for shooting the red spiders in the sample acquisition device, transmitting the acquired images to the server with image recognition and statistical functions, and performing image preprocessing, image recognition and classified statistics on the server;

the sampling box body (1) comprises a box body (11), a rotating shaft seat (12) and a box cover rotating shaft inserting guide groove (13); the box body (11) is a cuboid box body lacking a top surface and one side surface adjacent to the top surface, the lacking one side surface is an open end of the sampling box body (1), the open end is aligned with the roots of the wheat seedlings when in use, the lacking top surface is used for accommodating a box cover (21) of a box cover component (2), and the lacking top surface is closed by the box cover (21) when in use; two side surfaces of two sides of the open end are provided with rotating shaft seats (12) near the top, the rotating shaft seats (12) are round holes, the diameter of each round hole is slightly larger than the outer diameter of the circle-cutting rotating shaft (22), and the circle-cutting rotating shaft (22) can freely rotate in the round hole; the box cover rotating shaft inserting guide groove (13) is a rectangular opening, the rectangular opening is communicated with the rotating shaft seat (12), the tangent circle rotating shaft (22) is inserted into the rotating shaft seat (12) through the rectangular opening, and the size of the rectangular opening is smaller than the outer diameter of the tangent circle rotating shaft (22).

The wheat red spider forecast sample collection and statistics system comprises a box cover (21), a tangent circle rotating shaft (22), a battery and controller box (23) integrated with the box cover, beating plate guide posts (24), guide post limit check rings (25) and beating plates (26), wherein four beating plate guide posts (24) are fixed below the box cover (21), a vibrating element (28) is also fixed below the box cover (21) and positioned at the central positions of the four beating plate guide posts (24), the beating plates (26) are fixed on the vibrating element (28), beating plate ribs (27) are arranged below the beating plates (26), and round holes are formed in four corners of the beating plates (26) and penetrate through the four beating plate guide posts (24) to freely vibrate on the four beating plate guide posts (24); guide post limit retainer rings (25) are fixed at the bottom ends of the four beating plate guide posts (24) and used for limiting the beating plates (26).

According to the wheat red spider forecast sample collection statistical system, a battery and controller box (23) is fixed with one side of a box cover (21) and one side of a beating plate (26), a tangential rotating shaft (22) is arranged on the upper portion of the battery and controller box (23), and after the tangential rotating shaft (22) is inserted into a rotating shaft seat (12), when the box cover (21) and the beating plate (26) rotate to be parallel to the bottom surface of a box body (11) or perpendicular to the bottom surface of the box body (11), the battery and controller box (23) can plug the open end of a sampling box body (1).

According to the wheat red spider forecast sample collection and statistics system, the cross section of the battery and the controller box (23) is in a sector shape of 90 degrees, the open end is well sealed by utilizing a peripheral arc, and the wheat plants cannot be damaged by the arc of the battery and the controller box (23) in the rotation process.

A method of sample collection according to any of the sample collection statistics systems, comprising the steps of:

a1, rightly arranging the open end of the sampling box body (1) at the root of the wheat seedling (6) to be sampled;

a2, vertically placing the box cover assembly (2) with the vibrator, and then pushing the tangent circle rotating shaft (22) on the side surface of the box cover into the bottom of the sampling box body (1) along the guide groove (13) to enable the rotating shaft to reach the rotating shaft seat (12);

a3, closing the box cover assembly (2), wherein in the process of installing the box cover assembly (2), a battery box shell of the box cover assembly (2) pushes upright wheat seedlings to be pressed against the edge of the bottom of the sampling box (1), in the process of closing the box cover, the arc-shaped battery box shell ensures that the wheat seedlings (6) are approximately horizontal when the box cover is closed, and after the box cover is closed, a beating plate and beating plate ribs (27) at the lower part of the box cover assembly are in contact with the wheat seedlings (6);

a4, opening a switch of a vibration controller (3) to connect a vibration element (28), wherein the vibration element (28) drives a beating plate and beating plate ribs to rapidly vibrate up and down along a beating plate guide post (24) to beat wheat;

a5, beating the wheat seedlings by vibration, and enabling the wheat red spiders to fall into the bottom of the sampling box body (1);

a6, after beating, closing the power supply, taking down the box cover assembly (2) and completing sample collection.

The identification statistical method of any sample collection statistical system comprises the following steps:

b1, image acquisition: collecting red spiders in the field, and shooting the red spiders in a collection box by a smart phone client;

b2, image preprocessing and graphic transformation: the server receives an image to be processed transmitted by the client, background removal, graying removal, noise removal and image binarization processing are firstly carried out, then a sliding window is adopted for target positioning, and a square frame is used for marking out the target to be detected;

(1) inputting a picture, and removing the background image by adopting a Lazy snapping method to perform gray processing; carrying out image binarization after Gaussian filtering;

(2) extracting and generating about 2000 candidate regions for the binarized image by using a selective search algorithm;

(3) performing expansion processing on the edge of each candidate region by adding p to 16 pixels, and framing the candidate region after expansion; transform each candidate region size to 227x 227;

b3, image recognition:

adopting a tensoflow frame to construct a parallelized AlexNet convolutional neural network model, inputting each candidate region into an R-CNN convolutional neural network to obtain a 4096-dimensional characteristic vector, judging whether the candidate region is a spider, and leaving a frame of the spider for the next processing;

the R-CNN convolutional neural network model is provided with two input ports, input data are subjected to multilayer processing of convolutional layers conv, activation functions relu, pooling layers pool and batch normalization BN, output data are subjected to cascade merge by adopting convolutional layers conv with parallel structures and then are connected by using full connection layers fc, and forward propagation results logits are obtained;

adopting a transfer learning strategy by the R-CNN convolutional neural network, pre-training on ImageNet, fully learning shallow features by using the convolutional neural network, using an SGD algorithm during training, and taking an initial learning rate of 0.001 and a batch _ size of 128; then removing the last full-connection layer which is pre-trained, modifying the number of categories, carrying out large-scale training on a small-scale data set by a user, and finely adjusting the CNN when the user label data is deficient, so that the capability of identifying the object type is improved;

b4, importing the feature vector derived from the neural network into an SVM classifier to generate a class score;

b5, setting a IoU index threshold (>0.3), calculating a IoU index, removing overlapped frames by adopting non-maximum inhibition and taking the highest-score frame as a basis, and obtaining a candidate frame with the highest score in the class;

b6, classification statistics: the classification is to identify two spiders and then count the number of the spiders;

b7, outputting the result: and transmitting the classified statistical result back to the client through the server communication module, and storing and identifying the statistical result by the client.

Adopt novel wheat red spider to survey and report sample collection statistical system, can realize quick collection, the sample of wheat red spider sample in wheat field and keep somewhere and get and accurate count, labour saving and time saving improves sample collection and statistical efficiency.

Drawings

FIG. 1 is a schematic diagram of a wheat red spider test sample collection statistical system; .

FIG. 2 is a schematic structural diagram of a sampling box;

FIG. 3 is a schematic view of a cover assembly;

FIG. 4 is a schematic diagram of a method of use;

FIG. 5 is a schematic view of a method of installation and use;

FIG. 6 is a flow chart of a mobile client method;

FIG. 7 is a flow chart of a server-side method;

1 a sampling box comprising: the 11 box body, the 12 rotating shaft seat and the 13 box cover rotating shaft are inserted into the guide grooves; a lid assembly, comprising: 21 box cover, 22 tangent circle rotating shaft, 23 battery and controller box, 24 beating plate guide post, 25 guide post limit retainer ring, 26 beating plate, 27 beating plate rib and 28 vibrating element; 3, a vibration controller; 4, a battery; 5, a battery box cover; 6 wheat seedlings.

Detailed Description

The present invention will be described in detail with reference to specific examples.

As shown in fig. 1-5, the wheat red spider test and report sample collecting and counting system comprises a sample collecting device and a sample counting device; the sample collecting device comprises a sampling box body 1, a box cover assembly 2 and a vibration controller 3; the sample statistical device comprises a smart phone client and a server, wherein the smart phone client is used for shooting the red spiders in the sample acquisition device, transmitting the acquired images to the server with image recognition and statistical functions, and performing image preprocessing, image recognition and classified statistics on the server;

the sampling box body 1 comprises a box body 11, a rotating shaft seat 12 and a box cover rotating shaft inserting guide groove 13; the box body 11 is a rectangular parallelepiped box body lacking a top face and one side face adjacent to the top face, the lacking one side face being an open end of the sampling box body 1, the open end being aligned with the roots of the wheat seedlings in use, the lacking top face being for accommodating the cover 21 of the cover assembly 2, the lacking top face being closed by the cover 21 in use; the two side surfaces of the two sides of the open end close to the top are provided with rotating shaft seats 12, the rotating shaft seats 12 are round holes, the diameter of each round hole is slightly larger than the outer diameter of the tangent circle rotating shaft 22, and the tangent circle rotating shaft 22 can freely rotate in the round hole; the cover spindle insertion guide groove 13 is a rectangular opening that communicates with the spindle base 12, through which the tangential spindle 22 is inserted into the spindle base 12, and the size of the rectangular opening is smaller than the outer diameter of the tangential spindle 22.

The box cover component 2 comprises a box cover 21, a tangential rotating shaft 22, a battery and controller box 23 integrated with the box cover, a beating plate guide post 24, a guide post limiting check ring 25 and a beating plate 26, wherein four beating plate guide posts 24 are fixed on the lower surface of the box cover 21, a vibrating element 28 is also fixed on the lower surface of the box cover 21 and is positioned at the central positions of the four beating plate guide posts 24, the beating plate 26 is fixed on the vibrating element 28, beating plate ribs 27 (in contact with wheat seedlings and applying vibration to the wheat seedlings) are arranged on the lower surface of the beating plate 26, and round holes are formed in four corners of the beating plate 26 and penetrate through the four beating plate guide posts 24 and can freely vibrate on the four beating plate guide posts 24; the guide post limiting retainer rings 25 are fixed at the bottom ends of the four beating plate guide posts 24 and used for limiting the beating plates 26;

the battery and controller box 23, the box cover 21 and one side of the clapper board 26 are fixed together, the upper part of the battery and controller box 23 is provided with a tangential rotating shaft 22, after the tangential rotating shaft 22 is inserted into the rotating shaft seat 12, when the box cover 21 and the clapper board 26 rotate to be parallel to the bottom surface of the box body 11 or perpendicular to the bottom surface of the box body 11, the battery and controller box 23 can seal the open end of the sampling box body 1 to prevent the red spiders from escaping, preferably, the cross section of the battery and the controller box 23 is in a 90-degree fan shape, the open end is well sealed by using peripheral circular arcs to prevent the red spiders from escaping, and the battery and the controller box 23 cannot damage wheat plants in the rotating process.

The vibrating element 28 is realized by a fixed coil and a spring, a movable iron core; the vibration controller 3 consists of an oscillation circuit and a driving circuit, the frequency of the oscillation circuit is adjustable to control the beating frequency, and the output current of the driving circuit is adjustable to control the beating strength; the battery 4 is constituted by a rechargeable battery pack; the smart phone 6 adopts a smart phone with a high-definition camera which is amplified by 20 times.

First, sample collection

1. The open end of the sampling box body 1 which meets the sampling specification and has a clear bottom plane of 333mm multiplied by 200mm is rightly arranged at the root part of the wheat seedling 6 to be sampled (see figure 4);

2. vertically placing the box cover component 2 with the vibrator, and then pushing the tangential rotating shaft 22 on the side surface of the box cover into the bottom of the sampling box body 1 along the guide groove 13 to enable the rotating shaft to reach the rotating shaft seat 12 (see figure 4);

3. closing the box cover assembly 2, wherein in the process of installing the box cover assembly 2, a battery box shell of the box cover assembly 2 pushes upright wheat seedlings to be pressed against the edge of the bottom of the sampling box 1, in the process of closing the box cover, the arc-shaped battery box shell ensures that the wheat seedlings 6 are approximately horizontal when the box cover is closed, and after the box cover is closed, a beating plate and beating plate ribs 27 on the lower part of the box cover assembly are in contact with the wheat seedlings 6 (see figure 5);

4. the vibration controller 3 is turned on, the switch is connected with the vibration element 28, the vibration element 28 drives the beating plate and the beating plate ribs to quickly vibrate up and down along the beating plate guide post 24 to beat the wheat, the beating frequency and speed of the vibration element 28 are regulated and controlled by the vibration controller 3, and the electric power of the vibration controller 3 is supplied by the battery 4;

5. through vibrating and beating the wheat seedlings, the wheat red spiders fall into the bottom of the sampling box body 1;

6. after beating, the power supply is closed, the box cover assembly 2 is taken down, and sample collection is completed.

Secondly, image acquisition, identification and statistics

1. Image acquisition: the red spiders are small in size, and the intelligent mobile phone or the digital camera which needs a high-magnification near-focus high-definition lens and a high-resolution camera for image recognition shoots the red spiders in the collection box under a proper light condition. The method comprises the steps that the red spiders are collected in the field, an App mobile phone with the functions of image remote transmission and data return is generally used by a smart mobile phone client, and collected images are transmitted to a server with the functions of image preprocessing, image transformation, image recognition and statistics through the Internet to be processed.

2. Image preprocessing and graphic transformation: the server receives the image to be processed transmitted by the client, firstly carries out processing such as background removal, graying, noise removal, image binarization and the like, then adopts a sliding window to carry out target positioning, and marks out the target to be detected by a square frame.

(1) Inputting a picture, and removing the background image by adopting a Lazy snapping method to perform gray processing; carrying out image binarization after Gaussian filtering;

(2) extracting and generating about 2000 candidate regions for the binarized image by using a selective search algorithm;

(3) performing expansion processing on the edge of each candidate region by adding p to 16 pixels, and framing the candidate region after expansion; each candidate region size is transformed to 227x 227.

3. Image recognition:

and (3) constructing a parallelized AlexNet convolutional neural network model by adopting a tensoflow frame, inputting each candidate region into the R-CNN convolutional neural network to obtain a 4096-dimensional characteristic vector, judging whether the candidate region is a spider, and leaving a frame of the spider for the next processing.

The R-CNN convolutional neural network model is provided with two input ports, input data are subjected to multilayer processing of convolutional layers conv, activation functions relu, pooling layers pool and batch normalization BN, output data are subjected to cascade merge by adopting convolutional layers conv with parallel structures and then are connected by using full connection layers fc, and forward propagation results logits are obtained;

adopting a transfer learning strategy by the R-CNN convolutional neural network, pre-training on ImageNet, fully learning shallow features by using the convolutional neural network, using an SGD algorithm during training, and taking an initial learning rate of 0.001 and a batch _ size of 128; and then removing the last full-connection layer which is pre-trained, modifying the class number (3 classes are 2 targets +1 background), carrying out scale training on a small-scale data set by a user, and finely adjusting the CNN when the user labeling data is lack, so that the capability of identifying the object type is improved.

4. Importing the feature vector derived from the neural network into an SVM classifier to generate a class score;

5. setting IoU index threshold (>0.3), calculating IoU index, adopting non-maximum inhibition, and removing overlapped frames based on the highest-score frame to obtain the candidate frame with the highest score in the category.

6. And (4) classified statistics: the classification is to identify two spiders and then count their numbers.

7. And (4) outputting a result: and transmitting the classified statistical result back to the client through the server communication module, and storing and identifying the statistical result by the client.

It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.