Method and device for automatically cleaning ampullaria gigas eggs based on intelligent image recognition

Abstract

The invention belongs to the technical field of pomacea canaliculata egg cleaning, and provides an automatic pomacea canaliculata egg cleaning method and device based on intelligent image recognition. The method comprises the following steps: establishing a data set for training through ampullaria gigas egg image data of various scenes, training a neural network model-YOLOv8 target detection model through an image recognition technology, constructing an automatic ampullaria gigas egg recognition module based on the trained model, carrying out real-time image acquisition on a target water area based on a carrying main body and a mechanical arm, and carrying out real-time detection on the target water area. The ampullaria gigas eggs are automatically identified and subsequently processed; according to the device, manual intervention is not needed for daily cleaning operation of the ampullaria gigas eggs, the device is suitable for various scenes where the ampullaria gigas eggs appear, such as grass rods, soil and concrete, and therefore intelligent and automatic ampullaria gigas egg cleaning is achieved, the labor cost is reduced, potential safety hazards are avoided, and the cleaning efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of cleaning Pomacea canaliculata eggs. Specifically, it relates to an automatic cleaning method and device for Pomacea canaliculata eggs based on image intelligent recognition. Background Art

[0002] Pomacea canaliculata is a mollusk with a similar appearance to field snails. Its eggs are bright pink in color. Pomacea canaliculata carries a large number of parasites, and if accidentally ingested, it will cause great harm to the human body. The egg masses of Pomacea canaliculata are bright red and extremely easy to identify. Therefore, the prevention and control of Pomacea canaliculata should start from preventing and controlling its eggs.

[0003] Pomacea canaliculata generally lays its eggs on objects near water areas that are easy to attach to, such as stones, grass stalks, and soil. Currently, the methods for cleaning Pomacea canaliculata eggs basically rely on manual removal and manual killing. The main deficiencies are as follows: The efficiency of manual removal is very low, the speed is limited, and Pomacea canaliculata generally lays its eggs near water, where the environment is slippery, and there are safety hazards in manual cleaning; on the other hand, manual killing generally uses in-situ killing, such as spraying pesticides or burning, which has a certain impact on the environment. Summary of the Invention

[0004] The purpose of the present invention is to provide an automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition to solve the technical problems existing in the prior art.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows: An automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition, comprising the following steps: Step S1: Arrange the Pomacea canaliculata egg cleaning execution device in the target water area and operate the Pomacea canaliculata egg cleaning execution device according to the navigation path; Step S2: Based on the running path, the Pomacea canaliculata egg cleaning execution device real-time collects real-time image data of the target water area; Step S3: The Pomacea canaliculata egg automatic recognition module identifies and judges Pomacea canaliculata eggs according to the real-time image data. If Pomacea canaliculata eggs are identified, the next step is executed; the construction method of the Pomacea canaliculata egg automatic recognition module is as follows: Step S3.1: Obtain image data of Pomacea canaliculata eggs appearing in various scenarios, and build a data set for model training based on the image data; Step S3.2: Train the YOLOv8 target detection model based on the data set; Step S3.3: Build the Pomacea canaliculata egg automatic recognition module according to the trained YOLOv8 target detection model; Step S4: Plan a path according to the position of the identified Pomacea canaliculata eggs, and control the Pomacea canaliculata egg cleaning execution device to move to the target position based on the planned path; Step S5: After reaching the target position, control the Pomacea canaliculata egg cleaning execution device to collect the Pomacea canaliculata eggs; Step S6: Place the collected Pomacea canaliculata eggs in the disinfection frame and disinfect and clean the Pomacea canaliculata eggs.

[0006] In one embodiment, it further includes step S7 of iteratively training the YOLOv8 object detection model using the real-time image data identifying the Pomacea canaliculata eggs as training data.

[0007] In one embodiment, it further includes the step of optimizing the navigation path: predicting the high-incidence areas of Pomacea canaliculata eggs in the target water area based on time series and meteorological data, judging whether the current navigation path covers the high-incidence areas, if not, incorporating the high-incidence areas into the navigation path, and then regenerating the navigation path.

[0008] In one embodiment, the scenarios in step S3.1 include straw, soil, and concrete.

[0009] To achieve the above object, the present invention further provides an automatic Pomacea canaliculata egg cleaning device based on image intelligent recognition, including: A Pomacea canaliculata egg cleaning execution device arranged in the target water area and used to execute Pomacea canaliculata egg cleaning operations; A visual acquisition module for collecting and transmitting the real-time image data of the target water area in real time; A navigation module that provides a navigation path and is used for the automatic cruise and return of the Pomacea canaliculata egg cleaning execution device; A Pomacea canaliculata egg automatic recognition module that receives the real-time image data and identifies and judges Pomacea canaliculata eggs based on the real-time image data; A control module that controls the Pomacea canaliculata egg cleaning execution device according to a control command.

[0010] In one embodiment, the Pomacea canaliculata egg cleaning execution device includes a carrier body, a robotic arm and a disinfection frame arranged on the carrier body, wherein a battery for power supply is arranged on the carrier body.

[0011] In one embodiment, chemical agents for disinfecting Pomacea canaliculata eggs are stored in the disinfection frame.

[0012] In one embodiment, the control module includes: An automatic control unit that receives the recognition result of the Pomacea canaliculata egg automatic recognition module and automatically controls the carrier body and the robotic arm according to the recognition result; A remote control unit for receiving control instructions from remote / on-site operators and controlling the carrier body and the robotic arm according to the control instructions.

[0013] In one embodiment, the automatic control unit includes: An identification information receiving module for receiving the recognition result of the Pomacea canaliculata egg automatic recognition module; A path planning module that plans the operation path of the carrier based on the current position of the carrier according to the identified position of the apple snail eggs. A carrier control module that controls the carrier to move to the target position based on the planned path. A robotic arm control module that controls the robotic arm to collect the apple snail eggs and controls the robotic arm to place the collected apple snail eggs in the disinfection box.

[0014] In one implementation, the visual acquisition module includes: A camera for collecting real-time image data. An image transmission module for transmitting the real-time image data to the apple snail egg automatic recognition module.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention collects apple snail egg image data in various scenarios and establishes a training dataset, then trains a neural network model - YOLOv8 object detection model through image recognition technology, constructs an apple snail egg automatic recognition module based on the trained model, collects real-time images of the target water area based on the carrier and the robotic arm, and automatically recognizes and subsequent processes the apple snail eggs; according to the present invention, the daily cleaning operation of apple snail eggs does not require manual intervention, and is applicable to various scenarios where apple snail eggs appear, such as grass stems, soil, concrete, etc. Thus, the intelligent and automatic cleaning of apple snail eggs is realized, reducing labor costs, avoiding safety hazards, and improving cleaning efficiency. Description of the Drawings

[0016] Figure 1 It is a flowchart of Embodiment 1 of the present invention.

[0017] Figure 2 It is a structural schematic diagram of Embodiment 2 of the present invention.

[0018] Among them, the names corresponding to the reference numerals are as follows: 1, carrier; 2, control module; 3, camera; 4, robotic arm; 5, disinfection box. 101, double hull; 102, paddle wheel. 301, hull camera A; 302, robotic arm camera; 303, hull camera B. Detailed Embodiments

[0019] In order to enable those skilled in the art to have a clearer understanding and knowledge of the present invention, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described below are only used to explain the present invention for easy understanding, and the technical solutions provided by the present invention are not limited to the technical solutions provided by the following embodiments, nor should the technical solutions provided by the embodiments limit the protection scope of the present invention.

[0020] Embodiment 1 As Figure 1 shown, this embodiment provides an automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition. This cleaning method is used to deal with Pomacea canaliculata eggs that cause great ecological harm. Taking the Pomacea canaliculata egg cleaning execution device as the execution component, combined with the automatic recognition technology of Pomacea canaliculata eggs, based on the automatic cruise of the target water area, the Pomacea canaliculata eggs in the target water area are automatically cleaned, thereby realizing intelligent and automatic cleaning of Pomacea canaliculata eggs, which not only reduces labor costs and has high cleaning efficiency, but also avoids the drawbacks existing in manual cleaning.

[0021] In this embodiment, the automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition includes the following steps: I. Cruise start Arrange the Pomacea canaliculata egg cleaning execution device in the target water area and run the Pomacea canaliculata egg cleaning execution device according to the navigation path. The Pomacea canaliculata egg cleaning execution device realizes automatic cruise and return; the composition of the water area is relatively complex. Although the setting of the conventional cruise path can realize the automatic cruise and return of the device, the situation of each water area will be different. If the conventional cruise path (navigation path) is adopted for all, it is easy to have missed inspection. To solve the above problems, in this embodiment, the navigation path is further optimized as follows: Based on time series and meteorological data, predict the high-incidence area of Pomacea canaliculata eggs in the target water area, and judge whether the current navigation path covers the high-incidence area. If not, incorporate the high-incidence area into the navigation path and then regenerate the navigation path. The prediction model can adopt the ecological niche model. The distribution of Pomacea canaliculata is significantly affected by climate factors (such as temperature, precipitation). The ecological niche model can predict its potential suitable habitat. For example: MaxEnt (Maximum Entropy Model): It performs best (high AUC value) in the prediction of the suitable habitat of Pomacea canaliculata, is suitable for small sample data, and can be combined with the data of local meteorological stations in the current water area (such as annual average temperature, annual precipitation) for prediction. At the same time, it can also be combined with the time series prediction model. For example: ARIMA (Autoregressive Integrated Moving Average Model): Based on the historical data of time series data of temperature and rainfall, predict the short-term density change of Pomacea canaliculata eggs. For the areas where the density change shows an increasing trend, they are used as preferred cruise path points, or these areas are cruised repeatedly (multiple cruise path points are arranged in the same area) to further provide the effect of avoiding missed inspection.

[0022] In this embodiment, the apple snail egg cleaning execution device includes a carrier body, a robotic arm and a disinfection frame arranged on the carrier body. Among them, a battery for power supply is arranged on the carrier body, and the carrier body adopts a double-hull structure and is equipped with paddle wheels and a driving motor.

[0023] II. Visual Acquisition The apple snail egg cleaning execution device automatically cruises in the target water area according to the navigation path. At the same time, during the cruising process, the apple snail egg cleaning execution device collects real-time image data of the target water area. The acquisition of the image data is achieved through a camera. Based on the structure of the above-mentioned carrier body, two cameras A can be arranged on the hull and one camera can be arranged on the robotic arm for the layout of the cameras. Considering the color change of apple snail eggs, the camera is preferably a multi-spectral camera to enhance the ability to capture color changes.

[0024] III. Image Recognition The apple snail egg automatic recognition module identifies and judges apple snail eggs according to the real-time image data. If apple snail eggs are identified, the next step is executed; in this step, the construction method of the apple snail egg automatic recognition module is as follows: Step S3.1: Obtain the image data of apple snail eggs appearing in various scenarios, and construct a data set for model training based on the image data; Step S3.2: Train the YOLOv8 object detection model based on the data set. This model can frame and identify the apple snail eggs appearing in the collected images; Step S3.3: According to the trained YOLOv8 object detection model, construct the apple snail egg automatic recognition module, and realize the recognition of apple snail eggs in the real water area environment through the apple snail egg automatic recognition module; among them, the scenarios include grass stems, soil and concrete. The image data of apple snail eggs appearing in various scenarios can be real-time collected data, historical image data, and synthetic data generated by using a generative adversarial network (GAN) to simulate egg mass images under different water quality and light conditions (which can improve the detection stability in complex scenarios and avoid missed detection and misdetection); for the real-time image data in which apple snail eggs are identified, use it as training data to iteratively train the YOLOv8 object detection model to improve the accuracy and reliability of the model.

[0025] IV. Path Planning According to the position of the identified apple snail eggs and the current position of the apple snail egg cleaning execution device, plan the moving path of the apple snail egg cleaning execution device. This moving path includes the path to move to the target position and the path to return to the cruising path at the target position; based on the planned path, control the apple snail egg cleaning execution device to move to the target position, and this target position is the position of the water area where the apple snail egg cleaning execution device can reach the apple snail eggs and clean them.

[0026] V. Cleaning and Collecting Apple Snail Eggs After reaching the target position, control the apple snail egg cleaning execution device to collect apple snail eggs. The collection can be achieved in combination with a robotic arm, and the collection methods can be grasping, shoveling, etc.

[0027] VI. Centralized killing Place the collected apple snail eggs in a disinfection box for disinfection and cleaning of the apple snail eggs. The disinfection box stores chemical reagents that can kill apple snail eggs. There are various choices of chemical reagents. For example: niclosamide, mechanism of action: inhibiting oxidative phosphorylation of egg mitochondria and blocking energy metabolism, effect: the inhibition rate of apple snail egg hatching reaches over 95% (soaking in a concentration of 5 mg / L for 48 hours); cypermethrin, mechanism of action: penetrating the eggshell and interfering with nerve conduction, effect: the hatching rate decreases by 90% at a concentration of 0.5 mg / L; copper sulfate, mechanism of action: copper ions destroying the enzyme system in the eggs, effect: spraying with a 5% solution can kill the eggs.

[0028] After completing the above operations, the apple snail egg cleaning execution device returns to the cruising path at the target position according to the planned path and continues cruising detection and cleaning.

[0029] Embodiment 2 As Figure 2 shown, the present invention also provides an automatic apple snail egg cleaning device based on image intelligent recognition, including: An apple snail egg cleaning execution device, arranged in the target water area and used to execute apple snail egg cleaning operations; the apple snail egg cleaning execution device includes a carrier body, a robotic arm and a disinfection box provided on the carrier body. Among them, a battery for power supply is provided on the carrier body, and the carrier body adopts a double-hull structure, equipped with paddle wheels and drive motors; the disinfection box stores chemical agents for disinfecting apple snail eggs. There are various choices of chemical reagents. For example: niclosamide, mechanism of action: inhibiting oxidative phosphorylation of egg mitochondria and blocking energy metabolism, effect: the inhibition rate of apple snail egg hatching reaches over 95% (soaking in a concentration of 5 mg / L for 48 hours); cypermethrin, mechanism of action: penetrating the eggshell and interfering with nerve conduction, effect: the hatching rate decreases by 90% at a concentration of 0.5 mg / L; copper sulfate, mechanism of action: copper ions destroying the enzyme system in the eggs, effect: spraying with a 5% solution can kill the eggs.

[0030] The visual acquisition module is used to collect real-time image data of the target water area in real time and transmit it. The visual acquisition module includes a camera for collecting real-time image data, and an image transmission module for transmitting the real-time image data to the automatic recognition module of Pomacea canaliculata eggs. Preferably, three cameras are arranged, two of which are respectively arranged on both sides of the bow of the double hull: hull camera A and hull camera B, and the other is arranged on the robotic arm: robotic arm camera. Preferably, the image transmission module and the camera are integrated. Considering the color change of Pomacea canaliculata eggs, the camera is preferably a multi-spectral camera to enhance the ability to capture color changes.

[0031] The navigation module provides a navigation path and is used for the automatic cruise and return of the Pomacea canaliculata egg cleaning execution device. In a further preferred solution, the navigation module further includes a navigation path optimization module, which is implemented as follows: based on time series and meteorological data, predict the high-incidence area of Pomacea canaliculata eggs in the target water area, and judge whether the current navigation path covers the high-incidence area. If not, incorporate the high-incidence area into the navigation path, and then regenerate the navigation path.

[0032] The automatic recognition module of Pomacea canaliculata eggs receives real-time image data and identifies and judges Pomacea canaliculata eggs according to the real-time image data. The construction method of the automatic recognition module of Pomacea canaliculata eggs is as follows: (1) Obtain the image data of Pomacea canaliculata eggs appearing in various scenarios, and construct a data set for model training based on the image data; (2) Train the YOLOv8 object detection model based on the data set, and this model can frame and identify the Pomacea canaliculata eggs appearing in the collected images; (3) According to the trained YOLOv8 object detection model, construct the automatic recognition module of Pomacea canaliculata eggs, and realize the identification of Pomacea canaliculata eggs in the real water environment through the automatic recognition module of Pomacea canaliculata eggs.

[0033] The control module controls the apple snail egg cleaning execution device according to control commands. In this embodiment, the control module includes: an automatic control unit and a remote control unit. Among them, the automatic control unit receives the recognition results of the apple snail egg automatic recognition module and automatically controls the carrier body and the robotic arm according to the recognition results; the remote control unit is used to receive control instructions from remote / on-site operators and control the carrier body and the robotic arm according to the control instructions. Specifically, the automatic control unit includes: a recognition information receiving module for receiving the recognition results of the apple snail egg automatic recognition module; a path planning module that, based on the position of the recognized apple snail eggs and the current position of the carrier body, plans the running path of the carrier body, and this running path includes the path to move to the target position and the path to return to the cruising path at the target position; a carrier body control module that controls the carrier body to move to the target position based on the planned path, and after completing the cleaning operation, controls the carrier body to return from the destination position to the cruising path; a robotic arm control module that controls the robotic arm to collect the apple snail eggs and controls the robotic arm to place the collected apple snail eggs in the disinfection box, and in the disinfection box, the apple snail eggs are centrally killed based on chemical reagents.

[0034] In this embodiment, the navigation module, the apple snail egg automatic recognition module, and the control module are functional modules, and their implementation principles correspond to the technical solutions provided in Embodiment 1. Their hardware implementation may include a PCB circuit board and an IC chip. For example, it may be composed of a single packaged IC, or may be composed of multiple packaged ICs with the same or different functions connected together.

[0035] The above embodiments merely illustrate the principles and effects of the present invention, rather than limiting the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes completed by those with ordinary knowledge in the technical field without departing from the spirit and technical idea disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. An automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition, characterized in that It includes the following steps: Step S1: Arrange the apple snail egg cleaning execution device in the target water area and run the apple snail egg cleaning execution device according to the navigation path; Step S2: Based on the running path, the apple snail egg cleaning execution device real-time collects the real-time image data of the target water area; Step S3: The apple snail egg automatic recognition module recognizes and judges apple snail eggs according to the real-time image data. If apple snail eggs are recognized, the next step is executed; The construction method of the apple snail egg automatic recognition module is as follows: Step S3.1: Obtain the image data of apple snail eggs appearing in various scenarios, and construct a data set for model training based on the image data; Step S3.2: Train the YOLOv8 target detection model based on the data set; Step S3.3: According to the trained YOLOv8 target detection model, construct the apple snail egg automatic recognition module; Step S4: Plan a path according to the position of the recognized apple snail eggs, and control the apple snail egg cleaning execution device to move to the target position based on the planned path; Step S5: After reaching the target position, control the apple snail egg cleaning execution device to collect the apple snail eggs; Step S6: Place the collected apple snail eggs in the disinfection box and disinfect and clean the apple snail eggs.

2. The automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition according to claim 1, wherein, It also includes step S7, using the real-time image data of the recognized apple snail eggs as training data to iteratively train the YOLOv8 target detection model.

3. The automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition according to claim 2, characterized in that, It also includes the step of optimizing the navigation path: Based on time series and meteorological data, predict the high-incidence areas of apple snail eggs in the target water area, judge whether the current navigation path covers the high-incidence areas. If not, include the high-incidence areas in the navigation path, and then regenerate the navigation path.

4. The automatic cleaning method for Pomacea canaliculata eggs based on image intelligent recognition according to claim 3, characterized in that, The scenarios in step S3.1 include straw, soil, and concrete.

5. An automatic cleaning device for Pomacea canaliculata eggs based on image intelligent recognition, characterized in that, It includes: An apple snail egg cleaning execution device, arranged in the target water area and used to execute apple snail egg cleaning operations; A visual acquisition module, used to real-time collect the real-time image data of the target water area and transmit it; A navigation module, providing a navigation path and used for the automatic cruise and return of the apple snail egg cleaning execution device; An apple snail egg automatic recognition module, receiving the real-time image data and recognizing and judging apple snail eggs according to the real-time image data; A control module, controlling the apple snail egg cleaning execution device according to the control command.

6. The automatic cleaning device for Pomacea canaliculata eggs based on image intelligent recognition according to claim 5, characterized in that, The apple snail egg cleaning execution device includes a carrier body (1), a robotic arm (2) and a disinfection box (3) arranged on the carrier body (1). Among them, a battery for power supply is arranged on the carrier body (1).

7. The automatic cleaning device for apple snail eggs based on image intelligent recognition according to claim 6, characterized in that, The disinfection box (3) stores chemical agents for disinfecting apple snail eggs.

8. The automatic cleaning device for Pomacea canaliculata eggs based on image intelligent recognition according to claim 5, characterized in that, The control module includes: An automatic control unit, receiving the recognition result of the apple snail egg automatic recognition module, and automatically controlling the carrier body and the robotic arm according to the recognition result; A remote control unit, used to receive control instructions from remote / on-site operators, and controlling the carrier body and the robotic arm according to the control instructions.

9. The automatic cleaning device for Pomacea canaliculata eggs based on image intelligent recognition according to claim 8, characterized in that, The automatic control unit includes: A recognition information receiving module, used to receive the recognition result of the apple snail egg automatic recognition module; A path planning module that plans the running path of the carrier based on the current position of the carrier according to the identified position of the Pomacea canaliculata eggs; A carrier control module that controls the movement of the carrier to the target position based on the planned path; A robotic arm control module that controls the robotic arm to collect the Pomacea canaliculata eggs and controls the robotic arm to place the collected Pomacea canaliculata eggs in the disinfection box.

10. The automatic cleaning device for apple snail eggs based on image intelligent recognition according to claim 5, characterized in that, The visual acquisition module includes: A camera for acquiring real-time image data; An image transmission module for transmitting the real-time image data to the Pomacea canaliculata egg automatic recognition module.

Images