Intelligent feeding method, device and equipment, storage medium and computer program product

Abstract

The invention relates to the technical field of intelligent feeding, and discloses an intelligent feeding method, device and equipment, a storage medium and a computer program product. The method comprises the steps that the current fish school density and the current fish school shape distribution of a target fish school in a target fishpond are obtained through a three-dimensional monitoring network module, and the current fish school ingestion activeness is determined based on the current fish school density and the current fish school shape distribution; the current water area temperature of the target fishpond is obtained, the fish-food ratio is determined through a preset metabolism energy efficiency model according to the current water area temperature and the current fish school shape distribution, and the preset metabolism energy efficiency model is obtained by training the initial neural network model through sample fish school shape distribution and sample water area temperature; a target feeding decision is generated based on the current fish school ingestion activeness and the fish-food ratio, and feeding is conducted in the target fishpond based on the target feeding decision.

Description

Technical Field

[0001] This application relates to the field of intelligent feeding technology, and in particular to an intelligent feeding method, apparatus, device, storage medium, and computer program product. Background Technology

[0002] Intelligent feeding is an advanced feeding technology that uses artificial intelligence algorithms to dynamically generate precise feeding decisions and control the execution of feeding equipment. It automatically adjusts the amount, frequency, and location of feeding based on environmental conditions, improving breeding efficiency, reducing feed waste, and minimizing environmental impact.

[0003] Existing intelligent feeding systems mostly adjust feeding amounts based on a single water temperature parameter (e.g., increasing feeding when the water temperature rises because fish have a faster metabolic rate at higher temperatures), ignoring the complexity of fish behavior and environmental conditions. Adjusting feeding amounts solely based on water temperature changes cannot accurately match the actual needs of the fish and may lead to overfeeding or underfeeding. Summary of the Invention

[0004] The main purpose of this application is to provide an intelligent feeding method, which aims to solve the technical problem of how to perform feeding.

[0005] To achieve the above objectives, this application proposes an intelligent feeding method, which is applied to an intelligent feeding device equipped with a three-dimensional monitoring network module. The method includes: The current fish density and current fish size distribution of the target fish population in the target fish pond are obtained through the three-dimensional monitoring network module, and the current feeding activity of the fish population is determined based on the current fish density and current fish size distribution. The current water temperature of the target fish pond is obtained, and the fish feed ratio is determined based on the current water temperature and the current fish population size distribution using a preset metabolic energy efficiency model. The preset metabolic energy efficiency model is obtained by training an initial neural network model using the sample fish population size distribution and sample water temperature. Based on the current feeding activity of the fish population and the fish feed ratio, a target feeding decision is generated, and the target fish population is fed in the target fish pond according to the target feeding decision.

[0006] In one embodiment, the step of determining the fish feed ratio based on the current water temperature and the current fish population size distribution using a preset metabolic efficiency model includes: Statistical moment processing is performed on the current fish group size distribution to obtain the body size feature vector; The current water temperature is discretized, and the discretization result and the body shape feature vector are concatenated to obtain the target feature vector; The target feature vector is input into the preset metabolic efficiency model to obtain the fish feed ratio.

[0007] In one embodiment, the step of generating a target feeding decision based on the current feeding activity of the fish population and the fish feed ratio includes: A primary decision vector is determined by a preset feeding decision network based on the current feeding activity of the fish population and the fish feed ratio. Obtain the species constant corresponding to the target fish population and the effective water volume corresponding to the target fish pond; The average body length of the target fish group is determined based on the current fish group size distribution, and the saturation coefficient is obtained based on the species constant, the effective water volume, and the average body length. Based on the satiety coefficient, a target feeding decision is generated from the primary decision vector.

[0008] In one embodiment, the three-dimensional monitoring network includes a camera unit and a sonar unit. The step of obtaining the current fish density and current fish size distribution in the target fishpond through the three-dimensional monitoring network module includes: The camera unit acquires images of fish in the target fishpond, and performs network segmentation on the fish images to obtain the current fish size distribution in the target fishpond. The sonar unit acquires sonar images of the target fishpond and detects the sonar images to obtain the current fish density in the target fishpond.

[0009] In one embodiment, the step of performing network segmentation on the fish image to obtain the current fish size distribution within the target fishpond includes: The image of the fish school is segmented based on network segmentation technology to obtain the contour information of each member of the target fish school; Based on the contour information, feature extraction is performed to obtain the body shape characteristics of each member of the fish group; The current fish size distribution in the target fish pond is obtained based on the aforementioned body size characteristics.

[0010] In one embodiment, the step of detecting the sonar image to obtain the current fish density in the target fishpond includes: Based on a preset target detection algorithm, feature extraction is performed on the sonar image to obtain the reflective pixels in the sonar image; Target reflection pixels are selected from each of the reflection pixels according to a preset selection rule, and the number of members in the target fish group is determined based on the target reflection pixels. Obtain the area of ​​the target fish pond, and determine the current fish density in the target fish pond based on the area of ​​the fish pond and the number of members.

[0011] Furthermore, to achieve the above objectives, this application also proposes an intelligent feeding device, the device comprising: The activity acquisition module is used to acquire the current fish density and current fish size distribution of the target fish group in the target fish pond through the three-dimensional monitoring network module, and determine the current feeding activity of the fish group based on the current fish density and the current fish size distribution. The fish feed ratio module is used to obtain the current water temperature of the target fish pond and determine the fish feed ratio based on the current water temperature and the current fish population size distribution through a preset metabolic energy efficiency model. The preset metabolic energy efficiency model is obtained by training an initial neural network model with the sample fish population size distribution and sample water temperature. The intelligent feeding module is used to generate a target feeding decision based on the current feeding activity of the fish population and the fish feed ratio, and to feed the target fish population in the target fish pond according to the target feeding decision.

[0012] In addition, to achieve the above objectives, this application also proposes an intelligent feeding device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the intelligent feeding method as described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the intelligent feeding method described above.

[0014] In addition, to achieve the above objectives, this application also proposes a computer program product comprising a computer program that, when executed by a processor, implements the steps of the intelligent feeding method described above.

[0015] This application proposes an intelligent feeding method, device, equipment, storage medium, and computer program product. The method, applied to an intelligent feeding device, includes: acquiring the current fish density and body size distribution of a target fish population in a target fishpond through a three-dimensional monitoring network module, and determining the current feeding activity level of the fish population based on the current fish density and body size distribution; acquiring the current water temperature of the target fishpond, and determining the fish feed ratio based on the current water temperature and current fish body size distribution using a preset metabolic efficiency model, wherein the preset metabolic efficiency model is obtained by training an initial neural network model using sample fish body size distribution and sample water temperature; generating a target feeding decision based on the current fish feeding activity level and the fish feed ratio, and feeding the target fish population in the target fishpond according to the target feeding decision. By acquiring fish density and body size distribution through a three-dimensional monitoring network module, combining the preset metabolic efficiency model, comprehensively considering water temperature and fish body size distribution to determine the fish feed ratio, and then generating a precise feeding decision based on the fish feeding activity level and the fish feed ratio, precise feeding is achieved. This method solves the problem of adjusting feeding based solely on a single water temperature parameter, ignoring the complexity of fish behavior and environmental conditions. It effectively avoids low aquaculture efficiency caused by overfeeding or underfeeding, thereby improving aquaculture efficiency. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the first embodiment of the intelligent feeding method proposed in this application; Figure 2 This is a flowchart of the second embodiment of the intelligent feeding method proposed in this application; Figure 3 This is a flowchart of the third embodiment of the intelligent feeding method proposed in this application; Figure 4 A diagram of an intelligent feeding device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an intelligent feeding device suitable for implementing the embodiments of this application.

[0019] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of this application and are not intended to limit this application.

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0022] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0023] Understandably, intelligent feeding is an advanced feeding technology that uses artificial intelligence algorithms to dynamically generate precise feeding decisions and control the execution of feeding equipment. This technology can automatically adjust the feeding amount, frequency, and location based on changes in environmental conditions, such as water temperature, fish density, and dissolved oxygen levels. In this way, intelligent feeding not only improves aquaculture efficiency but also effectively reduces feed waste and lowers the burden on the environment.

[0024] However, most existing intelligent feeding systems adjust feeding amounts based solely on a single water temperature parameter. For example, when the water temperature rises, the system increases the feeding amount because it's generally believed that fish have a faster metabolic rate at higher temperatures. But this approach ignores the complexity of fish behavior and environmental conditions. In reality, fish feeding needs are also influenced by various factors such as the fish's health status, body size distribution, and dissolved oxygen levels in the water. Adjusting feeding amounts based solely on water temperature changes cannot accurately match the actual needs of the fish, easily leading to overfeeding or underfeeding.

[0025] Therefore, to address the technical challenge of feeding, this embodiment proposes an intelligent feeding method. This method is applied to an intelligent feeding device equipped with a three-dimensional monitoring network module. The method includes: acquiring the current fish density and size distribution of the target fish population in the target fishpond through the three-dimensional monitoring network module, and determining the current feeding activity level of the fish population based on the current fish density and size distribution; acquiring the current water temperature of the target fishpond, and determining the feed ratio based on the current water temperature and current fish size distribution using a preset metabolic efficiency model. The preset metabolic efficiency model is obtained by training an initial neural network model using sample fish size distribution and sample water temperature; generating a target feeding decision based on the current feeding activity level and feed ratio, and feeding the fish in the target fishpond based on the target feeding decision. By acquiring fish density and size distribution through the three-dimensional monitoring network module, combining the preset metabolic efficiency model, comprehensively considering water temperature and fish size distribution to determine the feed ratio, and then generating a precise feeding decision based on the feeding activity level and feed ratio, precise feeding is achieved. This method solves the problem of adjusting feeding based solely on a single water temperature parameter, ignoring the complexity of fish behavior and environmental conditions. It effectively avoids low aquaculture efficiency caused by overfeeding or underfeeding, thereby improving aquaculture efficiency.

[0026] For ease of understanding, the following is combined with Figures 1 to 5 The present application provides a detailed description of the intelligent feeding method provided in the embodiments, as well as the intelligent feeding method, apparatus, device, storage medium, and computer program product provided in the following embodiments.

[0027] This application provides an intelligent feeding method, referring to... Figure 1 , Figure 1 This is a flowchart of the first embodiment of the intelligent feeding method proposed in this application.

[0028] like Figure 1 As shown, the method includes: Step S10: Obtain the current fish density and current fish size distribution of the target fish group in the target fish pond through the three-dimensional monitoring network module, and determine the current feeding activity of the fish group based on the current fish density and the current fish size distribution.

[0029] It should be noted that the aforementioned three-dimensional monitoring network module can be a system integrating multiple sensors and data processing units, such as underwater cameras, sonar sensors, and data processing units, used to collect data such as fish density and body size distribution in real time. The aforementioned target fishpond can be a specific body of water used for fish farming, such as a large freshwater or seawater aquaculture pond, the scope and parameters of which can be set according to aquaculture needs. The aforementioned target fish population can be a specific fish group monitored and fed within the target fishpond, such as a particular species of salmon or sea bass. The aforementioned current fish population density can be the number of fish per unit volume of water, such as the number of fish per cubic meter of water, used to measure the distribution of the fish population. The aforementioned current fish population body size distribution can be the proportion of individuals of different body sizes within the fish population, such as the fish population structure divided by body length or weight, reflecting the growth status and structure of the fish population. The aforementioned feeding activity can be an indicator measuring the feeding enthusiasm of the fish population, determined, for example, by analyzing the activity status, swimming speed, and feeding behavior patterns of the fish population.

[0030] Specifically, underwater cameras capture images of fish activity, sonar sensors detect the location and number of fish, and a data processing unit analyzes and processes the data. Based on the activity images, location, and number of fish, the equipment uses preset algorithms and models to calculate the current feeding activity level of the fish.

[0031] For ease of understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, in a freshwater aquaculture scenario, the aforementioned equipment is installed next to a large aquaculture pond. The underwater camera and sonar sensor in the three-dimensional monitoring network module collect real-time images and sound signals of the fish's activity. The underwater camera captures multiple frames per second, using image processing algorithms to identify the fish's outlines and quantity; the sonar sensor detects the fish's location and depth through sound wave reflection. The data processing unit integrates this data, calculating the current fish density to be 50 fish per cubic meter of water, with the size distribution showing most fish being between 10 and 15 centimeters in length. Based on this data, the equipment analyzes the fish's activity status and feeding behavior patterns using a preset algorithm, determining that the current feeding activity of the fish is high.

[0032] Step S20: Obtain the current water temperature of the target fish pond, and determine the fish feed ratio based on the current water temperature and the current fish population size distribution using a preset metabolic energy efficiency model. The preset metabolic energy efficiency model is obtained by training an initial neural network model using the sample fish population size distribution and sample water temperature.

[0033] It should be noted that the target fishpond mentioned above can be a specific body of water used for fish farming, such as a large freshwater or seawater aquaculture pond, the scope and parameters of which can be set according to aquaculture needs. The current water temperature mentioned above can be the real-time temperature value of the fishpond water, such as a Celsius value measured by a temperature sensor, used to reflect the thermal environment of the water. The preset metabolic efficiency model mentioned above can be a neural network-based computational model, such as a model trained using machine learning methods, used to calculate the optimal feed ratio based on the input fish population characteristics and environmental parameters. The sample fish population size distribution mentioned above can be a fish population size dataset used to train the model, such as containing body size parameters of different species and growth stages of fish, used by the model to learn the relationship between fish population size and feed requirements. The sample water temperature mentioned above can be water temperature data corresponding to the sample fish population size distribution, such as water temperature values ​​measured under different seasons and geographical conditions, used by the model to learn the impact of water temperature on fish metabolism. The initial neural network model mentioned above can be an untrained neural network architecture, such as a multilayer perceptron model containing input, hidden, and output layers, whose parameters are optimized through the training process to adapt to specific computational tasks. The above-mentioned fish feed ratio can be the proportion of various nutrients in the feed, such as the percentage of protein, fat, and carbohydrates, and can be adjusted according to the growth needs of the fish and environmental conditions.

[0034] In its implementation, the aforementioned equipment acquires real-time water temperature data through temperature sensors installed around or inside the target fishpond. Simultaneously, the equipment utilizes a three-dimensional monitoring network module to collect information on the body size distribution of the target fish population, such as analyzing fish length and weight characteristics using underwater cameras and image processing algorithms. The acquired water temperature and fish body size distribution data are input into a pre-defined metabolic efficiency model. This model, trained on a large amount of sample data, can calculate the optimal feed ratio based on the input environmental and biological parameters. The model's output is the required nutrient ratio of the feed for the fish population under the current conditions, such as 30% protein and 10% fat. In this way, the equipment can accurately determine the feed ratio based on the real-time environment and fish growth status, meeting the growth needs of the fish population and improving aquaculture efficiency and resource utilization.

[0035] To facilitate understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, in a marine aquaculture scenario, the aforementioned equipment is installed in a large bass rearing pond. The equipment monitors the current water temperature in real time, which is 20 degrees Celsius, using a temperature sensor. Simultaneously, underwater cameras in the 3D monitoring network module capture images of the fish's activity. After image processing and analysis, the current fish size distribution data is obtained, showing that most bass are between 20 and 25 centimeters in length. The equipment inputs this data into a preset metabolic efficiency model, which is trained based on previously collected data on bass size distribution and feed requirements at corresponding water temperatures. After calculation, the model outputs the required feed ratio for the bass under the current conditions, such as 35% protein, 12% fat, and 50% carbohydrates. Based on this ratio, the equipment can control the feeding system to deliver feed according to the calculated nutrient proportions, ensuring that the bass receive sufficient nutrition at a water temperature of 20 degrees Celsius, promoting their healthy growth, while avoiding feed waste and pollution of the aquatic environment.

[0036] Step S30: Generate a target feeding decision based on the current feeding activity of the fish population and the fish feed ratio, and feed the target fish population in the target fish pond according to the target feeding decision.

[0037] It should be noted that the aforementioned current feeding activity level of the fish school can be a quantitative indicator reflecting the current feeding enthusiasm of the fish school, such as a value calculated by comprehensively analyzing characteristics such as the swimming speed, aggregation degree, and feeding behavior frequency of the fish school. The aforementioned feed ratio can be the proportion of various nutrients in the feed determined according to the growth stage of the fish school and environmental conditions, such as the percentage of protein, fat, and carbohydrates. The aforementioned target feeding decision can be a feeding plan generated by the equipment based on feeding activity and feed ratio, including specific plans such as feeding amount, feeding frequency, and feeding time. The aforementioned target fish pond can be the specific aquaculture area where the equipment performs feeding operations, such as a freshwater aquaculture pond or a marine cage. The aforementioned target fish school can be a specific fish group that the equipment feeds, such as salmon or sea bass at a specific growth stage.

[0038] In its implementation, the aforementioned device generates precise target feeding decisions based on the current feeding activity level and feed ratio of the fish population, using a built-in decision-making algorithm. The device comprehensively considers the feeding activity level reflecting the fish's feeding intentions and the nutritional requirements provided by the feed ratio to calculate the optimal feeding amount and timing. For example, when feeding activity is high, the device appropriately increases the feeding amount and frequency to meet the fish's feeding needs; simultaneously, based on the nutrient ratio in the feed, the device controls the feed composition to ensure the fish receive balanced nutrition. Following the generated target feeding decisions, the device precisely feeds the target fish population in the target pond, improving feed utilization efficiency and promoting healthy fish growth.

[0039] To facilitate understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, in a freshwater aquaculture scenario, the equipment monitors the feeding activity of the target fish population at 85 (out of 100), indicating a strong feeding desire. Simultaneously, based on the feed formulation, the equipment determines that the fish population requires feed with a protein content of 35% and a fat content of 12%. Based on this, the equipment generates a target feeding decision, deciding to feed the fish once per hour, with each feeding being 100 kg. Following this decision, the equipment feeds the target fish population in the target pond, ensuring that the fish receive the appropriate amount of nutritionally balanced feed at the right time, thereby improving aquaculture efficiency and resource utilization.

[0040] Furthermore, to achieve more accurate intelligent feeding, the step of determining the fish feed ratio based on the current water temperature and the current fish population size distribution using a preset metabolic efficiency model includes: Step S21: Perform statistical moment processing on the current fish group size distribution to obtain the body size feature vector.

[0041] It should be noted that the aforementioned current fish group size distribution can be a statistical characteristic of the body size of individuals within the fish group, such as the fish group structure divided by body length or weight, reflecting the growth status and structure of the fish group. The aforementioned statistical moment processing can be a mathematical transformation method used to extract key features from the body size distribution data, such as calculating statistical moments like mean, variance, skewness, and kurtosis. The aforementioned body size feature vector can be a multidimensional array used to store and represent the main features of the fish group size distribution, such as including feature values ​​like average body length, standard deviation, and body length skewness.

[0042] In its implementation, the aforementioned device acquires the body size distribution data of the target fish group through a three-dimensional monitoring network module. The device uses image processing algorithms to analyze the activity images of the fish group, identifying the outline and size of each fish, and calculating body size parameters such as length and weight. Then, the device performs statistical moment processing on these body size parameters, calculating statistical characteristics such as mean, variance, skewness, and kurtosis, and combines these feature values ​​into a body size feature vector.

[0043] To facilitate understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, in a freshwater aquaculture scenario, the aforementioned device is installed beside a large aquaculture pond. The device captures images of fish activity using an underwater camera, analyzes them using image processing algorithms to identify the outline of each fish, and calculates the body length of each fish. The body length data collected by the device forms a fish population size distribution, showing that the fish population's body length ranges from 10 to 20 centimeters. The device performs statistical moment processing on this body length data, calculating an average body length of 15 centimeters, a standard deviation of 2 centimeters, a skewness of 0.5, and a kurtosis of 3.2. These statistical characteristics are combined into a body size feature vector, for example, [15, 2, 0.5, 3.2]. This vector reflects the central tendency, dispersion, symmetry, and steepness of the fish population size distribution, providing crucial data support for the device to further assess the feeding activity of the fish population and generate feeding decisions.

[0044] Step S22: Discretize the current water temperature and concatenate the discretization result with the body shape feature vector to obtain the target feature vector.

[0045] It should be noted that the aforementioned current water temperature can be the real-time temperature value of the fishpond water monitored by the equipment, such as the Celsius value collected by a temperature sensor every minute. The aforementioned discretization can be a data quantization method used to convert continuous temperature data into finite discrete values, such as dividing the temperature range into multiple intervals, each interval corresponding to a discrete value. The aforementioned body shape feature vector can be a multi-dimensional array used to store and represent the main body shape characteristics of the fish population, such as including statistical features like average body length, standard deviation of body length, skewness, and kurtosis. The aforementioned concatenation can be a data fusion operation used to combine different types of feature data into a unified feature vector, such as concatenating the discretized temperature values ​​with the body shape feature vector in sequence. The aforementioned target feature vector can be a comprehensive feature representation used to integrate the current water temperature and the fish population's body shape characteristics.

[0046] In its implementation, the device acquires the current water temperature of the target fishpond using a temperature sensor and performs discretization processing on the temperature data, converting continuous temperature values ​​into discrete values ​​within a preset range. Simultaneously, the device retrieves previously calculated fish size feature data from the body shape feature vector. Then, the device concatenates the discretized temperature values ​​with the body shape feature vector to form a target feature vector.

[0047] To facilitate understanding, the following example illustrates the concept, but does not impose specific limitations on this embodiment. For instance, assuming the current water temperature is 22 degrees Celsius, the device discretizes the temperature data, dividing it into 5 intervals, each corresponding to a discrete value. Assuming 22 degrees Celsius falls in the third interval, the corresponding discrete value is 3. Simultaneously, the device obtains a body shape feature vector of [15, 2, 0.5, 3.2], which includes an average body length of 15 cm, a standard deviation of 2 cm, a skewness of 0.5, and a kurtosis of 3.2. Then, the device concatenates the discrete value 3 with the body shape feature vector to obtain the target feature vector [3, 15, 2, 0.5, 3.2]. This target feature vector integrates the current water temperature and the body shape characteristics of the fish population, and can be used as input data for subsequent model calculations, such as inputting it into a preset metabolic efficiency model to calculate a suitable fish feed ratio for the current conditions.

[0048] Step S23: Input the target feature vector into the preset metabolic efficiency model to obtain the fish feed ratio.

[0049] It should be noted that the aforementioned target feature vector can be a multidimensional array integrating discrete values ​​of the current water temperature and fish population characteristics. The aforementioned preset metabolic efficiency model can be a machine learning-based model, such as a neural network model trained on a large amount of sample data, used to calculate a suitable feed ratio for the current conditions based on the input feature vector. The aforementioned feed ratio can be a specific proportion of various nutrients in the feed, such as the percentages of protein, fat, and carbohydrates, used to meet the growth needs of the fish population in a specific environment.

[0050] In its implementation, the device takes the previously processed target feature vector as input and passes it to a preset metabolic efficiency model. The model calculates and analyzes the input feature vector based on its internal algorithm and trained parameters. Through the model's output, the device can obtain the required feed ratio for the fish population under the current conditions, such as 30% protein, 10% fat, and 60% carbohydrates.

[0051] For ease of understanding, the following example illustrates the concept, but does not limit the scope of this embodiment. For instance, suppose the target feature vector is [3, 15, 2, 0.5, 3.2], where 3 represents the discrete water temperature value, and 15, 2, 0.5, and 3.2 represent the fish population's average body length, standard deviation, skewness, and kurtosis, respectively. The device inputs this feature vector into a preset metabolic efficiency model. The model processes the input data according to its internal calculation logic, such as through the weights and activation functions of a multi-layer neural network. Finally, the model outputs a fish feed ratio, for example, 35% protein, 12% fat, and 53% carbohydrates. Based on this ratio, the device controls the feeding system to feed the fish according to the corresponding nutrient proportions, meeting the nutritional needs of the fish at the current water temperature and body size, thereby achieving precise feeding and improving aquaculture efficiency and resource utilization.

[0052] Furthermore, before the step of obtaining the current fish density and current fish size distribution of the target fish population in the target fishpond through the three-dimensional monitoring network module, the method further includes: The initial neural network model was trained based on the sample temperature and the distribution of fish population to obtain the preset metabolic energy efficiency model.

[0053] It should be noted that the sample temperature mentioned above can be the water temperature data of the fishpond used for model training, such as water temperature values ​​measured under different seasons and environmental conditions. The sample fish body size distribution mentioned above can be fish body size data corresponding to the sample temperature, such as the statistical distribution of fish body length, weight, and other characteristics under different temperature conditions. The initial neural network model mentioned above can be an untrained neural network architecture, such as a multilayer perceptron model containing input, hidden, and output layers, whose parameters need to be optimized through the training process. The preset metabolic efficiency model mentioned above can be a trained neural network model used to calculate the fish feed ratio based on the input temperature and fish body size distribution data.

[0054] In its implementation, the aforementioned device collects a large amount of sample temperature and fish population size distribution data to train an initial neural network model using supervised learning. The device uses this sample data as input and the corresponding fish feed ratio as the output label, continuously adjusting the model's parameters using a backpropagation algorithm until the model can accurately predict the fish feed ratio. After training, the device obtains a preset metabolic efficiency model, which can quickly calculate the appropriate fish feed ratio based on real-time temperature and fish population size distribution data, providing support for precise feeding.

[0055] For ease of understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, suppose the device collects 1000 sets of sample data. Each set includes the water temperature of the fishpond and the corresponding fish size distribution characteristics, such as average body length and weight. Each sample also has a corresponding feed ratio label, indicating the optimal nutrient ratio of the feed for the fish under those conditions. The device inputs this sample data into an initial neural network model and continuously optimizes the model's weights and bias parameters through multiple rounds of iterative training. Ultimately, the model learns the relationship between temperature, body size distribution, and feed ratio, forming a preset metabolic efficiency model. When the device acquires current water temperature and fish size distribution data in an actual aquaculture scenario, it can use this model to quickly calculate a suitable feed ratio, thereby achieving precise feed delivery.

[0056] In this embodiment, during intelligent feeding, the fish density and body size distribution are acquired through a three-dimensional monitoring network module. Combined with a preset metabolic efficiency model, the feed ratio is determined by comprehensively considering water temperature and fish body size distribution. Then, based on the fish's feeding activity and the feed ratio, a precise feeding decision is generated, achieving precise feeding. This solves the problem of existing methods that adjust feeding based solely on a single water temperature parameter, ignoring the complexity of fish behavior and environmental conditions. It effectively avoids low aquaculture efficiency caused by overfeeding or underfeeding, thereby improving aquaculture efficiency.

[0057] Based on the first embodiment, in the second embodiment, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 , Figure 2 This is a flowchart of the second embodiment of the intelligent feeding method proposed in this application. Further, to achieve more accurate intelligent feeding, the step of generating a target feeding decision based on the current feeding activity of the fish population and the fish feed ratio includes: Step S31: Determine the primary decision vector based on the current feeding activity of the fish swarm and the fish food ratio through a preset feeding decision network.

[0058] It should be noted that the aforementioned pre-set feeding decision network can be a machine learning-based model, such as a neural network model trained on a large amount of sample data, used to generate a preliminary decision vector based on the input fish feeding activity level and feed ratio. The aforementioned current fish feeding activity level can be a quantitative indicator reflecting the current feeding enthusiasm of the fish, such as a value calculated by analyzing characteristics such as the fish's swimming speed, aggregation degree, and feeding behavior frequency. The aforementioned feed ratio can be the proportion of various nutrients in the feed determined according to the fish's growth stage and environmental conditions, such as the percentage of protein, fat, and carbohydrates. The aforementioned preliminary decision vector can be a multi-dimensional array used to represent the initial feeding decision results, including parameters such as feeding amount, feeding frequency, and feeding time.

[0059] In its implementation, the device takes the current feeding activity level of the fish and the feed ratio as input data and passes them to a pre-set feeding decision network. The network calculates and analyzes the input data based on its internal algorithm and trained parameters, generating a preliminary decision vector. This vector contains preliminary feeding decision information, such as a suggested feeding amount of 100 kg per hour, a feeding frequency of once per hour, and a feeding time from 8:00 AM to 4:00 PM.

[0060] For ease of understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, assume the current feeding activity level of the fish population is 85 (out of 100), and the feed ratio is 35% protein, 10% fat, and 55% carbohydrates. The device inputs this data into a preset feeding decision network. The network processes the input data according to its internal computational logic, such as through the weights and activation functions of a multi-layer neural network. Finally, the network outputs a preliminary decision vector, such as [100, 1, 8-16], suggesting a feeding frequency of 100 kg of feed per hour, once per hour, from 8:00 AM to 4:00 PM.

[0061] Step S32: Obtain the species constant corresponding to the target fish population and the effective water volume corresponding to the target fish pond.

[0062] It should be noted that the target fish population mentioned above can be a specific fish group monitored and fed by the equipment, such as a certain species of salmon or sea bass. The species constant mentioned above can be a physiological characteristic coefficient related to that fish species, such as a value reflecting its metabolic characteristics and feeding behavior, used in the model to calculate the saturation risk or other relevant indicators of the fish population. The target fish pond mentioned above can be a specific body of water used for fish farming, such as a large freshwater or marine aquaculture pond. The effective water volume mentioned above can be the actual usable water volume in the fish pond, such as the water volume after deducting the space occupied by obstacles, equipment, etc., used to more accurately calculate the fish density and feeding amount.

[0063] In its implementation, the device obtains the species constants corresponding to the target fish population by accessing a pre-stored fish species database, which contains physiological characteristic data of different fish species. Simultaneously, the device calculates the total volume of the fishpond using its dimensional parameters (e.g., length, width, and height) and adjusts these parameters based on the space occupied by equipment and obstacles within the pond to obtain the effective water volume.

[0064] For ease of understanding, the following example is provided, but it does not limit the scope of this embodiment. For example, assume the target fish population is Atlantic salmon, and the device obtains its species constant from the database as 0.02 (this constant reflects the metabolic characteristics of salmon at a specific temperature). The target fish pond is 10 meters long, 5 meters wide, and 3 meters deep, with a total volume of 150 cubic meters. Feeding and monitoring equipment occupy approximately 15 cubic meters of space within the pond. The device calculates the effective water volume to be 135 cubic meters.

[0065] Step S33: Determine the average body length of the target fish group based on the current fish group size distribution, and obtain the satiation coefficient based on the species constant, the effective water volume, and the average body length.

[0066] It should be noted that the aforementioned current fish population size distribution can be a statistical characteristic of the individual body size within the fish population, such as a fish population structure divided by body length or weight, reflecting the growth status and distribution of the fish population. The aforementioned target fish population can be a specific fish population monitored and fed by the equipment, such as a particular species of salmon or bass. The aforementioned average body length can be the average body length of all individuals in the fish population size distribution, for example, the average value obtained by calculating the body length of each fish in the population. The aforementioned species constant can be a physiological characteristic coefficient related to that fish species, such as a value reflecting its metabolic characteristics and feeding behavior, used to calculate the fish population's satiety coefficient. The aforementioned effective water volume can be the actual usable water volume in the fishpond, for example, the water volume after deducting the space occupied by obstacles, equipment, etc. The aforementioned satiety coefficient can be an indicator measuring the degree of feeding saturation in the fish population, used to assess the feeding status of the fish population and guide feeding decisions.

[0067] In its implementation, the aforementioned device calculates the average body length of the target fish group based on the current fish group size distribution. The device acquires the fish group's size data through a three-dimensional monitoring network module, calculates the body length of each fish, and then calculates the average body length. Next, the device uses pre-stored species constants, the calculated average body length, and previously acquired effective water volume to calculate the saturation coefficient using a preset saturation coefficient calculation formula. The preset saturation coefficient calculation formula is: ; in, The satiety coefficient is denoted as , where Where N is a species constant and N is the fish population size. For average body length, The effective water volume.

[0068] To facilitate understanding, the following example is provided, but it does not limit the scope of this embodiment. For example, suppose the target fish population is a type of freshwater fish. Data obtained by the device from the 3D monitoring network module shows that the fish's body length ranges from 10 to 20 centimeters, and the calculated average body length is 15 centimeters. The device obtains the species constant of this fish species from the database as 0.018. The previously calculated effective water volume is 100 cubic meters. Substituting the values ​​into the formula, the calculation yields... By calculating the satiety coefficient S, the equipment adjusts the feeding amount based on this coefficient. When S exceeds a certain threshold, the feeding amount is reduced to avoid overfeeding, ensuring the healthy growth of the fish and improving aquaculture efficiency. Step S34: Generate a target feeding decision based on the satiety coefficient for the primary decision vector.

[0069] It should be noted that the aforementioned satiety coefficient can be an indicator measuring the feeding saturation level of a fish population, such as a value calculated using a specific formula, used to assess the feeding status of the fish population. The aforementioned primary decision vector can be preliminary feeding decision data generated by the device based on the current feeding activity level and feed ratio of the fish population, such as a multi-dimensional array containing parameters such as feeding amount, feeding frequency, and feeding time. The aforementioned target feeding decision can be the final feeding decision after adjustment and optimization, such as a feeding plan obtained by correcting the primary decision vector based on the satiety coefficient, used to guide actual feeding operations.

[0070] In its implementation, the aforementioned device adjusts the primary decision vector based on the calculated satiety coefficient to generate a target feeding decision. The device first determines the current satiety coefficient, which reflects the fish's current feeding status. If the satiety coefficient is high, indicating the fish are nearing saturation, the device will correspondingly reduce the feeding amount in the primary decision vector or extend the feeding interval. Conversely, if the satiety coefficient is low, indicating strong feeding demand, the device may appropriately increase the feeding amount or shorten the feeding interval. In this way, the device can generate more accurate target feeding decisions, ensuring the feeding amount matches the actual needs of the fish, optimizing aquaculture results, and reducing feed waste.

[0071] For ease of understanding, the following example illustrates the concept, but does not impose specific limitations on this embodiment. Assume the device calculates the current fish population's satiety coefficient to be 0.7, indicating the fish have consumed a significant amount of food and are nearing saturation. The initial decision vector suggests a feeding rate of 100 kg per hour and a feeding frequency of once per hour. Based on the satiety coefficient, the device adjusts the initial decision vector, reducing the feeding rate to 70 kg per hour and adjusting the feeding frequency to once every two hours. This generated target feeding decision satisfies the fish population's feeding needs while avoiding overfeeding, contributing to the healthy growth of the fish and the stability of the aquaculture environment.

[0072] Furthermore, the three-dimensional monitoring network includes a camera unit and a sonar unit. The step of obtaining the current fish density and current fish size distribution in the target fishpond through the three-dimensional monitoring network module includes: Step S11: Obtain images of fish in the target fishpond through the camera unit, and perform network segmentation on the fish images to obtain the current fish size distribution in the target fishpond; Step S12: Obtain sonar images of the target fishpond through the sonar unit, and detect the sonar images to obtain the current fish density in the target fishpond.

[0073] It should be noted that the aforementioned camera unit can be an underwater camera device installed around or above the target fishpond, such as a high-definition underwater camera, used to capture images of fish activity. The aforementioned fishpond image can be image data showing the activity of the fish, captured by the camera unit, such as video frames containing information such as the fish's location, swimming state, and body size. The aforementioned network segmentation can be an image processing technique, such as using a neural network algorithm to segment an image to identify and distinguish different objects or regions within it. The aforementioned sonar unit can be a sonar sensor installed in the fishpond, such as a device for transmitting and receiving sound wave signals to detect the location and density of the fish. The aforementioned sonar image can be image data generated by the sonar unit based on sound wave reflection signals, such as a sonar scan showing the distribution of fish in the fishpond. The aforementioned current fish group size distribution can be a statistical distribution of individual fish size characteristics obtained through network segmentation of the fishpond image, such as the proportion of fish of different sizes. The aforementioned current fish group density can be the number of fish per unit volume of water, detected by sonar imagery, such as the number of fish per cubic meter of water.

[0074] In its implementation, the device acquires images of fish in the target fishpond using a camera unit and processes these images using network segmentation technology to determine the current fish population size distribution. Simultaneously, the device uses a sonar unit to acquire sonar images of the target fishpond and performs detection processing to determine the current fish density. These two processes complement each other, providing the device with detailed information about the fish population within the target fishpond.

[0075] To facilitate understanding, the following example illustrates the concept but does not limit the scope of this embodiment. For instance, suppose in a freshwater aquaculture scenario, the device's camera unit captures a video of a school of fish swimming in a pond. By performing network segmentation on each frame of this video, the device can identify the size of each fish and count the number of fish of different sizes, thus obtaining the current fish school size distribution. Simultaneously, the sonar unit emits sound wave signals and receives the reflected signals to generate sonar images. The device analyzes the sonar images to calculate the number of fish per cubic meter of water, thereby obtaining the current fish school density.

[0076] Based on the first and second embodiments, in the third embodiment, the content that is the same as or similar to that in Embodiments 1 and 2 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 , Figure 3 This is a flowchart of the third embodiment of the intelligent feeding method proposed in this application. Further, the step of performing network segmentation on the fish swarm image to obtain the current fish morphology distribution within the target fishpond includes: Step S111: Perform image segmentation on the fish group image based on network segmentation technology to obtain the contour information of each member of the target fish group; Step S112: Extract features based on the contour information to obtain the body shape features of each member of the fish group; Step S113: Obtain the current fish size distribution in the target fish pond based on the described body size characteristics.

[0077] It should be noted that the aforementioned network segmentation technology can be a deep learning-based image segmentation method, such as using a convolutional neural network (CNN) to perform pixel-level classification of images to identify and distinguish different objects or regions in the image. The aforementioned fish school image can be image data showing the activity of the fish school acquired by a camera unit, such as video frames containing information such as the fish's position, swimming state, and body shape. The aforementioned contour information can be the boundary line data of the fish school members in the image, such as the outline of each fish identified by an image processing algorithm, used to describe the shape and size of the fish. The aforementioned feature extraction can be the process of obtaining the fish's body shape features from the contour information, such as calculating parameters such as the fish's body length, body width, and area. The aforementioned body shape features can be specific numerical values ​​describing the fish's shape and size, such as body length, body width, and body surface area, used to characterize the fish's body shape. The aforementioned current fish school body shape distribution can be the proportion of individuals of different body shapes in the fish school, such as the fish school structure divided by body length or weight, reflecting the growth status and distribution of the fish school.

[0078] In its implementation, the aforementioned device uses network segmentation technology to segment images of fish schools, accurately identifying the contour information of each member of the target fish school. The device uses a convolutional neural network to classify each pixel in the image, distinguishing between the fish body and the background, thereby obtaining the contour of each fish. Next, the device extracts features based on this contour information, calculating the body shape characteristics of each fish, such as body length, body width, and area. Finally, the device statistically analyzes these body shape characteristics to generate current fish school body shape distribution data within the target fish pond.

[0079] To facilitate understanding, a specific example is used below, but this embodiment is not intended to be limiting. Assume a freshwater aquaculture scenario where a device acquires a frame of an image of a school of fish. Using network segmentation technology, the device accurately identifies each fish in the image and extracts their contour information. Then, the device analyzes these contours to calculate the body length and width of each fish. For example, out of 100 identified fish, 30 are 10-12 cm long, 40 are 12-14 cm long, and 30 are 14-16 cm long. The device statistically analyzes this data to generate the current fish group size distribution: 30% are 10-12 cm, 40% are 12-14 cm, and 30% are 14-16 cm.

[0080] Furthermore, the step of detecting the sonar image to obtain the current fish density in the target fishpond includes: Based on a preset target detection algorithm, feature extraction is performed on the sonar image to obtain the reflective pixels in the sonar image; Target reflection pixels are selected from each of the reflection pixels according to a preset selection rule, and the number of members in the target fish group is determined based on the target reflection pixels. Obtain the area of ​​the target fish pond, and determine the current fish density in the target fish pond based on the area of ​​the fish pond and the number of members.

[0081] It should be noted that the aforementioned preset target detection algorithm can be an algorithm used to identify and locate reflective pixels in sonar images, such as threshold-based segmentation or machine learning-based detection algorithms, used to extract features from sonar images. The aforementioned sonar images can be image data generated by sonar units based on sound wave reflection signals, such as a sonar scan map showing the distribution of fish in a fishpond. The aforementioned feature extraction can be the process of obtaining reflective pixels from sonar images, for example, identifying pixels with significant reflective properties in the image through algorithms.

[0082] The aforementioned reflective pixels can be bright pixels in sonar images formed by the reflection of sound waves from objects such as fish, used to indicate the location of the fish school. The aforementioned preset selection rules can be rules used to filter out target reflective pixels from the reflective pixels, such as based on pixel intensity, distribution density, or correlation with other pixels. The aforementioned target reflective pixels can be reflective pixels that, after filtering, are considered to belong to the fish school, used to calculate the number of fish school members. The aforementioned number of members can be the number of fish in the target fish school, for example, by counting the number of target reflective pixels and converting it into the number of fish. The aforementioned pond area can be the water area of ​​the target fish pond, for example, the area calculated by measuring the length and width of the pond. The aforementioned fish school density can be the number of fish per unit area of ​​the pond, for example, the number of fish per square meter of pond.

[0083] In its implementation, the device extracts features from sonar images using a preset target detection algorithm to identify reflective pixels. The device then selects target reflective pixels from these pixels according to preset selection rules and determines the number of members in the target fish population through counting and conversion. Finally, the device obtains the area of ​​the target fishpond and calculates the current fish population density based on the number of members.

[0084] For ease of understanding, the following example illustrates the concept, but does not limit the scope of this embodiment. For instance, the device acquires a sonar image and, using a preset target detection algorithm, identifies pixels with high reflectivity in the sonar image as reflective pixels. The device then selects target reflective pixels that match the reflectivity characteristics of fish bodies according to preset selection rules. Assuming the device selects 100 target reflective pixels and calculates the number of fish in the target fish population to be 50 based on the area represented by each pixel, and the area of ​​the target fishpond is measured to be 100 square meters, the device calculates the current fish density to be 0.5 fish per square meter.

[0085] This embodiment also provides a first embodiment of an intelligent feeding device; please refer to [reference needed]. Figure 4 , Figure 4 This is a diagram of an intelligent feeding device provided in an embodiment of this application. The intelligent feeding device includes: The activity acquisition module is used to acquire the current fish density and current fish size distribution of the target fish group in the target fish pond through the three-dimensional monitoring network module, and determine the current feeding activity of the fish group based on the current fish density and the current fish size distribution. The fish feed ratio module is used to obtain the current water temperature of the target fish pond and determine the fish feed ratio based on the current water temperature and the current fish population size distribution through a preset metabolic energy efficiency model. The preset metabolic energy efficiency model is obtained by training an initial neural network model with the sample fish population size distribution and sample water temperature. The intelligent feeding module is used to generate a target feeding decision based on the current feeding activity of the fish population and the fish feed ratio, and to feed the target fish population in the target fish pond according to the target feeding decision; The fish feed formulation module is also used to perform statistical moment processing on the current fish group size distribution to obtain a body size feature vector; to perform discrete processing on the current water temperature, and to concatenate the discrete processing result and the body size feature vector to obtain a target feature vector; and to input the target feature vector into the preset metabolic efficiency model to obtain the fish feed formulation.

[0086] Referring to the first embodiment of the intelligent feeding device, this embodiment also proposes a second embodiment of the intelligent feeding device. The contents that are the same as or similar to those in the first embodiment of the intelligent feeding device can be referred to the above description, and will not be repeated hereafter.

[0087] The intelligent feeding module is further configured to: determine a primary decision vector based on the current feeding activity of the fish population and the fish feed ratio using a preset feeding decision network; obtain the species constant corresponding to the target fish population and the effective water volume corresponding to the target fish pond; determine the average body length corresponding to the target fish population based on the current fish population size distribution, and obtain a satiety coefficient based on the species constant, the effective water volume, and the average body length; and generate a target feeding decision based on the satiety coefficient and the primary decision vector. The activity acquisition module is also used to acquire images of fish in the target fish pond through the camera unit, and to perform network segmentation on the fish images to obtain the current fish size distribution in the target fish pond; and to acquire sonar images of the target fish pond through the sonar unit, and to detect the sonar images to obtain the current fish density in the target fish pond.

[0088] Referring to the first and second embodiments of the intelligent feeding device, this embodiment also proposes a third embodiment of the intelligent feeding device. The contents that are the same as or similar to the first and second embodiments of the intelligent feeding device can be referred to the above description, and will not be repeated hereafter.

[0089] The activity acquisition module is further configured to perform image segmentation on the fish group image based on network segmentation technology to obtain the contour information of each fish group member in the target fish group; perform feature extraction based on the contour information to obtain the body shape features of each fish group member; and obtain the current fish group body shape distribution in the target fish pond based on the body shape features. The activity acquisition module is further configured to extract features from the sonar image based on a preset target detection algorithm to obtain reflective pixels in the sonar image; select target reflective pixels from each of the reflective pixels according to a preset selection rule, and determine the number of members in the target fish group based on the target reflective pixels; obtain the area of ​​the fish pond corresponding to the target fish pond, and determine the current fish density in the target fish pond based on the fish pond area and the number of members.

[0090] The intelligent feeding device provided in this embodiment, employing the intelligent feeding method described in the above embodiments, can solve the technical problem of how to perform feeding. Compared with the prior art, the beneficial effects of the intelligent feeding device provided in this embodiment are the same as those of the intelligent feeding method provided in the above embodiments, and other technical features in the intelligent feeding device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0091] This embodiment provides an intelligent feeding device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the intelligent feeding method in the first embodiment described above.

[0092] The following is for reference. Figure 5 , Figure 5 This is a schematic diagram of the structure of an intelligent feeding device suitable for implementing the embodiments of this application. The intelligent feeding device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (such as vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The intelligent feeding device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0093] like Figure 5 As shown, the intelligent feeding device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the intelligent feeding device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the smart feeding device to communicate wirelessly or wiredly with other devices to exchange data. Although the figures show smart feeding devices with various systems, it should be understood that implementing or having all of the systems shown is not required. More or fewer systems may be implemented alternatively.

[0094] Specifically, according to this embodiment, the process described above with reference to the flowchart can be implemented as a computer software program. For example, this embodiment includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the disclosed embodiments of this embodiment.

[0095] The intelligent feeding device provided in this embodiment, employing the intelligent feeding method described in the above embodiments, can solve the technical problem of how to perform feeding. Compared with the prior art, the beneficial effects of the intelligent feeding device provided in this embodiment are the same as those of the intelligent feeding method provided in the above embodiments, and other technical features of this intelligent feeding device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0096] It should be understood that the various parts disclosed in this embodiment can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0097] The above description is merely a specific implementation of this embodiment, but the protection scope of this embodiment is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this embodiment should be included within the protection scope of this embodiment. Therefore, the protection scope of this embodiment should be determined by the protection scope of the claims.

[0098] This embodiment provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the intelligent feeding method in the above embodiment.

[0099] The computer-readable storage medium provided in this embodiment may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0100] The aforementioned computer-readable storage medium may be included in the intelligent feeding device; or it may exist independently and not be assembled into the intelligent feeding device.

[0101] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the intelligent feeding device, cause the intelligent feeding device to perform intelligent feeding.

[0102] Computer program code for performing the operations of this embodiment can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0103] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this embodiment. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0104] The modules described in this embodiment can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0105] The readable storage medium provided in this embodiment is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-described intelligent feeding method, and can solve the technical problem of how to perform feeding. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this embodiment are the same as the beneficial effects of the intelligent feeding method provided in the above embodiments, and will not be repeated here.

[0106] The above descriptions are only some embodiments and do not limit the patent scope of this embodiment. All equivalent structural transformations made based on the technical concept of this application and the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included within the patent protection scope of this application.

Claims

1. An intelligent feeding method, characterized in that, The method is applied to an intelligent feeding device provided with a three-dimensional monitoring network module, and the method comprises the following steps: A current fish school density and a current fish school body type distribution of a target fish school in a target fish pond are acquired through the three-dimensional monitoring network module, and a current fish school feeding activity is determined based on the current fish school density and the current fish school body type distribution; A current water area temperature of the target fish pond is acquired, and a fish food ratio is determined according to the current water area temperature and the current fish school body type distribution through a preset metabolic energy efficiency model, wherein the preset metabolic energy efficiency model is obtained by training an initial neural network model through sample fish school body type distribution and sample water area temperature; A target feeding decision is generated based on the current fish school feeding activity and the fish food ratio, and the target fish school is fed in the target fish pond according to the target feeding decision.

2. The method of claim 1, wherein, The step of determining the fish food ratio according to the current water area temperature and the current fish school body type distribution through the preset metabolic energy efficiency model comprises the following steps: The current fish school body type distribution is subjected to statistical moment processing to obtain a body type feature vector; The current water area temperature is subjected to discrete processing, and the discrete processing result and the body type feature vector are spliced to obtain a target feature vector; The target feature vector is input into the preset metabolic energy efficiency model to obtain the fish food ratio.

3. The method of claim 1, wherein, The step of generating the target feeding decision based on the current fish school feeding activity and the fish food ratio comprises the following steps: A primary decision vector is determined based on the current fish school feeding activity and the fish food ratio through a preset feeding decision network; A breed constant corresponding to the target fish school and an effective water volume corresponding to the target fish pond are acquired; An average body length corresponding to the target fish school is determined according to the current fish school body type distribution, and a satiation coefficient is obtained according to the breed constant, the effective water volume and the average body length; The target feeding decision is generated based on the satiation coefficient and the primary decision vector.

4. The method of claim 1, wherein, The three-dimensional monitoring network comprises a camera unit and a sonar unit, and the step of acquiring the current fish school density and the current fish school body type distribution in the target fish pond through the three-dimensional monitoring network module comprises the following steps: A fish school image in the target fish pond is acquired through the camera unit, and the fish school image is subjected to network segmentation to obtain the current fish school body type distribution in the target fish pond; A sonar image in the target fish pond is acquired through the sonar unit, and the sonar image is subjected to detection to obtain the current fish school density in the target fish pond.

5. The method of claim 4, wherein, The step of subjecting the fish school image to network segmentation to obtain the current fish school body type distribution in the target fish pond comprises the following steps: The fish school image is subjected to image segmentation based on a network segmentation technology to obtain contour information of each fish school member in the target fish school; Body type features of each fish school member are obtained through feature extraction based on the contour information; The current fish school body type distribution in the target fish pond is obtained based on the body type features.

6. The method of claim 4, wherein, The step of subjecting the sonar image to detection to obtain the current fish school density in the target fish pond comprises the following steps: The sonar image is subjected to feature extraction based on a preset target detection algorithm to obtain reflected pixel points in the sonar image; A target reflected pixel point is selected from the reflected pixel points according to a preset selection rule, and a member quantity of fish group members in the target fish group is determined according to the target reflected pixel point; A fish pool area corresponding to the target fish pool is obtained, and a current fish group density in the target fish pool is determined according to the fish pool area and the member quantity.

7. An intelligent feeding device, characterized in that, The device comprises: An activity acquisition module is configured to acquire a current fish group density and a current fish group body type distribution of a target fish group in a target fish pool through a stereoscopic monitoring network module, and determine a current fish group feeding activity based on the current fish group density and the current fish group body type distribution; A fish food ratio module is configured to acquire a current water area temperature of the target fish pool, and determine a fish food ratio based on the current water area temperature and the current fish group body type distribution through a preset metabolic energy efficiency model, which is obtained by training an initial neural network model based on sample fish group body type distributions and sample water area temperatures; An intelligent feeding module is configured to generate a target feeding decision based on the current fish group feeding activity and the fish food ratio, and feed a target fish group in the target fish pool according to the target feeding decision.

8. An intelligent feeding device, characterized in that, The device comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, which is configured to implement the steps of the intelligent feeding method according to any one of claims 1 to 6.

9. A storage medium, characterized by The storage medium is a computer readable storage medium, and the storage medium stores a computer program, which is executed by the processor to implement the steps of the intelligent feeding method according to any one of claims 1 to 6.

10. A computer program product, characterised in that, The computer program product comprises a computer program, which is executed by the processor to implement the steps of the intelligent feeding method according to any one of claims 1 to 6.

Images