Fish disease monitoring method and system based on feature image recognition

Abstract

The invention relates to the field of data processing, in particular to a fish disease monitoring method and system based on feature image recognition. The method comprises the following steps: monitoring water quality change conditions in a target area in real time through a water quality sensor; acquiring fish image information in the target area; performing behavior identification on the fish image information to obtain fish behavior information of each fish object in the target area; performing lesion recognition on the fish behavior information, the fish image information and the water quality change condition through a fish lesion recognition model to obtain potential lesion information of each fish object; and generating lesion risk information of each fish object according to the potential lesion information, and displaying the lesion risk information in a visual model corresponding to the target water area. According to the method, individualized potential fish lesion recognition can be achieved for different types of fishes, the accuracy of fish lesion monitoring is improved, the fish lesion monitoring process is simplified, and the fish lesion monitoring efficiency is further improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular, to a fish disease monitoring method and system based on feature image recognition. Background Art

[0002] Fish live in water, and the water body provides them with a certain degree of concealment, which makes it difficult for monitoring personnel to directly observe the activities and health conditions of fish. Especially for fish in some large water bodies such as lakes and oceans, the monitoring difficulty is even greater. Moreover, the behaviors of fish in water are complex and diverse, and the behavioral habits of different species of fish vary greatly. Some fish may be nocturnal or prefer to inhabit hard-to-observe positions such as the bottom of the water, increasing the difficulty of comprehensively and accurately observing their behaviors, and abnormal behaviors are often one of the important signals of disease occurrence.

[0003] Therefore, there is an urgent need for a brand-new solution to overcome the above problems and achieve automated monitoring of fish diseases. Summary of the Invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a fish disease monitoring method and system based on feature image recognition, which is used to realize individualized identification of potential fish lesions for different types of fish, improve the accuracy of fish lesion monitoring, simplify the fish lesion monitoring process, and further improve the monitoring efficiency of fish lesions.

[0005] In a first aspect, an embodiment of the present application provides a fish disease monitoring method based on feature image recognition, including: Real-time monitoring of water quality changes in the target area through a water quality sensor; the water quality changes include the following parameters: chlorophyll concentration, dissolved oxygen, temperature, pH value; Collecting fish image information in the target area; Performing behavior recognition on the fish image information to obtain fish behavior information of each fish object in the target area; Through a fish lesion recognition model, performing lesion recognition on the fish behavior information, the fish image information, and the water quality changes to obtain potential lesion information of each fish object; Generating lesion risk information for each fish object according to the potential lesion information and displaying it in the visualization model corresponding to the target water area.

[0006] In a second aspect, an embodiment of the present application provides a fish disease monitoring system based on feature image recognition, and the system includes the following units: A monitoring unit for real-time monitoring of water quality changes in a target area through a water quality sensor; the water quality changes include the following parameters: chlorophyll concentration, dissolved oxygen, temperature, pH value; and collecting fish image information in the target area. An identification unit for performing behavior identification on the fish image information to obtain fish behavior information of each fish object in the target area; and performing lesion identification on the fish behavior information, the fish image information, and the water quality changes through a fish lesion identification model to obtain potential lesion information of each fish object. A display unit for generating lesion risk information of each fish object according to the potential lesion information and displaying it in the visualization model corresponding to the target water area.

[0007] In a third aspect, an embodiment of the present application provides an electronic device, which includes: At least one processor, a memory, and an input / output unit; Wherein, the memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the fish disease monitoring method based on feature image recognition in the first aspect.

[0008] In a fourth aspect, a computer-readable storage medium is provided, which includes instructions that, when the instructions are run on a computer, cause the computer to execute the fish disease monitoring method based on feature image recognition in the first aspect.

[0009] The beneficial effects of the present invention are: providing a fish disease monitoring method and system based on feature image recognition. In this technical solution, the water quality changes in the target area are real-time monitored through a water quality sensor; the water quality changes include the following parameters: chlorophyll concentration, dissolved oxygen, temperature, pH value; the fish image information in the target area is collected; the fish image information is subjected to behavior identification to obtain fish behavior information of each fish object in the target area; through a fish lesion identification model, the fish behavior information, the fish image information, and the water quality changes are subjected to lesion identification to obtain potential lesion information of each fish object; and lesion risk information of each fish object is generated according to the potential lesion information and displayed in the visualization model corresponding to the target water area. This technical solution can achieve individualized potential fish lesion identification for different types of fish, improve the accuracy of fish lesion monitoring, simplify the fish lesion monitoring process, and further improve the monitoring efficiency of fish lesions. Description of the Drawings

[0010] Figure 1 is a schematic flowchart of a fish disease monitoring method based on feature image recognition according to an embodiment of the present application; Figure 2It is a schematic structural diagram of a fish disease monitoring system based on feature image recognition according to an embodiment of the present application; Figure 3 It is a schematic structural diagram of an electronic device according to an embodiment of the present application; Figure 4 It is a schematic structural diagram of a medium device according to an embodiment of the present application. Specific embodiments

[0011] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0012] The embodiments of the present application provide a fish disease monitoring method and system based on feature image recognition.

[0013] In the embodiments of the present application, initial image data to be processed is obtained; the initial image data includes target objects to be analyzed in an industrial monitoring scenario; the initial image data is three-dimensionally stitched to obtain a stitched image corresponding to the target object; virtual modeling is performed based on the stitched image to obtain an initial virtual model of the target object; the virtual initial model is a virtual three-dimensional model of the target object in the stitched image space; the initial virtual model is optimized in spatial form through an adaptive optimization model to obtain an optimized virtual model; the optimized virtual model is a virtual three-dimensional model of the target object in the image optimization space; three-dimensional point cloud data of the target object is extracted from the optimized virtual model to complete the intelligent extraction of three-dimensional point cloud data.

[0014] In the embodiments of the present application, firstly, the water quality sensor is used to monitor the water quality change situation in real time, and the dynamic information of the water quality in the target area can be obtained in time, such as the change of parameters such as chlorophyll concentration, dissolved oxygen, temperature, pH value, etc. At the same time, by collecting fish image information and performing behavior recognition, the behavior state of fish can also be grasped in real time. This real-time monitoring can detect abnormalities at the early stage of disease occurrence, providing the possibility for taking prevention and control measures in time and avoiding the further spread and diffusion of diseases. By using a fish lesion recognition model to comprehensively analyze and recognize fish behavior information, fish image information, and water quality change situation, compared with the traditional method that relies solely on naked-eye observation or partial parameter judgment, it can more comprehensively and accurately identify the potential lesion information of fish. Multi-dimensional data fusion analysis can improve the accuracy of recognition and reduce the possibility of misjudgment and missed judgment.

[0015] Second, it not only focuses on the images and behavioral information of fish themselves, but also includes water quality changes in the monitoring scope. Water quality is an important factor affecting fish health, and the occurrence of many diseases is closely related to water quality changes. By comprehensively considering these factors, it is possible to more comprehensively understand the environment and health status of fish, and judge whether there is a risk of fish diseases from multiple perspectives. It can perform individual lesion identification and risk assessment for each fish object in the target area, and obtain the potential lesion information and lesion risk information of each fish object. This individual-level analysis can more accurately locate problems, help target the treatment and management of diseased fish, and at the same time better monitor the health status of the entire fish population.

[0016] Third, the lesion risk information of each fish object generated based on the potential lesion information will be displayed in the visualization model corresponding to the target water area, and presented to the user in an intuitive way. Users can clearly understand the distribution of the health status of the fish population through the visualization interface, quickly locate the areas and individuals with lesion risks, and facilitate taking corresponding measures in a timely manner. It provides a scientific decision-making basis for breeders or relevant management personnel. They can formulate reasonable breeding strategies according to the visualized lesion risk information, such as adjusting water quality, optimizing feed feeding, and isolating diseased fish in a timely manner, so as to effectively prevent and control the occurrence and spread of fish diseases, improve breeding efficiency, and reduce economic losses.

[0017] Fourth, through real-time monitoring and comprehensive analysis, it is possible to detect potential lesion risks before fish diseases are significantly manifested, realizing early warning. This enables breeders to take preventive measures in advance, such as strengthening water quality management and increasing fish immunity, to control diseases in the budding state and reduce the likelihood and harm of disease outbreaks. The large amount of monitoring data accumulated by the system can be used to analyze the changing trends of fish health status and the occurrence patterns of diseases. By mining and analyzing these data, it can help breeders better understand the occurrence mechanisms and influencing factors of fish diseases, so as to formulate more forward-looking disease prevention and control plans, optimize the breeding environment and management measures, and improve the overall health level of fish.

[0018] Fifth, when collecting fish image information and monitoring water quality changes, there is no need to directly contact the fish, avoiding stress and harm to the fish caused by operations such as catching or sampling. This non-contact monitoring method helps to maintain the natural living state of the fish, reduce changes in the health status of the fish caused by human interference, and improve the authenticity and reliability of the monitoring results. Compared with some traditional disease detection methods (such as anatomical examination), this system will not cause damage to the fish, will not affect the growth and development of the fish, and will not damage the overall ecological environment of the fish population. This is of great significance for protecting fish resources and maintaining the balance of the breeding ecosystem.

[0019] In summary, in the embodiments of the present application, individualized potential fish lesion recognition is achieved for different types of fish, improving the accuracy of fish lesion monitoring, simplifying the fish lesion monitoring process, and further enhancing the monitoring efficiency of fish lesions. The fish disease monitoring solution based on feature image recognition provided by the embodiments of the present application can also be executed by an electronic device, which can be a server, a server cluster, or a cloud server. The electronic device can also be a terminal device such as a mobile phone, a computer, a tablet computer, a wearable device, or a dedicated device (such as a dedicated terminal device with a fish disease monitoring system based on feature image recognition). These electronic devices can also be equipped with the chips introduced in the above embodiments. Alternatively, these electronic devices can also install a service program for executing the fish disease monitoring solution based on feature image recognition.

[0020] Figure 1 Schematic diagram of a fish disease monitoring method based on feature image recognition provided by the embodiments of the present application, as Figure 1 shown, the method includes the following steps: 101. Real-time monitor the water quality changes in the target area through a water quality sensor; 102. Collect fish image information in the target area; 103. Perform behavior recognition on the fish image information to obtain the fish behavior information of each fish object in the target area; 104. Through a fish lesion recognition model, perform lesion recognition on the fish behavior information, the fish image information, and the water quality changes to obtain the potential lesion information of each fish object; 105. Generate lesion risk information for each fish object according to the potential lesion information and display it in the visualization model corresponding to the target water area.

[0021] In the embodiments of the present application, the water quality change conditions include the following parameters: chlorophyll concentration, dissolved oxygen, temperature, and pH value. Among them, the chlorophyll concentration mainly reflects the quantity and growth conditions of phytoplankton (such as algae) in the water body. In a pond for cultivating crucian carp, if the chlorophyll concentration suddenly increases, for example, from the normal 10 μg / L to 30 μg / L, this may mean a large reproduction of algae in the water body. On the one hand, the photosynthesis of algae can produce oxygen and increase the dissolved oxygen content in the water body. On the other hand, if the algae reproduce excessively, it may lead to a decrease in water transparency, affecting the penetration of light. Moreover, when a large number of algae die, they will consume the oxygen in the water and decompose to produce some harmful substances, threatening the living environment of crucian carp and increasing the risk of crucian carp getting sick, such as possibly causing gill diseases, etc. On the contrary, if the chlorophyll concentration is too low, such as below 5 μg / L, it may indicate that the water body is too barren, lacking plankton, and the natural bait of crucian carp is insufficient, affecting its growth and nutrient intake.

[0022] Dissolved oxygen is an essential condition for the survival of fish. Taking the water body for cultivating tilapia as an example, the suitable dissolved oxygen content for the growth of tilapia is generally above 4 mg / L. When the dissolved oxygen content drops to 3 mg / L, tilapia will show a decrease in appetite and slow growth; if it continues to drop below 2 mg / L, tilapia may float on the water surface and even suffocate to death in severe cases. For example, during the high-temperature period in summer, due to the vigorous activities of microorganisms in the water, the oxygen consumption increases. If oxygen is not added in time, the dissolved oxygen in the water body may decrease, resulting in tilapia lacking oxygen, a decline in resistance, and being easily infected with various diseases, such as gill rot disease, red skin disease, etc. When the dissolved oxygen content is sufficient, such as above 5 mg / L, the feeding and growth conditions of tilapia are good, and its resistance to diseases is relatively strong.

[0023] Temperature has an important impact on the physiological activities of fish, and different species of fish have their suitable survival temperature ranges. Taking the cultivation of rainbow trout as an example, rainbow trout is a cold-water fish, and the suitable water temperature for its growth is generally between 12 - 18 °C. If the water temperature rises above 20 °C, the metabolism of rainbow trout will accelerate, the oxygen consumption will increase, and at the same time its immunity may decline, making it vulnerable to the invasion of pathogens such as bacteria and viruses, leading to diseases, such as infectious hematopoietic necrosis disease, etc. On the contrary, when the water temperature is too low, such as below 8 °C, the appetite of rainbow trout will decrease, the growth rate will slow down, and long-term exposure to low-temperature environments may also cause it to be frostbitten, thereby leading to diseases such as saprolegniasis.

[0024] The pH value reflects the acidity or alkalinity of water bodies. In ponds for cultivating white - leg shrimp, the suitable pH value range for white - leg shrimp is generally between 7.8 and 8.6. If the pH value is too high, reaching above 9.0, the water body is too alkaline, which will corrode the gill tissue of the shrimp, affecting its respiratory function. At the same time, it may also enhance the toxicity of some substances in the water body, and the shrimp is prone to problems such as slow growth and difficulty in molting, and may even die. If the pH value is too low, such as below 7.0, the water body is too acidic, which will affect the shrimp's absorption of minerals such as calcium, resulting in softening of the shrimp's carapace, decreased resistance, and being easily infected with bacterial diseases such as black - gill disease.

[0025] These water quality parameters are interrelated and interact with each other, jointly affecting the survival and health of fish. By monitoring these parameters in real - time, the aquaculture environment can be adjusted in a timely manner to prevent the occurrence of fish diseases.

[0026] In step 101, assume that the target area is a large freshwater fishpond stocked with a large number of grass carp. Water quality sensors are installed in the fishpond, and these sensors can monitor parameters such as chlorophyll concentration, dissolved oxygen, temperature, and pH value in real - time.

[0027] For example, during the high - temperature period in summer, the sensors detect that the water temperature gradually rises and reaches above 30°C (exceeding the upper limit of the suitable growth water temperature of 20 - 30°C for grass carp), and at the same time, the dissolved oxygen content drops from the normal 6 mg / L to below 4 mg / L.

[0028] Thus, through the above - mentioned step 101, water quality change information can be obtained in real - time, providing an environmental data basis for subsequent analysis of fish health status. Aquaculture workers can adjust the water quality in a timely manner according to these data, such as turning on oxygen - increasing equipment to increase dissolved oxygen, to avoid fish stress or disease caused by water quality deterioration.

[0029] In step 102, a high - definition camera is installed above the fishpond to take pictures of the grass carp in the fishpond at regular intervals. The camera can cover most of the fishpond area to ensure that fish in different positions can be photographed.

[0030] For example, a high - definition picture is taken every 15 minutes, and in the picture, the body shape, body surface color, and state of the grass carp can be clearly seen.

[0031] Thus, through the above - mentioned step 102, intuitive visual information of the fish can be obtained, providing image data for subsequent behavior recognition and lesion recognition. These image information can be one of the important bases for judging the health status of fish.

[0032] In step 103, computer vision and machine - learning algorithms are used to analyze the photographed fish images.

[0033] For example, behavioral information such as the swimming speed, swimming direction, and aggregation state of grass carp is identified through an algorithm. If it is found that the originally dispersed grass carp begin to gather in a certain corner of the fishpond, or the swimming speed of individual grass carp is significantly slower than that of other fish.

[0034] In this way, the behavioral characteristics of fish are extracted from the images, and this behavioral information can reflect the health and physiological state of the fish. Abnormal behavioral manifestations may be early signals of fish diseases, which helps to detect potential health problems in a timely manner.

[0035] In step 104, it is assumed that the fish lesion recognition model is a deep learning model trained with a large amount of data. Based on this assumption, the previously obtained fish behavioral information (such as abnormal swimming), fish image information (such as spots on the body surface), and water quality changes (such as too high water temperature, low dissolved oxygen) are input into the model. The model analyzes these multi-dimensional data and determines that a certain grass carp may have gill rot disease (due to a combination of factors such as water quality deterioration, abnormal behavior of the fish, and possible minor injuries on the body surface).

[0036] Identifying lesions by integrating information from multiple aspects improves the accuracy and reliability of the identification. It no longer solely relies on a single type of information to judge whether a fish is sick, and can analyze the health status of fish more comprehensively to discover potential lesion risks.

[0037] In step 105, the potential lesion information of each grass carp judged by the model is quantified to generate lesion risk information.

[0038] For example, for a grass carp that may have gill rot disease, a relatively high lesion risk level is given (such as an 80% probability of being sick), and this information is displayed in a visualization model with the fishpond as the background, and the locations of grass carp with a higher disease risk are marked with different colors or icons.

[0039] In this way, the health status of the fish population is presented to the farmers in an intuitive manner, facilitating the farmers to quickly understand which fish have a relatively high lesion risk, so as to take targeted measures, such as isolating the sick fish and treating them, improving the efficiency and effectiveness of aquaculture management, and reducing economic losses.

[0040] As an optional embodiment, in 103, behavioral recognition is performed on the fish image information to obtain the fish behavioral information of each fish object in the target area, including: Extract the local feature image containing the object to be recognized from the fish image information; compare the local feature image with the standard fish feature images in the fish species database to obtain the fish species to which the object to be recognized belongs; based on the fish species, perform behavior recognition on the fish image information to obtain the fish behavior types and corresponding fish behavior parameters included in the fish image information as the fish behavior information.

[0041] For example, in an actual fish farming environment, such as a large pond where multiple fish species are mixed, the fish image information collected by a camera often contains numerous fish individuals and surrounding environmental information. To accurately identify the behavior of each fish, it is first necessary to extract the local feature image containing the object to be recognized (i.e., a single fish) from the overall image. This can be achieved using object detection algorithms in computer vision, such as Faster R-CNN, YOLO, etc. based on deep learning. These algorithms can accurately locate fish individuals in the image and segment them from the background and other fish to obtain a local feature image containing only a single fish. For example, for a carp swimming in the pond, the algorithm can accurately frame this carp and extract its local image for subsequent analysis.

[0042] Build a rich fish species database containing standard feature images of various common cultured fish species. These images record the appearance characteristics of different fish species in their normal state, such as body shape, color, fin shape and position, etc. Compare the extracted local feature image of the object to be recognized with the standard images in the database. Through image matching algorithms, such as SIFT (Scale-Invariant Feature Transform), SURF (Speeded-Up Robust Features) algorithms based on feature points, or deep learning-based image classification algorithms, calculate the similarity between the image to be recognized and each standard image in the database, and find the standard image with the highest similarity to determine the fish species to which the object to be recognized belongs. For example, for the extracted local feature image, after comparison with the database, it is determined that the fish is a grass carp.

[0043] Different species of fish have different behavioral patterns and habits. After determining the fish species, the image information of the fish is subjected to behavioral recognition according to the behavioral characteristics of the species. Using a behavioral recognition model specifically trained for this fish species (which can be a model based on machine learning or deep learning), information such as the actions and postures of the fish is analyzed to determine its behavioral type, such as swimming, feeding, resting, avoiding, etc. At the same time, parameters related to these behaviors can be further extracted, such as swimming speed, swimming direction, feeding frequency, etc. For example, for the identified grass carp, the model can determine that it is currently performing a feeding behavior and can calculate parameters such as its feeding frequency of 3 times per minute and swimming speed of 10 centimeters per second. These behavioral types and parameters together constitute the fish behavior information of the grass carp.

[0044] By first determining the fish species and then performing behavioral recognition based on the species characteristics, the behavioral differences between different fish species can be fully considered, avoiding the confusion of behavioral patterns of different fish species, thus improving the accuracy of behavioral recognition. For example, the swimming patterns and speeds of carp and crucian carp may be different. After accurately identifying the species, it is possible to more precisely judge whether their behaviors are normal. In actual fish farming, polyculture models are relatively common. This method can accurately identify the behaviors of each fish species in a complex polyculture environment, providing detailed fish group behavior information for farmers and helping to better manage and monitor the health status of the fish group in the polyculture pond. For example, in a pond where grass carp, carp, and crucian carp are cultured simultaneously, the behaviors of different fish species can be monitored and analyzed separately. It can not only identify the behavioral types of fish but also obtain detailed parameters related to them, providing more valuable information for in-depth research on the physiological state and health status of fish. Farmers can timely detect abnormal behaviors of fish based on these parameters. For example, a sudden decrease in feeding frequency may indicate a problem with the fish's health, and corresponding measures can be taken for prevention and treatment. Based on accurate fish species identification and behavior information, farmers can develop personalized aquaculture management strategies for different fish species, such as adjusting the feed feeding amount and optimizing the aquaculture environment, improving aquaculture efficiency and economic benefits, and also helping to ensure the healthy growth of fish.

[0045] Further optionally, the fish behavior parameters are determined based on the fish behavior type. If the fish behavior type includes a movement behavior, the fish behavior parameters at least include: fish movement distance, movement direction, movement time, movement trajectory type. If the fish behavior type includes predation, the fish behavior parameters at least include: prey object type, predation time, predation times, predation frequency. If the fish behavior type includes dormancy, the fish behavior parameters at least include: dormancy time, dormancy times, dormancy location.

[0046] Specifically, the above-mentioned fish behavior parameters are introduced below with reference to specific examples.

[0047] Fish movement distance: Assume that in a perch pond, an image recognition system monitors a perch swimming from one side of the pond to the other over a period of time. Using the scale or other reference information in the image, combined with the image recognition algorithm's determination of the perch's position at different times, its movement distance can be calculated. For example, it was calculated that the perch moved about 5 meters in 10 minutes. Farmers can determine the range of activity and vitality of the perch by analyzing the movement distance. If the movement distance suddenly decreases, it may mean that there is a problem with the health of the perch, or that environmental factors such as water quality have changed.

[0048] Movement direction: Similarly, for the above-mentioned perch, the image recognition system can determine its movement direction based on its position changes in different frames. For example, it is found that this perch is moving from the northwest to the southeast of the fish pond. Knowing the movement direction of fish can help farmers understand the activity patterns of fish schools. For example, some fish may have a fixed migration direction in a specific time period. If the direction is abnormal, it may indicate the existence of external interference factors, such as water pollution, water flow changes, etc.

[0049] Movement time: Record the time points when the bass starts and ends its movement to get its movement time. For example, if this bass starts moving at 10:05 am and stops at 10:15 am, the movement time is 10 minutes. The analysis of movement time can be combined with other parameters to determine the activity intensity and behavior patterns of fish. For example, if the movement time of the bass during the day is significantly reduced, and the movement time at night is increased, it may indicate that its adaptation to environmental factors such as light has changed.

[0050] Movement track type: By continuously recording the location information of the perch at different times, its movement track can be plotted. Movement track types may include straight lines, curves, spirals, etc. For example, the movement track of this perch was observed to be irregularly curved, which may reflect its behavior of looking for food, exploring the environment, or avoiding predators. Different movement track types can provide farmers with clues about the motivation of fish behavior and environmental adaptability.

[0051] Prey Object Type: In a pond where salmon are farmed, it is identified through an image recognition system that the salmon are preying. Further analysis of the images can determine the type of prey object of the salmon, such as small shrimps, plankton, or other fish. For example, it is found that this salmon preys on a type of small shrimp in the pond. Understanding the prey object type is very important for farmers to rationally formulate feed and manage the aquaculture ecosystem. If the salmon mainly preys on a certain specific organism and the quantity of this organism is insufficient, it may be necessary to adjust the feed formula to meet the nutritional needs of the salmon.

[0052] Prey Time: Record the start and end time points of the salmon's predation to determine the prey time. For example, this salmon starts preying at 2:30 pm and ends at 2:40 pm, with a prey time of 10 minutes. The length of the prey time can reflect the predation efficiency and hunger level of the salmon. If the prey time is too long, it may mean that the number of prey objects is scarce or the predation ability of the salmon has declined, and further observation and analysis are needed.

[0053] Prey Frequency: Count the number of times the salmon preys within a certain time period. For example, within one hour, this salmon preys 5 times. The change in the prey frequency can be used as an indicator to judge the health status and appetite of the salmon. If the prey frequency suddenly decreases, it may indicate that the salmon is sick or not interested in the current food.

[0054] Predation Frequency: Calculate the predation frequency based on the number of prey times and time. For example, if the above-mentioned salmon preys 5 times within one hour, the predation frequency is once every 12 minutes. The analysis of the predation frequency can help farmers understand the feeding pattern of the salmon, reasonably arrange the feeding time and amount of feed, and improve the feed utilization rate.

[0055] Dormancy Time: Taking the farming of eels as an example, it is monitored through an image recognition system that the eels enter the dormant state. Record the start and end time points of the eels' dormancy to determine the dormancy time. For example, this eel starts to dorm at 10 pm and wakes up at 6 am the next morning, with a dormancy time of 8 hours. The length of the dormancy time has an important impact on the growth and health of the eels. If the dormancy time is too short or too long, it may indicate that there are problems with the physiological state of the eels, and it is necessary to further check the aquaculture environment and the health status of the eels.

[0056] Dormancy Frequency: Count the number of times the eels dorm within a certain time period. For example, within one week, this eel dormed 10 times. The change in the dormancy frequency can reflect the living habits and environmental adaptability of the eels. If the dormancy frequency suddenly increases or decreases, it may be related to changes in environmental factors such as water temperature and water quality.

[0057] Hibernation Location: Determine the specific location where eels hibernate, such as in a certain corner of a pond, among aquatic plants, or in a cave. Understanding the hibernation location of eels can help aquaculture personnel optimize the aquaculture environment and provide a more suitable hibernation place. For example, if eels often choose to hibernate in a specific corner, aquaculture personnel can add some shelters in this area to improve the sense of security and comfort of eels.

[0058] As an optional embodiment, in 103, after comparing the local feature image with the standard fish feature images in the fish species database to obtain the fish species to which the object to be identified belongs, the climate conditions at the current time period can also be monitored in real time. Furthermore, obtain the types of water body areas where each fish object appears, and the geographical location range of the target area. Then, based on the climate conditions, the types of water body areas, and the geographical location range of the target area, verify the matching degree between the fish species and the target area. If the matching degree is not lower than the set threshold, the verified fish species is used as the final output target fish species.

[0059] After identifying the fish species, use meteorological monitoring equipment (such as weather stations, satellite cloud image data, etc.) to obtain the climate conditions at the current time period in real time. The climate conditions include multiple key factors, such as temperature, humidity, light intensity, air pressure, wind force, and wind direction, etc. Taking a large fish farming pond located by the sea as an example, on a certain day in summer, it is monitored by a weather station that the temperature on that day reaches 32 °C, the humidity is 70%, the light intensity is strong, the wind force is level 3 and the wind direction is southeast. This climate condition information is crucial for subsequent judgment of the matching degree between the fish species and the target area.

[0060] Determine the type of water body area where the fish is located through means such as on-site investigation and Geographic Information System (GIS). The types of water body areas are rich and diverse, including but not limited to freshwater lakes, rivers, seawater aquaculture ponds, offshore waters, reservoirs, etc. For example, it is monitored that the target area is a small freshwater reservoir located in the mountains, and its water quality, water flow speed, water depth and other characteristics are significantly different from other types of water body areas.

[0061] With the help of GPS positioning technology, map data, etc., clarify the geographical location range of the target area. It can be accurate to longitude and latitude coordinates, or described by a more macroscopic geographical area, such as a specific province, city, village, etc. For example, the target area is located in a coastal city in Guangdong Province in the south of China, and the geographical environment such as climate and hydrology in this area has unique characteristics.

[0062] Each fish species has its suitable living climate conditions, preferred types of water body areas, and specific geographical distribution ranges. Compare and analyze these suitable living conditions of the identified fish species with the actual situation monitored currently, and calculate the matching degree between them.

[0063] For example, the identified fish species is salmon. Salmon are generally suitable to live in cold water environments with low water temperatures and clear water quality, and are mainly distributed in cold water sea areas at high latitudes or cold water streams in mountainous areas. Currently, the target area is a seawater aquaculture pond located in the tropical region, where the water temperature is relatively high in summer, and the water quality and water flow conditions are quite different from the suitable living environment of salmon. Through comprehensive analysis of climatic conditions (high temperature), water body area type (tropical seawater pond), and geographical location range (tropical region), it is concluded that the matching degree of salmon with this target area is relatively low.

[0064] The matching degree can be calculated by setting weights for multiple indicators and scoring. For example, climatic conditions account for 40% of the weight, water body area type accounts for 30% of the weight, and geographical location range accounts for 30% of the weight. Corresponding scores are given according to the compliance degree of each indicator, and finally the total matching degree is calculated by weighted averaging.

[0065] Set a matching degree threshold (such as 70%). If the matching degree of the identified fish species with the target area after calculation is not lower than this threshold, it indicates that it is reasonable for this fish species to survive in the current target area, and it will be used as the finally determined target fish species. For example, the identified fish species is tilapia. After verification, tilapia are suitable to live in warm waters. The target area is a freshwater aquaculture pond in a certain southern city of China. The current climate is warm, and both the water body area type and geographical location range match the suitable living conditions of tilapia, with a matching degree of 80%, which is higher than the set threshold of 70%. Then the finally determined target fish species is tilapia.

[0066] Thus, by verifying the identified fish species by combining various factors such as climatic conditions, water body area types, and geographical location ranges, the misjudgment that may occur when relying solely on image features is avoided. For example, at the juvenile stage or under the influence of the environment, the appearance of some fish may be similar to that of other species. Through the verification of environmental factors, their species can be determined more accurately, reducing the subsequent management problems caused by misidentification. Accurate fish species information is crucial for aquaculture decisions. After understanding the matching degree between the fish species and the target area, aquaculture personnel can reasonably adjust aquaculture strategies according to the suitable survival conditions of the fish, such as water quality regulation, feed feeding, and aquaculture density control. For example, if it is determined that the fish species is suitable for a cold water environment, but the water temperature in the current target area is relatively high, aquaculture personnel can take cooling measures, such as increasing the water change frequency and building a sunshade, to improve the success rate and economic benefits of aquaculture. Ensuring that the introduced fish species matches the target area can effectively prevent ecological risks caused by the invasion of alien species or the inability to adapt to the environment. If fish species that are not suitable for the local environment are introduced, it may damage the local ecological system, such as competing for resources with local species and spreading diseases. By verifying the matching degree, such situations can be avoided, protecting the balance and stability of the ecological environment. After determining the accurate fish species and suitable aquaculture environment, aquaculture resources can be utilized more reasonably, avoiding waste of resources. For example, according to the habits and needs of the fish, feed can be accurately put in and aquaculture facilities can be reasonably arranged to improve the resource utilization efficiency and reduce the aquaculture cost.

[0067] As an optional embodiment, in 103, if there are multiple fish species with a matching degree higher than the set threshold, then the fish species with the highest occurrence probability in the target area is selected from the multiple verified fish species as the target fish species. If the matching degree is lower than the set threshold, the step of comparing the local feature image with the standard fish feature image in the fish species database to obtain the fish species to which the object to be identified belongs is executed again.

[0068] After calculating the matching degree between the identified fish species and the climatic conditions, water body area type, and geographical location range of the current target area, there may be a situation where the matching degrees of multiple fish species are higher than the set threshold. For example, in a freshwater lake aquaculture area in the south, several possible fish species are initially determined through image recognition. After calculating the matching degree, it is found that the matching degrees of crucian carp, common carp, and grass carp are all higher than the set threshold (assumed to be 70%). This means that these three fish species theoretically have the possibility to survive in this area and have good adaptability to the current environmental conditions.

[0069] When there are multiple highly matching fish species, it is necessary to further determine the most likely target fish species. This can be achieved by referring to historical data, local fishery statistics, expert experience, etc., to evaluate the probability of each fish species appearing in the target area. For example, by consulting local fishery records and the experience sharing of fish farmers, it is found that in the aquaculture area of this freshwater lake, the aquaculture quantity and natural occurrence frequency of crucian carp are relatively higher than those of carp and grass carp, that is, the probability of crucian carp appearing in this area is the greatest. Then, in this case, the crucian carp is determined as the target fish species.

[0070] If, after calculation, the matching degrees of all initially identified fish species with the target area are lower than the set threshold, it indicates that the currently identified fish species do not match the environmental conditions of this area well, and there may be identification errors. For example, in a seawater aquaculture area, after calculating the matching degrees of several initially identified fish species, the matching degrees are all lower than 70%. This shows that these fish are unlikely to survive in this seawater area, and the current identification results may be inaccurate.

[0071] When the matching degree is lower than the set threshold, in order to obtain a more accurate fish species identification result, the system will re - compare the extracted local feature images with the standard fish feature images in the fish species database. By applying image comparison algorithms again, such as feature point matching algorithms, deep - learning - based image classification algorithms, etc., to re - identify the fish species, in order to find fish species that match the environmental conditions of the target area better.

[0072] From a technical effect perspective, when there are multiple highly matching fish species, by selecting the fish species with the highest probability of occurrence in the target area, the scope can be further narrowed, and the accuracy of the finally determined fish species can be improved. This avoids the uncertainty caused by multiple species all meeting the basic conditions and makes the recognition result more in line with the actual situation. When the matching degree is low, re-performing the recognition step can timely correct possible misidentifications, continuously optimize the recognition result, and ensure that the finally determined fish species is suitable for the environment of the target area. This processing method enables the system to handle different situations. Whether there are multiple possible fish species or the recognition result does not match the environment, corresponding operations can be carried out for adjustment and optimization. It improves the adaptability and reliability of the system in complex environments and can better meet the requirements for accurate identification of fish species in the actual aquaculture process. The target fish species obtained through the above steps is more accurate. Aquaculture personnel can formulate more reasonable aquaculture strategies according to the characteristics and requirements of this species, such as selecting appropriate feed, controlling aquaculture density, and adjusting water quality management measures. This helps to improve aquaculture efficiency, reduce aquaculture risks, and increase aquaculture benefits. Accurately identifying fish species and ensuring their match with the target area helps to reasonably plan and utilize fishery resources. It avoids introducing fish species that are not suitable for the local environment, reduces resource waste and ecological risks, protects the local fishery ecological balance, and promotes the sustainable development of fishery resources.

[0073] As an optional embodiment, the fish lesion recognition model at least includes the following structures: an extraction layer, a prediction layer, and a correction layer.

[0074] Based on the above model, in 104, through the fish lesion recognition model, the fish behavior information, the fish image information, and the water quality change situation are used for lesion recognition to obtain potential lesion information of each fish object, including: 201. Through the extraction layer, the fish behavior information, the fish image information, and the water quality change situation are subjected to feature extraction to obtain image feature information of each fish object; wherein, the image feature information of each fish object at least includes: fish behavior characteristics, body surface image characteristics, and water quality change characteristics of each fish object; 202. Through the prediction layer, according to the body surface image characteristics and water quality change characteristics of each fish object, the probability of body surface lesions of each fish object is predicted; 203. Through the correction layer, the fish behavior characteristics of each fish object are used to correct the probability of body surface lesions of each fish object to obtain potential lesion information of each fish object.

[0075] In step 201, fish behavior information, fish image information, and water quality change conditions are used as inputs. These information reflect the survival status of fish from different perspectives. For example, fish behavior information can reflect its daily activity patterns, fish image information can directly show the body surface condition, and water quality change conditions reflect the environmental factors for fish survival.

[0076] Using relevant algorithms or models in the extraction layer, feature extraction is performed on the input multi-source data. For fish behavior information, features such as movement speed, activity frequency, and social behavior may be extracted; for fish image information, body surface image features such as color, texture, and shape will be extracted, such as whether there are spots, ulcers, and the state of fins; for water quality change conditions, change features of parameters such as chlorophyll concentration, dissolved oxygen, temperature, and pH value will be extracted. Finally, image feature information containing fish behavior features, body surface image features, and water quality change features is obtained, providing a basis for subsequent analysis.

[0077] In step 202, the body surface image features and water quality change features obtained from the extraction layer are input into the prediction layer. This is because fish body surface lesions are often closely related to the appearance changes of the body surface and the water quality environment for survival. For example, too low dissolved oxygen in the water may cause fish to have difficulty breathing, which in turn affects their health, and some abnormal symptoms may appear on the body surface; while features such as spots and ulcers in the body surface image are directly related to the lesions.

[0078] The prediction layer uses existing models and algorithms to predict the probability of body surface lesions for each fish object based on the input feature information. This process may be based on methods such as machine learning and deep learning. By learning and analyzing a large amount of historical data, a relationship model between features and lesion probabilities is established. For example, if irregular white spots appear in the body surface image features and the water quality change features show that the water temperature has risen abnormally, the model may predict that the fish has a relatively high probability of body surface lesions.

[0079] In step 203, the fish behavior features obtained from the extraction layer are introduced into the correction layer. The behavior changes of fish are also important bases for judging their health status. For example, when fish are sick, they may show abnormal behaviors such as slow swimming, swimming alone away from the group, and losing balance.

[0080] The correction layer corrects the probability of body surface lesions obtained by the prediction layer according to the fish behavior features. If the fish behavior features show obvious abnormalities, such as staying still for a long time or jumping frequently, even if the probability of body surface lesions given by the prediction layer is low, the correction layer will appropriately increase this probability to more accurately reflect the potential lesion situation of the fish. On the contrary, if the fish behavior is normal and the probability of the prediction layer is high, the correction layer may reduce this probability to avoid misjudgment.

[0081] Thus, by comprehensively considering fish behavior information, fish image information, and water quality changes, the health status of fish is evaluated from multiple dimensions, avoiding inaccurate judgments that may result from relying on a single information source. For example, relying solely on the body surface image may misjudge some body surface abnormalities caused by temporary water quality changes as lesions, while combining water quality change characteristics and fish behavior characteristics can more accurately identify true lesions. The model can detect potential problems at the early stage of fish lesions. Even if there are no obvious lesion symptoms on the fish body surface, but there are already some abnormal behaviors, or the water quality environment has changed unfavorably to fish health, the model can analyze this information, predict and correct the lesion probability in advance, providing valuable time for timely prevention and control measures. Accurate lesion identification results can help aquaculture personnel formulate more scientific aquaculture management strategies. If the model finds that the potential lesion probability of certain fish is relatively high, aquaculture personnel can adjust the water quality in a timely manner, increase the feed nutrition, isolate and treat the diseased fish, etc., thus effectively reducing the incidence and mortality of fish diseases and improving aquaculture efficiency. The multi-layer structure of the model and the utilization of multi-source data enable it to adapt to different aquaculture environments and fish species. Whether in pond aquaculture, reservoir aquaculture, or marine aquaculture, whether for common fish species or some special species, the model can achieve relatively accurate lesion identification through learning and analysis of corresponding data, with strong generalization ability and adaptability.

[0082] Further optionally, in 202, according to the body surface image characteristics and water quality change characteristics of each fish object, predicting the body surface lesion probability of each fish object includes: Determining the visibility level in the target area according to the water quality change characteristics of each fish object; performing sharpening processing on the body surface image characteristics of each fish object based on the visibility level to obtain the optimized body surface image characteristics of each fish object; extracting multi-dimensional visual features from the optimized body surface image characteristics to obtain the multi-dimensional body surface image characteristics of each fish object; the multi-dimensional body surface image characteristics at least include: body surface texture characteristics, body surface brightness characteristics, and body surface color characteristics; predicting the correlation between the multi-dimensional body surface image characteristics and the standard body surface lesion images based on a pre-set standard body surface lesion image library, and calculating the body surface lesion probability of each fish object based on the correlation.

[0083] Specifically, some parameters in the water quality change characteristics, such as chlorophyll concentration, suspended solid content, etc., will directly affect the transparency of the water body, and then affect the visibility in the target area. For example, when the chlorophyll concentration is high, the water body may show a darker green color, resulting in a decrease in visibility; and an increase in the suspended solid content will also make the water body turbid and the visibility poor. By establishing relevant mathematical models or empirical formulas, these parameters in the water quality change characteristics are associated with the visibility level. For instance, it is set that the chlorophyll concentration within a certain range corresponds to a low visibility level, within another range corresponds to a medium visibility level, and so on. In this way, the visibility level of each area where fish objects are located can be determined according to the water quality change characteristics of that area.

[0084] Different visibility levels will affect the clarity of the fish body surface images captured. For the case of a low visibility level, the images may be relatively blurred and details are difficult to identify. Therefore, according to the determined visibility level, corresponding image sharpening algorithms are used to process the body surface image features. For example, for images with a low visibility level, the Laplacian sharpening operator or Gaussian sharpening filter, etc. can be used to enhance the edges and details of the image, making the features such as the texture and spots on the fish body surface clearer. After the sharpening process, the optimized body surface image features are obtained, which are more conducive to subsequent analysis.

[0085] The optimized body surface image features contain rich information. Through multi-dimensional visualization feature extraction methods, this information is further explored. Among them, the body surface texture features can reflect the surface structure of the fish body surface, such as the arrangement of scales, whether there are protrusions or depressions, etc.; the body surface brightness features can reflect the light and dark degrees of different parts of the fish body surface, and some lesions may cause changes in local brightness; the body surface color features can show the color distribution and hue of the fish body surface, and some diseases may change the color of the fish body surface. Using specialized image analysis algorithms, such as texture analysis algorithms (such as gray-level co-occurrence matrix), brightness and color space conversion algorithms, etc., these multi-dimensional body surface image features are extracted from the optimized body surface image features.

[0086] Pre-collect and organize a large number of fish body surface images with clear lesion types and degrees to construct a standard image library of body surface lesions. Compare and analyze the multi-dimensional body surface image features of each fish object extracted with the image features in the standard image library. By calculating the similarity or correlation index (such as Euclidean distance, cosine similarity, etc.) between the multi-dimensional body surface image features and the standard image features, evaluate the matching degree between the two. The higher the correlation, the greater the possibility that the fish has the corresponding lesion. According to the size of the correlation, combined with a certain probability calculation model (such as a probability prediction model based on machine learning), calculate the body surface lesion probability of each fish object. For example, if the multi-dimensional body surface image features of a certain fish object have a high correlation with the image features of a certain lesion type in the standard image library, then it can be predicted that the fish object has a high probability of body surface lesions.

[0087] Thus, by sharpening the body surface image features according to the visibility level, the problem of image blurring caused by water quality factors can be effectively improved, making the image clearer, thereby improving the accuracy of subsequent feature extraction. Accurate feature extraction is the basis for lesion prediction and provides more reliable data support for subsequent analysis.

[0088] The multi-dimensional visualization feature extraction method can analyze the fish body surface image from multiple angles and comprehensively mine the information in the image. By considering multiple feature dimensions such as body surface texture, brightness, and color, the features of the fish body surface can be described more meticulously, capturing some subtle changes, which may be early signs of lesions and help detect fish diseases early.

[0089] Based on the standard image library of body surface lesions for correlation analysis and probability calculation, the body surface features of fish can be compared with known lesion features, making the prediction results more objective and accurate. By quantifying the correlation to calculate the lesion probability, the error of subjective judgment is avoided, the reliability of the prediction is improved, and a scientific basis is provided for breeders to take prevention and control measures in a timely manner.

[0090] Considering the impact of water quality changes on visibility and processing the image accordingly, the system can adapt to different water quality environments. Whether in clear water or turbid water, accurate fish body surface image features can be obtained through corresponding processing methods, ensuring the stability and reliability of lesion prediction, and improving the adaptability and generality of the system.

[0091] In the embodiment of the present application, the calculation process of the correlation between the body surface image feature point Pi in the multi-dimensional body surface image features and the standard image of body surface lesions is expressed as the following formula: ; Wherein, Indicates the correlation between the body surface image feature point Pi and the j-th body surface lesion standard image, and the weight factor and are used to adjust the contribution degrees of the j-th body surface lesion standard image in the x-dimension and y-dimension respectively. x and y represent the coordinate values of the projection components of the body surface image feature point Pi in the x-dimension and y-dimension.

[0092] Represents the actual coordinate value of the body surface image feature point Pi in the x-th dimension, represents the standard coordinate value corresponding to the body surface image feature point Pi in the x-th dimension, represents the average coordinate value of all body surface image feature points in the multi-dimensional body surface image feature in the x-th dimension. m is the number of body surface image feature points in the multi-dimensional body surface image feature, is used to represent the average correlation fluctuation range between each body surface image feature point in the multi-dimensional body surface image feature and the j-th body surface lesion standard image in the x-th dimension.

[0093] Represents the actual coordinate value of the body surface image feature point Pi in the y-th dimension, represents the standard coordinate value corresponding to the body surface image feature point Pi in the y-th dimension, represents the average coordinate value of all body surface image feature points in the multi-dimensional body surface image feature in the y-th dimension, is used to represent the average correlation fluctuation range between each body surface image feature point in the multi-dimensional body surface image feature and the j-th body surface lesion standard image in the y-th dimension.

[0094] This formula is used to calculate the correlation between the body surface image feature point Pi in the multi-dimensional body surface image feature and the body surface lesion standard image. By separately considering the coordinate values and related parameters in the x-dimension and y-dimension, this formula can comprehensively measure the relationship between the feature point and the standard image from the perspective of two-dimensional space. For example, by calculating in the x-dimension and in the y-dimension, it can reflect the deviation degree of the feature point Pi from the average coordinate in each dimension, so as to understand the position characteristics of the feature point in the overall image.

[0095] The setting of the weight factor can flexibly adjust its contribution degree to the correlation calculation in the x-dimension and y-dimension according to different body surface lesion standard images j. This means that for different types of body surface lesion standard images, according to their characteristics and importance, the influence of certain dimensions can be emphasized or weakened targeted, making the correlation calculation more flexible and accurate.

[0096] Considering the coordinate value differences respectively measures the differences between the actual coordinate values and the standard coordinate values of the feature point \(P_i\) in the \(x\) - dimension and the \(y\) - dimension. This difference reflects the similarity degree of the feature point and the standard image in morphology. The smaller the difference, the closer the feature point is to the standard image, and the higher the correlation between the two; conversely, the lower the correlation. In this way, the subtle differences between the feature point and the standard image can be accurately captured, so as to more precisely evaluate their correlation.

[0097] Measuring the overall fluctuation range respectively represents the average correlation fluctuation range between each body surface image feature point in the multi - dimensional body surface image features and the \(j\) - th body surface lesion standard image in the \(x\) - dimension and the \(y\) - dimension. They can help evaluate the dispersion degree of the feature point and the standard image in different dimensions, so as to more comprehensively understand the relationship between the feature point and the standard image. If the fluctuation range is small, it indicates that the correlation between the feature point and the standard image in this dimension is relatively stable. On the contrary, if the fluctuation range is large, it means that there are large changes in the correlation between the feature point and the standard image in this dimension, and more careful evaluation of their relationship is needed.

[0098] By comprehensively considering multiple factors, this formula can accurately calculate the correlation between the body surface image feature point and the body surface lesion standard image, providing accurate data support for calculating the body surface lesion probability of fish objects based on correlation subsequently, and helping to more accurately identify the body surface lesion conditions of fish.

[0099] In the embodiments of this application, individualized potential fish lesion recognition is realized for different types of fish, improving the accuracy of fish lesion monitoring, simplifying the fish lesion monitoring process, and further enhancing the monitoring efficiency of fish lesions.

[0100] Figure 2 This is a schematic structural diagram of a fish disease monitoring system based on feature image recognition provided by the embodiments of this application, as Figure 2 shown. The system includes the following steps: A monitoring unit, which is used to monitor the water quality change situation in the target area in real - time through a water quality sensor; the water quality change situation includes the following parameters: chlorophyll concentration, dissolved oxygen, temperature, pH value; and collect the fish image information in the target area; An identification unit, which is used to perform behavior recognition on the fish image information to obtain the fish behavior information of each fish object in the target area; through a fish lesion recognition model, perform lesion recognition on the fish behavior information, the fish image information, and the water quality change situation to obtain the potential lesion information of each fish object; A display unit, which is used to generate the lesion risk information of each fish object according to the potential lesion information and display it in the visualization model corresponding to the target water area.

[0101] Please refer to Figure 3 , Figure 3 which is a schematic diagram of an embodiment of an electronic device provided by an embodiment of the present application. As Figure 3 shown, an embodiment of the present application provides an electronic device 500, including a memory 510, a processor 520, and a computer program 511 stored on the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 511, the foregoing embodiments are implemented.

[0102] Please refer to Figure 4 , Figure 4 which is a schematic diagram of an embodiment of a computer-readable storage medium provided by an embodiment of the present application. As Figure 4 shown, this embodiment provides a computer-readable storage medium 600, on which a computer program 611 is stored. When the computer program 611 is executed by a processor, the foregoing embodiments are implemented.

[0103] It should be noted that in the above embodiments, the descriptions of each embodiment have their own emphases. For parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0104] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0106] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions in the flowFigure 1 one process or multiple processes and / or boxes Figure 1 the functions specified in one box or multiple boxes

[0107] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide for implementing in the process Figure 1 one process or multiple processes and / or boxes Figure 1 the steps of the functions specified in one box or multiple boxes

[0108] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the appended claims are intended to be interpreted to include the preferred embodiments and all changes and modifications that fall within the scope of the present invention

[0109] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations

Claims

1. A fish disease monitoring method based on feature image recognition, characterized in that: The method at least comprises: Monitor the water quality changes in the target area in real time through water quality sensors; the water quality changes include the following parameters: chlorophyll concentration, dissolved oxygen, temperature, and pH value; Collect fish image information in the target area; Performing behavior recognition on the fish image information to obtain fish behavior information of each fish object in the target area; Using a fish lesion recognition model, the fish behavior information, the fish image information and the water quality change are used for lesion recognition to obtain potential lesion information of each fish object; The disease risk information of each fish object is generated according to the potential disease information and displayed in the visualization model corresponding to the target water area.

2. The fish disease monitoring method based on feature image recognition according to claim 1 is characterized in that: The performing behavior recognition on the fish image information to obtain fish behavior information of each fish object in the target area includes: Extracting a local feature image containing an object to be identified from the fish image information; Comparing the local characteristic image with a standard fish characteristic image in a fish species database to obtain the fish species to which the object to be identified belongs; The fish image information is subjected to behavior recognition based on the fish species to obtain the fish behavior type contained in the fish image information and the corresponding fish behavior parameters as the fish behavior information.

3. The fish disease monitoring method based on feature image recognition according to claim 2 is characterized in that: The fish behavior parameter is determined based on the fish behavior type; If the fish behavior type includes movement behavior, the fish behavior parameters at least include: fish movement distance, movement direction, movement time, and movement trajectory type; If the fish behavior type includes predation, the fish behavior parameters at least include: predation object type, predation time, predation times, and predation frequency; If the fish behavior type includes hibernation, the fish behavior parameters at least include: hibernation time, hibernation times, and hibernation location.

4. The fish disease monitoring method based on feature image recognition according to claim 2 is characterized in that: After comparing the local characteristic image with the standard fish characteristic image in the fish species database to obtain the fish species to which the object to be identified belongs, the method further includes: Real-time monitoring of climate conditions during the current period; Obtain the type of water area where each fish object appears, and the geographical location range of the target area; Verify the matching degree between the fish species and the target area based on the climate conditions, the water area type, and the geographical location of the target area; If the matching degree is not lower than the set threshold, the verified fish species will be used as the target fish species for the final output.

5. The fish disease monitoring method based on feature image recognition according to claim 4 is characterized in that: After verifying the matching degree between the fish species and the target area based on the climate conditions, the water area type, and the geographical location range of the target area, the method further includes: If there are multiple fish species with a matching degree higher than the set threshold, the fish species with the highest probability of appearing in the target area is selected from the multiple verified fish species as the target fish species; If the matching degree is lower than the set threshold, the step of comparing the local feature image with the standard fish feature image in the fish species database is re-executed to obtain the fish species to which the object to be identified belongs.

6. The fish disease monitoring method based on feature image recognition according to claim 1 is characterized in that: The fish lesion recognition model at least includes the following structures: an extraction layer, a prediction layer, and a correction layer; The fish lesion recognition model is used to identify the fish behavior information, the fish image information and the water quality change to obtain potential lesion information of each fish object, including: Through the extraction layer, the fish behavior information, the fish image information and the water quality change are subjected to feature extraction to obtain image feature information of each fish object; wherein the image feature information of each fish object at least includes: fish behavior features, body surface image features, and water quality change features of each fish object; Through the prediction layer, the probability of surface lesions of each fish object is predicted based on the surface image characteristics of each fish object and the characteristics of water quality changes; Through the correction layer, the fish behavior characteristics of each fish object are used to correct the probability of surface lesions of each fish object to obtain the potential lesion information of each fish object.

7. The fish disease monitoring method based on feature image recognition according to claim 6 is characterized in that: The method of predicting the probability of body surface lesions of each fish object according to the body surface image characteristics of each fish object and the water quality change characteristics includes: Determine the visibility level in the target area based on the water quality change characteristics of each fish target; Sharpening the body surface image features of each fish object based on the visibility level to obtain optimized body surface image features of each fish object; Performing multi-dimensional visualization feature extraction on the optimized body surface image features to obtain multi-dimensional body surface image features of each fish object; the multi-dimensional body surface image features at least include: body surface texture features, body surface brightness features, and body surface color features; Based on a preset body surface lesion standard image library, the correlation between the multidimensional body surface image features and the body surface lesion standard images is predicted, and the body surface lesion probability of each fish object is calculated based on the correlation.

8. The fish disease monitoring method based on feature image recognition according to claim 7 is characterized in that: The calculation process of the correlation between the body surface image feature point Pi and the body surface lesion standard image in the multi-dimensional body surface image feature is expressed as the following formula: ； in, represents the correlation between the body surface image feature point Pi and the jth body surface lesion standard image, and the weight factor and It is used to adjust the contribution of the j-th body surface lesion standard image in the x dimension and the y dimension respectively, where x and y represent the projection component coordinate values ​​of the body surface image feature point Pi in the x dimension and the y dimension; Represents the actual coordinate value of the feature point Pi of the body surface image in the x-th dimension, Indicates the standard coordinate value corresponding to the feature point Pi of the body surface image in the xth dimension, represents the average coordinate value of all body surface image feature points in the multidimensional body surface image feature in the xth dimension, m is the number of body surface image feature points in the multidimensional body surface image feature, Used to represent the average correlation fluctuation range between each body surface image feature point in the multidimensional body surface image feature and the jth body surface lesion standard image in the xth dimension; Represents the actual coordinate value of the feature point Pi of the body surface image in the y-th dimension, Indicates the standard coordinate value corresponding to the feature point Pi of the body surface image in the yth dimension, represents the average coordinate value of all body surface image feature points in the multi-dimensional body surface image feature in the y-th dimension, Used to represent the average correlation fluctuation range between each body surface image feature point in the multidimensional body surface image feature and the jth body surface lesion standard image in the yth dimension.

9. A fish disease monitoring system based on feature image recognition, characterized in that: The system comprises at least the following units: A monitoring unit is used to monitor the water quality changes in the target area in real time through a water quality sensor; the water quality changes include the following parameters: chlorophyll concentration, dissolved oxygen, temperature, pH value; collect fish image information in the target area; An identification unit, used for performing behavior identification on the fish image information to obtain fish behavior information of each fish object in the target area; Using a fish lesion recognition model, the fish behavior information, the fish image information and the water quality change are used for lesion recognition to obtain potential lesion information of each fish object; The display unit is used to generate the disease risk information of each fish object according to the potential disease information, and display it in the visualization model corresponding to the target water area.

10. An electronic device, characterized in that: including a memory for storing a computer software program; A processor is used to read and execute the computer software program, thereby implementing the fish disease monitoring method based on feature image recognition as described in any one of claims 1-7.

Images