JP6401411B1 - Artificial intelligence catch identification system, management system and logistics system - Google Patents

Abstract

[PROBLEMS] To systematize a fish identification system that makes use of artificial intelligence to identify catches and track the history, a fish management system that systematizes quality control of fish, and logistics of fish Provide a catch logistics system. A fish type feature quantity storage unit 7 that stores fish type feature quantities for each fish species 14 of a fish catch 13, an image pickup unit 4 that individually captures the fish catch 13 and extracts feature quantities, and a feature quantity Is compared with the fish species feature amount, the fish species determination unit 8 that determines the fish species 14 of the catch 13, the identification number assigning unit 9 that assigns the identification number 10 to the catch 13, and other features of the same type from the feature amount An individual feature amount extraction unit 6a that extracts an individual feature amount that can be distinguished from the fish 13 and an individual fish database 12 that stores the individual feature amount in association with the identification number 10 for each fish 13 are provided. [Selection] Figure 1

Description

  The present invention relates to a catch identification system, a management system, and a logistics system using artificial intelligence, and in particular, catches using artificial intelligence for conventional fisheries such as coastal fishery and offshore fishery, cultivated fishery, and aquaculture fishery. The present invention relates to a catch identification system for individually identifying a catch, a catch management system for managing the quality and price of the catch, and a catch logistics system for transporting the catch to a final consumption place.

  The fisheries in Japan are classified into, for example, coastal fisheries, near-shore fisheries, aquaculture fisheries, and cultivated fisheries. The coastal fishery is a fishing method for catching seafood by, for example, small-scale fixed net fishing or ground net fishing, which is carried out on the coastal area relatively close to our land. It is said that 85% of our fishermen are engaged in this coastal fishery 3a. The near sea fishery 3b is said to be offshore fishery, and it can be reciprocated within 2 to 3 days from the sea near our country by a fishing boat of about 20 to 150 tons, for example, purse seine fishing, sardine, saury, mackerel, horse mackerel, squid, etc. It is a fishing method to catch. It is said that nearly 40% of Japan's catch is due to this offshore fishery. Here, the offshore fishery targeting the fishing grounds of the countries of the world is out of the scope of the present invention.

  Organize the general flow from catching to catching fish in the fish market. The catch caught in the waters near Japan is landed at a nearby fishing port when caught by a fishing boat. At the fishing port, the caught fish are sorted according to fish type and size. Live fish in the catch is auctioned for each basket by a middleman by a method called “Utai Seri”. And the middleman who said the higher price can buy the basket. In addition, for the fresh fish, a “bidding” is performed in which a tag with the desired price is placed in the basket of the fish that the middleman wants. And the middleman who has the highest price gets the right to buy. The fish caught by the “bidding” is carried by the middleman to the live fish tank or the handling area. Live fish are moved to live fish carts and transported alive to central wholesale markets, and fresh fish are boxed by size according to their fish type, loaded on cold cars and transported to the central wholesale market. Then, the fish caught in the wholesale market of the consumption area is sold to consumers at a general fish shop. In this way, the freshness of the catch decreases with the elapsed time since it was caught, so how quickly it reaches the end consumer determines the quality of the catch and affects the price.

  On the other hand, the catch of fish caught in Japan's seas has changed due to the influence of El Nino, global warming, etc., and the habitat of the catch has also changed accordingly. There may be a decrease in the landing of the catch at a fishing port that has previously landed a particular catch, or an increase in the catch of a catch at a fishing port where there has been little landing of a certain catch until now, Each fishing port is struggling to respond.

  Furthermore, in recent years, the catch of East Asian countries such as China, Taiwan, and South Korea has increased, contrary to the decrease in catch near Japan. This is a marked increase in global demand for catch including these countries. For this reason, large fishing boats form a fleet to catch fishery in the high seas away from the sea near Japan and inside and outside Japan's exclusive economic zone (200 nautical miles).

  In order to respond to these trends, in recent years, a method of transporting freshly caught fish to remote consumption areas by using a fish freezing technique is being adopted. That is, fish species that could only be consumed in the vicinity of the fishing port until now can be transported to remote consumption areas. In addition, a fresh fish distribution network that provides fish directly to the consumer from the fishing port has begun to be implemented. This fresh fish distribution network saves time and costs because it eliminates the middlemen who traded fish caught between middlemen and restaurants that had been auctioning off the fish at the fishing port. .

  Furthermore, aquaculture and cultivated fisheries have become active as new techniques for fisheries. Aquaculture is a fishery that grows under human control in an aquarium or “ikes” from the egg until it reaches the shipping size. On the other hand, cultivated fisheries means that the weakest period from egg to fry is raised by human hands, and when you can protect yourself from foreign enemies, the fry is released into the sea and grown in the natural sea. A fishery that catches fish.

  Patent Document 1 discloses a fish type sorting device that can sort a plurality of types of fish at high speed without damaging them, and can be easily adapted to changes in fish types that require sorting, and has a simple configuration. Here, the image processing device or the like discriminates each fish type for a plurality of fishes, a discrimination means provided so as to be able to send a discrimination signal indicating the discrimination result, and each fish that has been discriminated by the discrimination means Is provided so as to cover the opening of the cell from the end to the lower part on the belt conveyor. A guide plate and a sorting lid that opens and closes in response to each discrimination signal sent from the discrimination means and drops a fish of a specific fish species into the fish box are described in the middle of the guide plate.

  Patent Document 2 discloses a fish freshness estimation method that can objectively estimate the freshness of the whole fish with higher accuracy than before. Here, it is a fish freshness estimation method, an absorbance spectrum obtaining step (S11) for obtaining an absorbance spectrum obtained by irradiating the fish eye with light including all or part of a wavelength band of 315 nm to 450 nm. And a freshness estimation step (S12) for estimating the freshness of the fish using the shape of the absorbance spectrum.

  According to Patent Document 3, a fish market information management system includes a weighing device installed in a fish market, a fish market information management server that manages information in the fish market, and a buyer who bids for a bid for fish. And a buyer terminal to be operated. The weighing device measures the weight of the fish landed by the shipper and sends information indicating the weight of the weighed fish to the fish market information management server, and the fish market information management server receives the weight received from the weighing device. Using the information, a list of catches for the catch is created and transmitted to the buyer terminal. It is described that the purchaser inputs the bid amount to the list of auctions displayed on the purchaser's terminal, and realizes the efficiency of the entire work in the fish market from the landing of the fish catch to the auction.

  Patent Document 4 discloses a marine water circulation type natural seafood breeding system that cultivates fishery on the ground for the purpose of improving the food self-sufficiency rate of fishery resources that are in a critically decreasing trend due to the effects of global warming. Yes. Here, seaweed spores are settled to form seaweed reefs in an orbited artificial lagoon A composed of excavated ground surface, or FRP or concrete structure skeletons, A gentle slope B for breeding, a seawater purification action by seaweed and bivalves, and a sandy layer C for breeding sandy benthic seafood, with renewable energy as the main power source in the lagoon Circulating the ocean water introduced by the ocean water intake system and pipeline transport system to form an artificial seaweed reef by aquaculture technology such as seaweed, and by the action of various animal and plant plankton captured from the ocean water intake It describes the reproduction of marine ecosystems consisting of the natural food chain to grow seafood without feed.

  Patent Literature 5 discloses a cultivated fishery support system that supports non-experts so that they can perform fisheries in the same manner as skilled workers in the aquaculture industry. Here, the operation is performed based on the operation plan data registered in advance by the operation management device, and the current position information is acquired by the GPS of the terminal device on which the work boat is installed. More accurate operation can be performed by monitoring the current position information and issuing instructions suitable for the situation. Furthermore, it is described that the operation management device can manage the growth status of resources, the harvest status at each fishing ground, and manage the harvested location in a lump by collecting the operation result data created after the operation. ing.

JP-A-8-140562 Japanese Patent Laying-Open No. 2015-232543 JP 2005-352955 A JP 2011-10645 A JP 2006-320203 A

(Individual recognition of fish catch)
The above-mentioned catches have a problem that after a “bidding” is performed in a fishing port market, it is impossible to accurately trace which catches go to which consumption route through which route and where they are cooked. From the consumer's point of view, there is little information about where the catch was caught, who caught it, when it was caught, and there is no choice but to trust a restaurant or fish shop that provides fish dishes. There's a problem. Even if it is the same fresh food, there is a big difference between vegetables and meat products, etc. that display the production area and producers to ensure “food safety and security”. In addition, it is assumed that branded catches are generated by the evolution of aquaculture and cultivated fisheries, and the quality of the catches is assumed to be differentiated from general catches. In that case, the individual recognition of the catch becomes an important strategy.

(Fish freshness decline)
The catches landed at the fishing port are put on the market by the fishermen from the fishermen and auctioned by the middlemen for the catches. For this reason, the fish on the market are left unattended until the end of the auction, during which time the freshness of the fish falls over time. In addition, since the auction itself is negotiated between a fisherman who wants to sell higher and a middleman who wants to buy cheaper, there is a problem that it takes time until the price is determined and the freshness of the catch is lowered.

(Objective quality prediction of fish catch)
The price of each catch is determined based on the experience of the middleman, who is said to be “connoisseur”, for the catch landed at the fishing port. In other words, the quality prediction of the catch depends on the experience value of the broker, so statistical data on the quality of the catch at each fishing port cannot be accumulated, and an evaluation of objective quality is recorded. There is a problem that can not be done. In addition, when there are different middlemen in each fishing port, it is difficult to predict the quality of the catch uniformly, and there is a problem that successors who have experience due to the aging of middlemen are difficult to grow. Furthermore, in the case of cultivated fisheries and aquaculture fisheries as well as the conventional fisheries described above, there is a problem of how to predict the objective quality of fish catches.

(Information sharing between fishing ports)
In addition, auctions in this fish market are held at each fishing port, and it is difficult to say that information sharing between markets is progressing. For this reason, for example, regarding the catches landed at each fishing port in Hokkaido, the catch of each catch has changed due to the El Nino phenomenon or global warming, and the size and shape of the catch has changed. There is a problem that it is difficult to statistically or objectively grasp changes related to the quality of the catch. Furthermore, there is a problem that it is difficult to statistically or objectively understand how the ratio of cultivated fisheries and aquaculture fisheries changes with respect to the above-described conventional fisheries.

  The purpose of the present application is to solve such problems, utilize artificial intelligence, identify individual catches and track their history, and catch the systematized quality control of the identified catches. The object is to provide a fish management system and a fish distribution system that systematizes the logistics of the identified fish.

  In order to achieve the above object, the artificial intelligence fish identification system according to the present invention selects an image of the fish from the big data, extracts the feature of the fish by analyzing with a neural network, and catches the fish by deep learning. A fish identification system that identifies fish, a fish feature storage unit that stores the fish feature of each fish species as feature data, and features of the fish by individually capturing the fish An image pickup unit that extracts the amount, a fish type determination unit that compares the feature amount with the fish type feature amount, determines the fish type of the catch, and extracts an individual feature amount that can be distinguished from the fish of the same type, and an identification number An identification number assigning unit for assigning an individual feature amount, an individual feature amount extracting unit for extracting an individual feature amount that can be distinguished from other catches of the same type from the feature amount, and an individual identification number for each catch as individual feature amount data Save by linking to Characterized in that it and a separate catch database.

  With the above configuration, the fish catches landed at the fishing port are analyzed for the features of the catches using artificial intelligence-based neural networks and deep learning, etc., individually given identification numbers and stored in the individual catch database, Various information about the catch can be centrally stored. Further, by extracting the individual feature quantities that can be distinguished from the same type of fish and assigning the identification numbers, the fish can be handled individually as if they were product numbers in industrial products. In other words, in order to make fisheries that have been manpowered so far more efficient, it is necessary to assign individual numbers to the catches and manage the catches that were originally living creatures as if they were industrial products. It is done. As a premise, by utilizing artificial intelligence, it becomes possible to assign individual identification numbers to the fish that was impossible until now, and various information about the fish can be centrally stored. Furthermore, as will be described later, by using this catch identification system and managing the quality of the catch for each individual catch, the quality control of the catch can be performed more finely. Furthermore, by utilizing this catch identification system, the distribution process can be traced for each catch regarding the logistics of the catch, and a safer and more reliable food can be obtained.

  In addition, establishing an individual catch database and storing all data related to individual catches as big data makes it possible to predict with higher accuracy when predicting by analyzing artificial catches using artificial intelligence. Is possible. Note that the image captured by the imaging unit includes not only image data from the camera but also data from a sensor or the like. Furthermore, image data from a video camera is also included.

  However, this catch identification system does not have to be applied to all catches landed at the fishing port. For example, it may be applied only to the representative fish species of the fishing port, the special fish species, or the large fish, and the small fish may be applied to the representative fish in the fish while being put in the above-described cage. . This is because small fish circulate in the basket and are sent to the final consumption area.

  In addition, the catch identification system may store the individual fish database as an individual aquaculture fish database by associating information on feeding and data on the status of fish in the aquarium with respect to the fish raised in the aquarium. preferable. This catch identification system is applied not only to the form of fishery that collects fish that live in the general coastal and nearby seas but also to aquaculture. And in aquaculture, since it is possible to cultivate catches in aquariums and ginger, the individual catch database can be used more than in the conventional fishery, and the effect can be expected. In particular, the state of the catch is affected by the content of the feed. Therefore, the contents of feeding are stored as data in the individual fish database and paired with data on the state of the fish in the aquarium, such as health status, activity, weight, etc. It is possible to analyze how it affects the state of.

  In addition, the catch identification system adjusts the water temperature in the aquarium, the water temperature measurement unit that measures the water temperature in the aquarium, the water quality measurement unit that measures the water quality in the aquarium, the water flow measurement unit that measures the water flow in the aquarium A water temperature adjustment unit, a water quality adjustment unit that adjusts the water quality in the aquarium, and a water flow adjustment unit that regulates the water flow in the aquarium, and the individual cultured fish database includes a water temperature measurement unit, a water quality measurement unit, and a water flow measurement unit It is preferable to store the measurement result and the information on the state of the catch in the tank in association with each other. In this way, utilizing the individual catch database of the present invention in aquaculture, information on the water temperature in the aquarium, information on the water quality, information on the water flow in the aquarium and the catches in the aquarium such as health status, activity By recording data relating to the state such as body weight in association with each other, it is possible to make a system for raising fish more effectively.

  In addition, the fish identification system extracts a video capture unit that captures the movement of fish moving around in the aquarium with a video camera, and extracts individual features of the fish from the video captured by the video image capture unit. It is preferable to include a fish identification unit that recognizes an identification number, and the moving image is stored in a cultured fish database of the cultured fish having the identification number. In this way, it is possible to finely grow the cultured fish by imaging the ecology of the cultured fish and monitoring each individual cultured fish.

  Moreover, it is preferable that the catch identification system extracts the foraging behavior of the catch from the moving image, predicts the state of the catch from the foraging behavior, and stores it in the individual catch database. In this way, it is more effective to record the information relating to the water quality in the aquarium, the information relating to the water flow in the aquarium, and the health condition of the catch in the aquarium by utilizing the individual database of the present invention in the aquaculture industry. System.

  Moreover, it is preferable that the fish identification system includes a fish theft monitoring unit that prevents the fish in the aquarium from being stolen using a moving image stored in the individually cultured fish database. It is preferable. In this way, it is more effective to record the information relating to the water quality in the aquarium, the information relating to the water flow in the aquarium, and the health condition of the catch in the aquarium by utilizing the individual database of the present invention in the aquaculture industry. System.

  Further, in the catch identification system, it is preferable that the individual catch database stores the individual feature quantity image data and the individual juvenile feature quantity data at the stage of the fry in the individual fry database. Thereby, in cultivation fisheries, it can be set as a more effective system by making a database and recording in the stage of a fry.

  In addition, the catch identification system is designed so that when a juvenile that has been released into the ocean grows and is landed as a catch in the state of an adult fish, Match and link. Thereby, when the grown juvenile is landed as a catch, the survival rate of the juvenile of the cultivated fish can be grasped by checking the identification number of the cultivated fish with the identification number of the juvenile fish.

  In addition, the fish management system using the artificial intelligence fish identification system includes a weight measuring unit for measuring the weight of the fish, a three-dimensional shape measuring unit for measuring the three-dimensional shape of the fish, and a catch item. And a color image capturing unit that captures the color tone of the body surface and a quality prediction unit that predicts the quality of the catch, and the individual catch database includes fish weight data, three-dimensional shape data, eye color data, and It is preferable to store the color data of the body surface as individual quality data in association with the fish identification number. Thereby, the sorting operation regarding the quality of the fish that has been relied on manually can be systematized by the fish management system.

  In addition, the catch management system has a three-dimensional shape measurement unit that calculates fish length data from the three-dimensional shape, sorts the catches according to a predetermined range of body lengths, and catches fish length data and catch weight data. Preferably, the body shape data of the fish is calculated from the data, and the individual fish database preferably stores the body length data and the body shape data in association with the fish identification number. Thereby, when sorting the fish, it is possible to sort the fish that are fat and those that are thin without depending on the hand, and to arrange the fish for each price range.

  In the catch management system, the three-dimensional shape measurement unit calculates the volume data of the catch from the three-dimensional shape, calculates the density data of the catch from the weight data measured by the weight measurement unit, and the quality prediction unit , Predict the degree of meat quality in the catch from the volume data and density data as meat quality data, and the individual fish database stores the volume data, density data, and meat quality data as individual quality data in association with the fish identification number It is preferable. This is said that according to the “connoisseur” of the buyer, the catch has been stiff after death for 24 hours after death, and is said to be in a “good” state. If this exceeds 24 hours, the postmortem stiffness will end and it will soften rapidly. Therefore, the hardness of the fish is calculated from the volume of the catch from the three-dimensional shape by the three-dimensional shape measuring unit and the weight measured by the weight measuring unit. In other words, the freshness of the fish can be predicted from the density of the fish with respect to the “tightness of the meat of the fish”.

  In the fish management system, the color imaging unit measures the data relating to the color of the eye of the fish or the color of the body surface of the fish, and the quality prediction unit determines the fish from the data relating to the color of the eye and the color of the body surface. Preferably, the individual fish database stores the eye color data, body surface color data, and freshness data in association with the fish identification number. According to the “connoisseur” of the buyer, it is said that the fresh fish looks transparent, while the old fish are white and cloudy. Therefore, the freshness of the fish can be predicted by observing the color tone of the fish. That is, the freshness of the fish can be predicted by accumulating the eye color pattern and eye shape pattern of the fish species in the database unit and pattern matching with the eye color and shape of the fish. .

  The fish management system includes a male and female feature extraction unit that extracts physical characteristics of male and female for a specific fish catch, and a male and female determination unit that compares male and female characteristics of the fish for each fish species to determine the male and female of the fish catch. It is preferable that the individual fish database stores information relating to the sex of the fish in association with an identification number for each fish. Thereby, it is possible to quickly sort fish species whose prices vary greatly depending on sex.

  The fish distribution system using the artificial intelligence catch identification system acquires individual feature image data and individual feature data at the fish distribution stage, and individually captures the individual feature image data and individual feature data. The identification number is identified by collating with the individual feature image data and individual feature data stored in the product database, tracking the logistic history for each catch, and the individual fish database links the logistic history data to the identification number. It is preferable to store them. Based on the distribution history data linked to this identification number, it is possible to specifically grasp the distribution quantity of the fish catch, and it is possible to perform a specific simulation of distribution in each fishing port.

  The fish distribution system is a fish quantity calculation unit that calculates the quantity of fish for each fish type based on the identification number, and calculates the proportion of fish from aquaculture or cultivated fishery in the quantity of fish for each fish type. A fishery composition ratio calculation unit, and a fish distribution grasping unit that traces the route of the fish distribution from the distribution history data stored in the individual fish database. It is preferable to update to Thereby, it is possible to accurately grasp the quantity of fishery logistics, and a specific simulation of logistics at each fishing port is possible.

  In the fish distribution system, it is preferable that the fish distribution grasping unit estimates the demand for the fish in the final consumption area from the route of the fish distribution. As a result, fluctuations in demand in the final consumption area can be materialized for each catch and used as a study material for distribution strategies at each fishing port.

  In the catch logistics system, it is preferable that the catch logistics grasping unit predicts the catch amount of the aquaculture fishery or the cultivated fishery from the fluctuation of the catch amount including the aquaculture fishery or the cultivated fishery for each fish species. Thereby, about the fishing method, the possibility and urgency of the shift to aquaculture or cultivated fishery can be immediately determined and used as a guideline.

  In the fish distribution system, it is preferable that the fish distribution grasping unit performs simulation of the optimum transportation route and transportation means to the consumption area of the fish using individual recognition by the fish identification system. In this way, the catch distribution system traces the catch using the individual recognition of the catch, and if a catch quality problem occurs, it is immediately clear where the problem occurred. , Can quickly take countermeasures and improvement measures. Transparency in the distribution of fish catches can be secured, and the safety and security of foodstuffs can be guaranteed to consumers.

(Net auction)
In the fish distribution system, it is preferable that the fish distribution grasping unit holds a net auction instead of the auction in the fish market and uses a communication network for the distribution of the fish. This allows anyone to participate in the catch of fish anytime and anywhere. Anyone can purchase the fish with the identification number individually, and can easily distribute the fish in large quantities in a small lot.

  As described above, according to the artificial intelligence catch identification system, management system, and distribution system according to the present invention, the artificial intelligence can be used to identify individual catches and track the distribution history data. It is possible to provide an identification system, a fish management system for systematizing the quality control of the identified fish, and a fish logistics system for systematizing the logistics of the recognized fish.

It is a block diagram showing a schematic structure of one embodiment of a fish identification system. It is a block diagram which shows schematic structure of one Embodiment of the sorting method of the fished fish which was landed. It is explanatory drawing which shows schematic structure of one Embodiment regarding the procedure which extracts the feature-value of a fish catch from the image analysis by artificial neural networks etc., and memorize | stores it in a fish kind feature-value memory | storage part. It is explanatory drawing which shows schematic structure of one Embodiment regarding the procedure which compares the feature-value of a fish catch with fish-type feature-value by artificial intelligence deep learning etc., and determines a fish species. It is explanatory drawing which shows schematic structure of one Embodiment regarding the procedure which discriminate | determines from the catch of the same kind by artificial neural network, deep learning, etc., and provides an identification number. It is a block diagram which shows schematic structure of one Embodiment regarding the structure of aquaculture fishery, and the structure of an individual catch database. It is a block diagram which shows schematic structure of one Embodiment regarding the structure of cultivation fisheries, and the structure of an individual catch database. It is a block diagram showing a schematic structure of one embodiment of a fish management system. It is a block diagram showing a schematic structure of one embodiment of a fish distribution system.

  Below, it demonstrates in detail about the catch identification system 1 which concerns on this invention using drawing. FIG. 1 is a block diagram showing a schematic configuration of one embodiment of the catch identification system 1. FIG. 2 is a block diagram showing a schematic configuration of one embodiment of the method for sorting the caught fish 13. FIG. 3 shows a schematic configuration of one embodiment relating to a procedure for extracting the feature quantity 15 of the fish 13 from the image analysis by the neural network 11b of the artificial intelligence 11a and storing it in the fish species feature quantity storage unit 7 as the feature quantity data 7n. Is shown in an explanatory diagram. FIG. 4 is an explanatory diagram showing a schematic configuration of one embodiment relating to a procedure for determining the fish species 14 by comparing the feature quantity 15 of the catch 13 and the fish species feature quantity 15b by the deep learning 11c of the artificial intelligence 11a. . FIG. 5 shows a schematic configuration of one embodiment relating to a procedure for discriminating the fish 13 of the same kind by the neural network 11b and the deep learning 11c of the artificial intelligence 11a and assigning the identification number 10.

(Configuration of catch identification system)
As shown in FIG. 1, the target of the present fish identification system 1 is a coastal fishery 3a, a near-shore fishery 3b, an aquaculture fishery 3c, and a cultivated fishery 3d. The coastal fishery 3a and the near-shore fishery 3b are conventionally known as fisheries since they are fisheries that have been generally performed. In this specification, the fish identification system 1 in the conventional fishery will be described, and then the fish identification system 1 in the case of the aquaculture fishery 3c and the cultivated fishery 3d will be described focusing on differences from the conventional fishery. Here, the offshore fishery targeting the fishing grounds of the countries of the world is out of the scope of the present invention.

  In the conventional fishery, the catch 13 caught by a fishing boat is transported to a nearby fishing port and landed. At the fishing port, the caught fish 13 is conveyed to the market by a fish conveyor 17 such as a belt conveyor. The fish 13 is imaged by the imaging unit 4 at the time of transportation. The imaging unit 4 may be provided with not only the imaging camera 4a but also a sensor 4b. This sensor 4b measures the surface temperature, odor, color tone, etc. of the catch 13. Moreover, the video camera 4c may be provided and the movement of the catch 13 may be imaged. In particular, in the aquaculture 3c, it is used to image the movement of the fish 13 in the aquarium or ginger.

(Fish species feature extraction and storage)
As shown on the right side of the drawing of FIG. 1, for the fish 13 that may be caught in the fishing port, the fish species characteristic amount 15 b for each fish species 14 is based on the image captured by the imaging unit 4. Extracted and stored. The fish species feature value 15b is obtained by analyzing the image data of a large amount of fish 13 caught as big data 11d and extracted by the fish species feature value extraction unit 6b as fish species feature value image data 5b. The process of extracting the fish species feature 15b is executed mainly by the neural network 11b of the artificial intelligence 11a, and a specific method thereof is shown in FIG.

  The fish species feature amount 15b of each fish species 14 extracted by the fish species feature amount extraction unit 6b is stored in the fish species feature amount storage unit 7. For example, saury feature value data 7a, squid feature value data 7b, scallop feature value data 7c, eel feature value data 7d,..., Fish species 14 feature value data 7n are stored. Here, the landed fish 13 is not limited to moving on the fish carrier 17 such as a belt conveyor, but is stored in a fish tank or a container that accommodates the fish 13 such as “ike”. The fish species feature image data 5b may be taken by the sensor 4b or the video camera 4c. Moreover, these stored fish species feature image data 5b may be prepared in advance before the fish identification system 1 operates, and is stored when the fish identification system 1 is operating. The fish species feature image data 5b may be updated to the latest data.

(Fish species determination)
As shown on the left side of FIG. 1, a method of sorting the fish catches 13 by the fish species 14 is shown. For example, when the fish 13 is conveyed to the market by a fish conveyor 17 such as a belt conveyor, the individual feature amount image data 5a is imaged by the imaging unit 4. Then, the individual feature quantity 15a is extracted by the individual feature quantity extraction unit 6a. The imaging unit 4 may capture the individual feature amount image data 5a and use the individual feature amount image data 5a as fish species feature amount image data 5b at the same time. The extracted individual feature 15a is compared with the fish species feature 15b of each fish species 14 stored in the fish species feature storage 7 by the fish species determination unit 8. Then, the fish species 14 of the catch 13 is determined. This comparison is executed mainly by the deep learning 11c of the artificial intelligence 11a, and a specific method thereof is shown in FIG.

(Granting identification number)
When the fish type 14 of the fish 13 is determined by the fish type determination unit 8, the identification number giving unit 9 gives the identification number 10 to the fish 13. For example, a serial number in the same type of catch 13 is given together with data such as the year, day of the week, fish type, fished fishing ground, fishing boat, and fishing port.

  Specifically, when a plurality of saury are caught, among the fish species 14 called saury, saury 1, saury 2, saury 3. That is. As a result, the catch of saury is accurately determined, the quality of each saury is individually controlled, and the distribution route of each saury is clarified.

(Data storage)
In the individual catch database 12, the individual feature amount image data 5a and the individual feature amount data 2 are stored together with the identification number 10 and the fish type 14. That is, each catch 13 caught at the fishing port stores data such as the year, day of the week, fish species 14, fished fishing ground, fishing boat, fishing port, etc. caught by the identification number 10, and the individual feature 15a is determined. The individual feature amount image data 5a that is the basis and the individual feature amount data 2 obtained by converting the plurality of individual feature amounts 15a into data collectively are stored. Then, data on the aquaculture fishery 3c and the cultivated fishery 3d are added to the individual fish catch database 12. Furthermore, data in the catch management system 40 and the catch logistics system 50 are added, and one database is completed, which can be analyzed as the big data 11d with higher accuracy by the artificial intelligence 11a.

(Sorting work)
As shown in FIG. 2, the catch 13 determined by the fish type determination unit 8 is transported to the sorting unit 18 by a fish transporter 17 such as a belt conveyor, for example. Sorted into juvenile fish. The sorting unit 18 includes a fish type sorting unit 18a, a male / female sorting unit 18b, and an immature fish sorting unit 18c. Data such as the length of the fish 13, the sex of the fish 13, the age of the fish 13, etc. necessary for sorting other than the fish species 14 are retrieved from the individual fish database 12 and sent from the determination information transmission unit 8 b. It is transmitted to the sorting unit 18. That is, the sorting work that has been relied on manually can be automated by using the catch identification system 1. In addition, the determination information transmission part 8b is not only sorted according to the type of fish 14, whether it is female or male, adult or juvenile, whether it is a wound, whether it is sick, whether it is deformed, whether it is a deep sea fish Classification may be made according to whether the child has children or not. That is, the fish 13 having characteristics that affect the commercial value and price of the fish 13 can be selected.

(Image analysis using artificial intelligence)
FIG. 3 shows a procedure for extracting the feature value 15 of the catch 13 from the image analysis by the neural network 11b of the artificial intelligence 11a and storing it in the fish species feature value storage unit 7. The artificial intelligence 11a selects all the fish species 14 that may be caught at the fishing port from the big data 11d of the catch 13 and selects them as fish species α ₁ , fish species α ₂ , fish species α ₃ , fish species. α ₄ ... Fish species α _n . Image data 5b ₁ , image data 5b ₂ , image data 5b ₃ , image data 5b ₄ ,... Image data 5b _n are used as the fish species feature amount image data 5b captured by the imaging unit 4 for these fish species 14. . The fish type feature amount extraction unit 6b, for each piece of image data 5 from the fish type feature amount image data 5b, for example, has a characteristic dorsal fin (A) in the fish type 14 as compared with other fish types 14. , Tail fin (B), belly fin (C)... For fish alpha _1, species characteristic amount 15b of dorsal fin (A) is A1, the fish species characteristic amount 15b of the tail fin (B) is B1, fish feature amount 15b of ventral fin (C) is C1 It is judged that. For fish species alpha _2, species characteristic amount 15b of dorsal fin (A) is A2, a species characteristic amount 15b of the tail fin (B) is B1, fish feature amount 15b of ventral fin (C) is C2 It is judged that. For fish species alpha _3, species characteristic amount 15b of dorsal fin (A) is A1, a species characteristic amount 15b of the tail fin (B) is B1, fish feature amount 15b of ventral fin (C) is C2 It is judged that. For the fish species α _n , the fish species feature 15b of the dorsal fin (A) is A1, the fish species feature 15b of the tail fin (B) is B1, and the fish species feature 15b of the belly fin (C) is C2. It is judged that. The characteristic amount 15 of this fish species 14 is not limited to the dorsal fin (A), tail fin (B), belly fin (C), but may be a buttocks fin or a chest fin, and may be a body shape, length, eyes, mouth, etc. What is necessary is just the characteristic which is characteristic compared with the other fish species 14, and can be differentiated.

  The artificial intelligence 11a uses, for example, a neural network 11b or the like as a criterion for determining the fish species 14, for example, “The fish species feature 15b of the dorsal fin (A) is A1. However, for females, A2 is not A1, but A2 is the fish species feature 15b. ”“ The fish species feature 15b of the tail fin (B) is B1 is an essential condition for the fish species 14. "The fish species characteristic amount 15b of the stomach fin (C) is an essential condition for the fish species 14, and C1 is another type of fish species characteristic amount 15b." The artificial intelligence 11a stores the fish species (α) feature amount data 7α in the fish species feature amount storage unit 7.

(Comparison of image data by artificial intelligence)
FIG. 4 shows a procedure for determining the fish species 14 by comparing the feature quantity 15 of the catch 13 and the fish species feature quantity 15b by the deep learning 11c of the artificial intelligence 11a as the first stage. The picked fish 13 (β ₁ , β ₂ , β ₃ ,... Β _n ) is imaged by the imaging unit 4. And the fish species feature 15b of the catch 13 (β ₁ , β ₂ , β ₃ ,... Β _n ) is compared with each fish species feature 15b stored in the fish species feature storage unit 7; The fish species 14 of each catch 13 (β ₁ , β ₂ , β ₃ ,... Β _n ) is determined depending on which fish species 14 corresponds to. In FIG. 4, since the characteristic amount 15 matches the fish type X2, it can be determined that the fish type 14 is “fish type X2”. That is, if the “determination criteria for fish species 14” described above is applied to the catch 13, the fish species 14 is determined.

FIG. 5 shows a procedure for identifying the fish 13 of the same kind and assigning the identification number 10 by the neural network 11b and the deep learning 11c of the artificial intelligence 11a as the second stage. First, the same kind of fish 14 is selected. Then, as their individual characteristic quantity image data 5a, the image data 5a _1, the image data 5a _2, the image data _5a 3 · · · image data 5a _n collected. The individual feature amount extraction unit 6a extracts the individual feature amount 15a from these data. Unlike the fish species feature 15b, the individual feature 15a is a feature that is characteristic and can be differentiated compared to other fish 13 of the same species. For humans, it is like a fingerprint. The artificial intelligence 11a, for example, determines that the individuality of each catch 13 appears in the shape of the dorsal fin (A) among the dorsal fin (A), tail fin (B), belly fin (C). To do. That is, the artificial intelligence 11a searches for a part having an element that can be differentiated from other fish 13. Then, the individual feature quantity 15 a different from the other fish 13 of the same kind is stored as the individual feature quantity data 2. Further, the individual feature amount image data 5a is also stored together as an image for determining the individual feature amount data 2. And this operation | work is repeated and the identification number 10 is assign | provided respectively with respect to all the fish 13 of the fished fish 13 ((beta) ₁ , (beta) ₂ , (beta) ₃ , ... (beta) _n )) landed. In FIG. 4, it is determined that A1, B1, and C2 are all identifiable as individual feature values 15a. Then, 0001 is assigned as the identification number 10, and 0002, 0003, and 0004 are assigned to each.

  In the first stage, the artificial intelligence 11a compares the individual feature 15a with the fish species feature 15b by deep learning 11c. For example, the neural network 11b or the like can determine that the fish species 14 is based on the fact that the fish species feature 15b of the dorsal fin (A) is A1, which is a feature for males but A2 for females instead of A1. Is the fish species feature 15b. ”“ It is an essential condition for the fish species 14 that the fish species feature 15b of the tail fin (B) is B1. ”“ The fish of the belly fin (C) As for the species feature 15b, C2 is an indispensable condition for the fish species 14, and C1 is a feature of another species. "If the element indicates an indispensable condition, the fish species 14 to decide.

  In the second stage, the artificial intelligence 11a specifies the individual feature 15a of the catch 13 by the neural network 11b and the deep learning 11c. For example, when it is determined that the fish species feature 15b of the dorsal fin (A) has individuality, the fish species feature 2 is used as the individual feature data 2 so that a human fingerprint is collected for the fish species feature 15b of the dorsal fin (A). Save in the quantity storage unit 7. And the identification number 10 is provided to the catch 13. Therefore, the identification number 10 includes individual characteristic amount data 2 indicating the fish species 14 that is differentiated from other fish species 14 and unique features that are differentiated from the fish 13 of the same species in the fish species 14. Are stored together with the individual feature amount data 2 having.

(Aquaculture)
In FIG. 6, the schematic structure of one embodiment regarding the structure of the cultured fish industry 3c and the structure of the individual cultured fish database 12a is shown. In the aquaculture 3c, an aqua tank is provided on the land to cultivate aquaculture fish that becomes the catch 13. Here, the aquarium includes ginger.

(Water temperature, water quality, water flow)
The aquaculture fishery 3c includes a water temperature measurement unit 20, a water quality measurement unit 26, a water temperature adjustment unit 21, a water quality adjustment unit 22, a water flow measurement unit 27, a water flow adjustment unit 23, a moving image imaging unit 32, a catch identification unit 33, and a feed management unit 24. The farming fish management unit 25 and the fish theft monitoring unit 34 are configured. The water temperature measuring unit 20 measures the water temperature in the water tank and sends the measurement result to the water temperature adjusting unit 21. The water quality measurement unit 26 measures information about the water quality in the aquarium and sends the measurement result to the water quality adjustment unit 22. The water flow measurement unit 27 measures information related to the water flow in the water tank and sends the measurement result to the water flow adjustment unit 23. Water temperature, water quality and water flow are factors that have a significant impact on the growth of farmed fish. The cultured fish management unit 25 monitors the health of the cultured fish and gives an instruction regarding feeding to the feeding management unit 24. The individual cultured fish database 12a stores the measurement results of the water temperature measurement unit 20, the water quality measurement unit 26, and the water flow measurement unit 27 in association with the data on the state of the cultured fish in the aquarium.

(Individually cultured fish database)
The individual cultured fish database 12a stores the cultured fish identification number 10b assigned by the identification number assigning unit 9 to the cultured fish bred in the aquarium. Moreover, the individual cultured fish database 12a stores the individual cultured fish feature data 2a extracted by the individual feature data extraction unit 6a in association with the cultured fish identification number 10b. Then, the feeding management unit 24 associates the information regarding the given food with the cultured fish identification number 10b as the feeding data 30a and stores it in the individual cultured fish database 12a. In addition, the individual cultured fish database 12a uses the measurement results of the water temperature measurement unit 20, the water quality measurement unit 26, and the water flow measurement unit 27 as water temperature data 31a, water quality data 31b, and water flow data 31c, and data on the state of the cultured fish in the aquarium. Associate and save. Moreover, the individual cultured fish database 12a stores the feeding behavior moving image data 30b of the cultured fish having the cultured fish identification number 10b in the individual cultured fish database 12a.

(Video of farmed fish)
The moving image capturing unit 32 captures the motion of the cultured fish migrating in the aquarium with the video camera 4c. This is because the health status of the cultured fish can be predicted from the feeding behavior of the cultured fish migrating in the aquarium. In addition, it is possible to find a farmed fish that behaves differently from other farmed fish and find out the cause. The fish identification unit 33 extracts the individual characteristic amount 15a of the cultured fish from the moving image captured by the moving image capturing unit 32 and recognizes the cultured fish identification number 10b of the cultured fish.

(Theft of cultured fish)
Further, the fish theft monitoring unit 34 can prevent the cultured fish in the aquarium from being stolen by monitoring a moving image stored in the individually cultured fish database 12a. This is because the cultured fish grown in the aquarium may be easily stolen, but the theft is instantly revealed by monitoring the number of cultured fish in the aquarium using the artificial intelligence 11a. In addition, the fish theft monitoring unit 34 can monitor the theft with respect to the equipment around the aquarium as well as the cultured fish.

  The individual fish database 12 stores the individual feature amount image data 5a and the individual feature amount data 2 at the stage of fry in an individual fry database 12c provided in the individual fish database 12. In addition, the individual fish database 12 is used to identify and link the fry identification number with the cultivated fish identification number when the fry released to the ocean grows and is landed as a catch in the adult state. Attach.

(Cultivated fishery)
In FIG. 7, the schematic structure of one embodiment regarding the structure of the cultivation fishery 3d and the structure of the individual fish catch database 12 is shown. The cultivated fish industry 3d includes a cultivated fish verification unit 38 and a cultivated fish breeding unit 37. The cultivated fish verification unit 38 compares the fry and the cultivated fish 39 when the fry is released into the ocean and grows and landed as the cultivated fish 39 in an adult state. The cultivated fish breeding unit 37 monitors the growth of fry.

  The individual catch database 12 stores the individual feature quantity image data 5a at the stage of fry and the individual feature quantity data 2 that is the basis of the individual feature quantity image data 5a in the individual catch database 12c provided in the individual catch database 12. save. By doing so, the fry and the cultivated fish 39 can be collated when the fry becomes an adult fish. In addition, the individual fish database 12 collates and links the juvenile identification number and the cultivated fish identification number when the juvenile released to the ocean grows and is landed as a catch 13 in an adult state. Thereby, when the grown fry is landed as the catch 13, the survival rate of the fry of the cultivated fish 39 can be grasped by comparing the cultivated fish identification number 10c of the cultivated fish 39 with the fry identification number 10d.

(Catch management system)
FIG. 8 shows a schematic configuration of one embodiment of the catch management system 40 according to the present invention. The fish management system 40 includes a weight measuring unit 42 for measuring the weight of the fish 13, a three-dimensional shape measuring unit 43 for measuring the three-dimensional shape of the appearance of the fish 13, and the color of the eyes and body surface of the fish 13. It is comprised from the color tone imaging part 44 which image | photographs, and the quality prediction part 45 which estimates the quality of the catch 13. The individual catch database 12 uses the weight data 49 a of the catch 13, the three-dimensional shape data 49 b, the eye tone data 49 c, and the body surface tone data 49 d as the individual quality data 49, and the identification number 10 of the catch 13. Save it in association with. In this way, the fish management system 40 performs quality management such as the freshness of the fish 13 based on the form of the fish 13 using the individual recognition of the fish identification system 1.

(Assortment of fish catch length)
The three-dimensional shape measurement unit 43 calculates the body length data 49e of the fish 13 from the three-dimensional shape. It is used when sorting the catch 13 for each body length within a predetermined range. By utilizing this body length data 49e for sorting work, body data 49f relating to whether the catch 13 is fat or thin can be calculated from the weight data 49a weighed by the weight weighing unit 42. Then, the individual fish database 12 stores the body length data 49e and the body shape data 49f, which are the individual quality data 49, in association with the identification number 10 of the fish 13 in the individual fish database 12.

(Assortment of fish catches)
The three-dimensional shape measuring unit 43 calculates volume data 49g from the three-dimensional shape of the catch 13. The density calculation unit 48 calculates density data 49h that is the hardness of the catch 13 from the three-dimensional shape data 49b and the weight data 49a measured by the weight measurement unit 42. Then, the quality prediction unit 45 predicts and sorts the tightness of the meat quality in the catch 13 from the volume data 49g and the density data 49h. That is, whether the fish 13 is thin or thin is sorted according to the numerical value. The catch 13 is said to be in a “good” state because it remains rigid after death for 24 hours after death. If this exceeds 24 hours, the postmortem stiffness will end and it will soften rapidly. Accordingly, the density data 49 h of the fish 13 is obtained from the volume data 49 g of the fish 13 calculated from the three-dimensional shape measured by the three-dimensional shape measuring unit 43 and the weight data 49 a measured by the weight measuring unit 42. Then, the hardness data 49k of the catch 13 is calculated from the density data 49h. That is, the freshness of the fish 13 can be predicted from the density of the fish 13 with respect to the “tightness of the meat of the fish 13”.

(Assortment of freshness of fish)
The color image capturing unit 44 measures the color data 49c of the eye of the fish 13 or the color data 49d of the body surface of the fish 13. The quality prediction unit 45 predicts the freshness of the catch 13 from the eye color data 49c and the body surface color data 49d. The individual fish database 12 stores data relating to the color of the eyes, data relating to the color of the body surface, and prediction of freshness in association with the fish identification number 10.

(Assortment of fish and fish)
The fish management system 40 determines the sex of the fish 13 by comparing the male and female features 46 for extracting the physical characteristics of the male and female for the specific fish 13 and the characteristics of the fish 13 for each fish species 14. And a male / female determination unit 47. The individual fish database 12 stores the sex 13 of the fish 13 in association with the identification number 10 for each fish 13. Thereby, it can sort quickly about the fish species 14 from which a price differs greatly by sex.

(Catch history)
In order to check the history of the fish 13 at the logistics stage of the fish 13, first, the individual feature image data 5 a and the individual feature data 2 are acquired using the imaging camera 4 a or the like. Then, the acquired individual feature amount image data 5 a and individual feature amount data 2 are collated with the individual feature amount image data 5 a and the individual feature amount data 2 stored in the individual catch database 12. If the identification number 10 or the like matches, the fish 13 can be specified. In this manner, the distribution history data 24 is tracked for each catch 13, and the latest information of the distribution history data 24 is associated with the identification number 10 and stored in the individual catch database 12.

(Catch logistics system)
The fish 13 that has been auctioned off at the fish market is transported to the central wholesale market by the middleman and brought into the wholesale market of the consumption area through various distribution routes. The final catch 13 is sold to consumers at a general fish shop or cooked at a restaurant and provided to consumers. In this way, the catch 13 finally reaches the consumer through various routes. In the logistics of the fish 13, transparency regarding the logistics of the fish 13 can be secured by individually identifying the fish 13 in the fish identification system 1, and the safety and security of foodstuffs can be guaranteed for consumers.

  FIG. 9 is a block diagram schematically illustrating the configuration of one embodiment of the fish distribution system 50. The fish distribution system 50 according to the present invention includes a fish quantity calculation unit 51 that calculates the quantity of the fish 13 for each fish species 14 based on the identification number 10, and aquaculture that occupies the quantity of the fish 13 for each fish species 14. A fishery composition ratio calculating unit 52 that calculates a ratio of the fish 13 by the fishery 3c or the cultivated fishery 3d, and a fish logistics grasping unit that grasps the logistics of the fish 13 from the logistics history data 54 stored in the individual fish database 12. 53. Here, the distribution history data 54 is data that clarifies what route is followed by what trader in the process from the fish market to the final consumer. The catch 13 that is a perishable food product decreases in freshness over time and deteriorates such as corrosion. In addition, a person who ate the catch 13 may be poisoned by bacteria or viruses, or parasites may enter the human body, and the trader handling the catch 13 in the catch logistics system 50 has a problem. It may become. In order to quickly elucidate these causes and minimize damage, the present fish distribution system 50 is utilized.

  The fish distribution system 50 reads data from the individual fish database 12 created by the fish identification system 1 and the fish management system 40. The individual catch database 12 includes an individual farmed fish database 12a, an individual cultivated fish database 12b, and an individual fry database 12c.

  As an effect of the fish distribution system 50, the distribution route of the fish 13 can be grasped from the individual fish database 12 stored in the fish 13. That is, “whose possession of the fish 13 from what time to what time” is recorded for each fish 13. Therefore, if the catch logistics grasping | ascertainment part 53 traces the record, the trader who handled the catch 13 in the catch logistics system 50 will become clear easily.

(Demand in the final consumption area)
Another effect of the fish distribution system 50 is that the demand for the fish 13 in the final consumption area can be estimated. That is, the catch logistics grasping part 53 keeps track of which logistics route has been traced from the fish history to the final consumption area for a certain catch 13 from the logistics history data 54. Accordingly, analysis of the data reveals which fish species 14 are transported to the consumption area. In addition, by introducing the fish distribution system 50 to the national fish market, it is possible to easily grasp the distribution situation of the nationwide fish 13.

(Catch prediction for each catch method)
The fish catch logistics grasping unit 53 adjusts the individual fish species 14 by predicting the catch of the aquaculture fishery 3c or the cultivated fishery 3d from the fluctuation of the catch including the aquaculture fishery 3c or the cultivated fishery 3d. As for the coastal fishery 3a and the near sea fishery 3b, the amount of catches has decreased due to the various problems described above, and a shift to the aquaculture fishery 3c or the cultivated fishery 3d is expected. The aquaculture fishery 3c or the cultivated fishery 3d can be promoted by including the catch of the aquaculture fishery 3c or the cultivated fishery 3d in the market catch. In addition, by introducing this system to the national fish market, it is possible to grasp the actual situation of the nationwide aquaculture fishery 3c or the cultivated fishery 3d.

(Logistics simulation of fish catch)
The catch physical distribution grasping part 53 can perform simulation of the optimum transport route and transport means to the final consumption place of the catch 13 by using the individual recognition by the catch identification system 1. That is, the catch logistics grasping part 53 installs the imaging camera 4a in the central wholesale market where the catches 13 gather, or the wholesale market at the midpoint, etc., individually captures the catches 13 that are distributed, and the identification number 10 or the like From this, it is possible to grasp the distribution status of the catch 13. Based on this data, the optimum transport route and transport means to the final consumption place can be examined. For example, it is possible to consider whether or not a useless transportation route is used until a catch 13 that has been released from a certain fish market reaches a final consumer, or whether or not transportation means such as a truck or a freight car is appropriate. In addition, by introducing this system in the national fish market, nationwide logistics simulation is possible.

  The configuration, shape, size, and arrangement relationship described in the above embodiments are merely schematically shown to the extent that the present invention can be understood and implemented. Therefore, the present invention is not limited to the described embodiments, and can be variously modified without departing from the scope of the technical idea shown in the claims.

1 catch identification system, 2 individual feature data, 2a individually cultured fish feature data, 2b individually cultivated fish feature data, 2c individual fry feature data, 2d individual fry feature data, 3a coastal fishery, 3b inshore fishery, 3c Aquaculture, 3d Cultivation Fishery, 4 Imaging Unit, 4a Imaging Camera, 4b Sensor, 4c Video Camera, 5 Image Data, 5a Individual Feature Value Image Data, 5b Fish Species Feature Value Image Data, 6a Individual Feature Value Extraction Unit, 6b Fish species feature extraction unit, 7 fish species feature storage unit, 7a saury feature data, 7b squid feature data, 7c scallop feature data, 7d eel feature data, ..., 7n feature data Data, 8 fish type determination unit, 8b determination information transmission unit, 9 identification number assigning unit, 10 identification number, 10b cultured fish identification number, 10c cultivated fish identification number, 10d fry fish identification number No., 11a Artificial Intelligence, 11b Neural Network, 11c Deep Learning, 11d Big Data, 12 Individual Catch Database, 12a Individual Cultured Fish Database, 12b Individual Cultivated Fish Database, 12c Individual Fry Database, 13 Fish Catch, 14 Fish Species, 15 Features, 15a Individual features, 15b Fish species features, 16 Identification number transmission unit, 17 Fish transporter, 18 Sorting unit, 18a Fish species sorting unit, 18b Male and female sorting unit, 18c Underage fish sorting unit, 20 Water temperature measurement Part, 21 water temperature adjustment part, 22 water quality adjustment part, 23 water flow adjustment part, 24 feeding management part, 25 aquaculture fish management part, 26 water quality measurement part, 27 water flow measurement part, 30a feeding data, 30b feeding behavior video data, 31a water temperature Data, 31b water quality data, 31c water flow data, 32 video imaging unit, 33 Catch identification section, 34 Catch theft monitoring section, 37 Cultivated fish breeding section, 38 Cultivated fish verification section, 39 Cultivated fish, 40 Catch management system, 41 Individual quality data, 42 Weight measurement section, 43 Three-dimensional shape measurement section , 44 color tone imaging unit, 45 quality prediction unit, 46 gender characteristic unit, 47 gender determination unit, 48 density calculation unit, 49 individual quality data, 49a weight data, 49b three-dimensional shape data, 49c eye tone data, 49d body surface Tone data, 49e body length data, 49f body shape data, 49g volume data, 49h density data, 49k hardness data, 50 fish distribution system, 51 fish quantity calculation section, 52 fishery composition ratio calculation section, 53 fish catch distribution grasp Department, 54 Logistics history data.

Claims (19)

A fish identification system that selects an image of a fish from big data, analyzes it by a neural network, extracts the feature of the fish, and identifies the fish by deep learning,
A fish species feature storage unit that stores fish species features for each fish species of the fish catch as feature data;
An imaging unit that individually images the fish and extracts the feature amount of the fish;
A fish type determination unit that compares the characteristic amount with the fish type characteristic amount and determines the fish type of the catch;
An identification number assigning unit that extracts individual features that can be distinguished from the same kind of catch and assigns an identification number;
An individual feature amount extraction unit that extracts an individual feature amount that can be distinguished from other catches of the same type from the feature amount;
A fish product identification system using artificial intelligence, comprising: an individual fish database that stores the individual feature data as individual feature data in association with the identification number for each fish product.   The artificial intelligence fish identification system according to claim 1, wherein the individual fish database includes information regarding feeding and data regarding a state of the fish in the aquarium, with respect to the fish bred in an aquarium. A fish identification system using artificial intelligence, characterized in that it is stored as an individual cultured fish database in association with each other.   It is the catch identification system by the artificial intelligence of Claim 1 or 2, Comprising: The water temperature measurement part which measures the water temperature in the said tank, The water quality measurement part which measures the water quality in the said tank, The water flow in the said tank A water flow measuring unit that measures the water temperature in the aquarium, a water temperature adjusting unit that adjusts the water quality in the aquarium, and a water flow adjusting unit that regulates the water flow in the aquarium, The individually cultured fish database stores the measurement results of the water temperature measurement unit, the water quality measurement unit, and the water flow measurement unit in association with information on the state of the catch in the tank. Artificial intelligence catch identification system.   The artificial intelligence fish identification system according to claim 3, wherein a moving image capturing unit that captures an operation of the fish moving around in the aquarium with a video camera, and a moving image captured by the moving image capturing unit, A fish identification unit that extracts the individual feature of the fish and recognizes the identification number of the fish, and the moving image is stored in the individually cultured fish database of the cultured fish having the identification number. A catch identification system using artificial intelligence.   5. The artificial intelligence fish identification system according to claim 4, wherein the foraging behavior of the catch is extracted from the moving image, and the state of the catch is predicted from the foraging behavior, and the individually cultured fish database A catch identification system using artificial intelligence, characterized by being stored in   5. The artificial intelligence fish identification system according to claim 4, further comprising a fish theft monitoring unit that prevents the fish in the tank from being stolen by using the video stored in the individually cultured fish database. A catch identification system using artificial intelligence.   2. The artificial intelligence catch identification system according to claim 1, wherein the individual catch database stores the individual feature image data and the individual juvenile feature quantity data at the stage of fry in the individual fry database. A catch identification system using artificial intelligence.   The artificial intelligence fish identification system according to claim 7, wherein the individually cultivated fish database is a cultivated fish collation when fry released to the ocean grows and is landed as the fish in an adult state. An artificial intelligence catch identification system characterized in that a fry identification number and a cultivated fish identification number are collated and linked to a part. A catch management system using the artificial intelligence catch identification system according to any one of claims 1 to 8, further comprising: a weight measuring unit for weighing the catch; and A three-dimensional shape measuring unit that measures a three-dimensional shape, a color image capturing unit that images the color of the eyes and body surface of the fish, and a quality predicting unit that predicts the quality of the fish, the individual fish The database stores the weight data, the three-dimensional shape data, the eye tone data, and the body surface tone data of the catch as individual quality data in association with the identification number of the catch. Intelligent fish catch management system.   The artificial intelligence fish management system according to claim 9, wherein the three-dimensional shape measurement unit calculates body length data of the fish from the three-dimensional shape, and the fish for each body length within a predetermined range. And the body data of the fish is calculated from the length data of the fish and the weight data of the fish, and the individual fish database uses the body length data and the body data as the identification of the fish. Artificial intelligence catch management system characterized by storing in association with numbers.   The artificial intelligence fish management system according to claim 9 or 10, wherein the three-dimensional shape measurement unit calculates volume data of the fish catch from the three-dimensional shape, and measures the weight data by the weight measurement unit. The density data of the catch is calculated from the weight data, and the quality prediction unit predicts the tightness of the meat quality in the catch as the meat quality data from the volume data and the density data, and the individual catch database is An artificial intelligence-based fish management system, wherein the volume data, the density data, and the meat quality data are stored as individual quality data in association with the identification number of the fish.   The artificial intelligence fish management system according to claim 11, wherein the color imaging unit measures color data of eyes of the fish or body color of the fish, and the quality prediction unit includes: Predicting the freshness of the catch from the data relating to the color of the eye and the color of the surface of the body, and the individual catch database is configured to use the color data of the eye, the color data of the body surface, and the freshness data as the catch. An artificial intelligence-based fish management system characterized by storing the information in association with the identification number. It is the fish management system by artificial intelligence of any one of Claims 9 thru | or 12 , Comprising: It is related with the male and female characteristic extraction part which extracts the male and female physical characteristics about a specific fish, and the said male and female of the said fish A male / male determination unit that compares the characteristics of each fish species to determine the sex of the fish, and the individual fish database associates information on the sex of the fish with the identification number for each fish A catch management system using artificial intelligence, which is characterized by being preserved.   A catch logistics system using the artificial intelligence catch identification system according to any one of claims 1 to 8, wherein the individual feature image data and the individual feature amount at a logistics stage of the catch Data is obtained, the identification number is identified by comparing the individual feature image data and the individual feature data with the individual feature image data and individual feature data stored in the individual fish database, and A logistic history is tracked for each item, and the individual catch database stores logistic history data in association with the identification number, and stores the log information by artificial intelligence.   15. A fish logistics system using artificial intelligence according to claim 14, wherein a fish quantity calculating unit for calculating the quantity of the fish for each fish type based on the identification number, A fishery composition ratio calculation unit that calculates a proportion of fishery by aquaculture or cultivated fishery in the quantity, and fish that traces the logistics route of the fish from the logistics history data stored in the individual fish database An artificial intelligence fish logistics system comprising a logistics grasping unit, wherein the individual fish catch database updates the logistics history data to the latest information.   15. The artificial intelligence fish logistics system according to claim 14, wherein the fish logistics grasping unit estimates a demand for the fish in a final consumption area from a logistics route of the fish. Intelligent fish catch logistics system. The catch logistics system using artificial intelligence according to any one of claims 14 to 16, wherein the catch logistics grasping unit includes a change in a catch amount including an aquaculture fishery or a cultivated fishery for the individual fish species. An artificial intelligence fish catch logistics system characterized by predicting the catch of aquaculture fishery or cultivated fishery.   18. The fish logistics system using artificial intelligence according to claim 14, wherein the fish logistics grasping unit is configured to simulate an optimum transportation route and transportation means to a place where the fish is consumed. An artificial intelligence fish logistics system characterized by using individual recognition by a fish identification system. The catch logistics system using artificial intelligence according to any one of claims 14 to 18, wherein the catch logistics grasping unit holds a net auction instead of auction in a fish market, and Artificial intelligence catch distribution system characterized by using a communication network for distribution.

Images