KR20230161808A - Fish sorting system using ai-based fish shadow - Google Patents

Abstract

An artificial intelligence-based fish selection system using fish shadows is launched. The present invention provides a moving tube for moving fish from a first tank to another tank, at least one lighting device that irradiates light to the fish passing through the moving tube, and an image of a shadow generated by the at least one lighting device. At least one camera device that generates data, a determination device that derives size information and health information for the fish based on the image data, and a plurality of predefined water tanks based on the size information and health information. It may include a classification device for classifying fish.

Description

Fish sorting system using artificial intelligence-based fish shadows {FISH SORTING SYSTEM USING AI-BASED FISH SHADOW}

The present invention relates to a fish selection system using artificial intelligence-based fish shadows. More specifically, it is about a fish selection system using fish shadows using vision recognition technology based on data learned through big data.

In general, a sorter refers to a device used to commercialize several products of different sizes, such as fish or fruit, by sorting them by size or type and discharging them.

In particular, when sorting fish and other seafood, various types of fish are placed on a conveyor belt and sorted as they are transported. If the worker sorts the fish by looking and judging, productivity will significantly decrease.

In addition, the existing fish sorting method is based on a causal method by workers, so not only does the work efficiency decrease significantly, but it also causes an increase in labor costs, and due to problems such as physical strength, concentration decreases when sorting work continues for a long time, and the resulting selection rate decreases. The vicious cycle leading to a decline continues.

KR 20070102023 A (Title of invention: Fish selection method and device)

The purpose of the present invention is to provide a fish selection system that derives the size of the fish and generates information on health abnormalities of the fish using only the shadow image of the fish passing through the moving tube.

Additionally, an object of the present invention is to provide a screening system with high recognition accuracy through a screening system including a lens with strong contamination resistance.

Additionally, the purpose of the present invention is to provide a screening system with high recognition accuracy by applying a filter to image data.

In order to solve the above-described problem, the present invention includes a moving tube for moving fish from a first tank to another tank, at least one lighting device for irradiating light to the fish passing through the moving tube, and the at least one lighting. At least one camera device that generates image data for a shadow generated by the device, a determination device that derives size information and health information for the fish based on the image data, and based on the size information and health abnormality information , may include a classification device for classifying the fish into a plurality of predefined water tanks.

In addition, the at least one lighting device includes a first lighting device and a second lighting device, the first lighting device radiates light to the fish at a first position, and the second lighting device is at a second position. Light is irradiated to the fish, and the at least one camera device may generate image data for a first shadow generated by the first lighting device and a second shadow generated by the second lighting device.

In addition, the determination device includes a first channel for deriving primary size information and primary health abnormality information of the fish based on the image data, and a second channel for deriving secondary size information and primary health abnormality information of the fish by applying a filter to the image data. It may include a second channel that derives information on primary health abnormalities.

Additionally, the determination device may derive tertiary size information of the fish based on the primary size information and the secondary size information.

Additionally, the determination device may generate health abnormality information about the fish based on the image data and transmit the health abnormality information to an external device.

The present invention can provide a selection system that can automatically select fish using shadow image data through a vision recognition algorithm.

Additionally, the present invention can provide a screening system with high recognition accuracy through a screening system including a lens with strong contamination resistance.

Additionally, an object of the present invention is to provide a screening system with high recognition accuracy by applying a filter to image data.

The effects that can be obtained according to the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below. will be.

Figure 1 schematically shows an artificial intelligence-based fish selection system according to the present invention.
Figure 2 shows a moving tube according to the present invention.
Figure 3 briefly shows a method of generating shadow image data according to the present invention.
Figure 4 briefly shows a method for obtaining health abnormality information according to the present invention.
Figure 5 schematically shows a discrimination device according to the present invention.
Figure 6 shows the functional configuration included in the memory according to the present invention.
Figures 7 and 8 show a fish selection device including a discrimination device including a plurality of channels.
Figure 9 shows a camera device according to the present invention.
Figure 10 is a diagram showing an example of an image correction filter according to the present invention, and Figure 11 shows an HSV graph according to the present invention.
The accompanying drawings, which are included as part of the detailed description to aid understanding of the present specification, provide embodiments of the present specification and explain technical features of the present specification together with the detailed description.

Hereinafter, embodiments disclosed in the present specification will be described in detail with reference to the attached drawings. However, identical or similar components will be assigned the same reference numbers regardless of reference numerals, and duplicate descriptions thereof will be omitted.

Additionally, in describing the embodiments disclosed in this specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in this specification, the detailed descriptions will be omitted.

In addition, the attached drawings are only for easy understanding of the embodiments disclosed in this specification, and the technical idea disclosed in this specification is not limited by the attached drawings, and all changes included in the spirit and technical scope of the present invention are not limited. , should be understood to include equivalents or substitutes.

Terms containing ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

Singular expressions include plural expressions unless the context clearly dictates otherwise.

In this application, terms such as “comprise” or “have” are intended to designate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

Hereinafter, based on the above-described contents, an artificial intelligence-based fish selection system according to a preferred embodiment of the present specification will be described in detail as follows.

Figure 1 schematically shows an artificial intelligence-based fish selection system according to the present invention.

According to Figure 1, the artificial intelligence-based fish selection system according to the present invention includes a moving tube 100, at least one camera device 400, at least one lighting device 120, a discrimination device 200, and a classification device ( 300). The artificial intelligence-based fish selection system according to the present invention may be based on a fish movement device that moves fish from a first tank to another tank. The artificial intelligence fish sorting device can move fish while simultaneously deriving size information about the fish, and classify or select fish by size based on the derived size information.

The moving tube 100 may be configured to move fish in a desired direction in the first tank. The moving tube 100 may be connected to a pump that creates negative pressure, and may be in the form of a suction tube that suctions the fish in the first tank according to the negative pressure, or may be in the form of a general tube that allows the fish to move freely. The basic configuration of the moving tube 100 connected to the pump may adopt conventionally known technologies. Additionally, the moving tube 100 may be formed to have a diameter suitable for passing only one fish, preventing multiple fish from moving simultaneously while being overlapped. Additionally, the moving tube 100 may be completely or partially transparent so that the inside can be viewed from the outside.

The camera device 400 may generate image data for fish passing through the moving tube 100. The camera device 400 may be configured to automatically generate image data when an object passes by at an angle.

At least one lighting device 120 may be a device for generating a shadow by radiating light to the fish at a specific location of the moving tube. At least one lighting device 120 may include a first lighting device and a second lighting device, and may include more lighting devices. Each lighting device is installed in a different location and irradiates light toward a single fish, thereby creating multiple shadows.

The determination device 200 may derive size information about fish based on image data. In addition, the discrimination device 200 is a type of control unit and can control the operation of components constituting the present invention, such as a pump, a camera device 400, and a classification device 300.

The classification device 300 may classify fish into a plurality of water tanks predefined by fish size based on the size information and health abnormality information derived from the determination device 200. As an example, the plurality of water tanks may include second to fifth water tanks. That is, the classification device 300 can move fish to each tank according to the size information of the fish. The classification device may include a first extension tube extending in a first direction and a plurality of second extension tubes extending from the first extension tube in a second direction. A gate may be installed between the first extension pipe and the second extension pipe. Gates may be installed in a number corresponding to each second extension pipe, and may be controlled by the discrimination device 200.

The discrimination device 200 can largely classify fish into two criteria. One may be a standard based on health abnormality information, and the other may be a standard based on size information. The determination device 200 can learn image data, detect whether a lesion has occurred on a fish in the image data based on this, and derive the size of the fish. For convenience, the size information of the fish may include the length information of the fish (from the tip of the head to the tip of the tail). In addition, as a result of analyzing image data, information on fish health abnormalities was found to be

Figure 2 shows a moving tube according to the present invention.

According to FIG. 2, the moving tube 100 according to the present invention may include an observation unit 200 that can observe the inside of the moving tube 100 from the outside. At this time, the observation unit 200 may include a through hole 111 penetrating one wall of the moving tube 100 and a lens unit 1120 that closes the through hole 111. In this case, the lens unit 1120 May include a convex lens.

According to FIG. 2, the camera device 400 may be installed outside the observation unit 200. At this time, inside/outside may refer to the portion of the passage through which the fish passes based on the moving tube 100 as the inside, and conversely, the outer part (part other than the inside) of the moving tube 100 may be referred to as the outside. That is, the camera device 400 may be installed outside the lens unit 1120. The camera device 400 may generate image data by photographing the inside of the moving tube 100 from the outside of the observation unit 200 or the lens unit 1120. This is because the camera device 400 is easily corroded or damaged by moisture, etc.

According to FIG. 2, a lighting device 120 may be additionally installed inside or outside the moving tube 100. Since fish move in a fluid, errors may occur in the image data of the fish due to light absorption or refraction due to other foreign substances in the fluid or the fluid itself. Accordingly, the camera device 400 can generate image data based on light reflected by the lighting device 120 in a wavelength range that easily penetrates fluid.

Figure 3 briefly shows a method of generating shadow image data according to the present invention.

According to FIG. 3, one lighting device 120 may irradiate light to fish passing through a moving tube and create a shadow on the other side of the lighting device 120. The camera device 400 may generate image data by photographing a shadow created on the opposite side of the lighting device 120. In order to generate image data, the moving tube 100 may include an observation unit, and the observation unit 110 may be formed at a position corresponding to and opposite to the lighting device, so that a shadow may be generated outside the moving tube 100. It can be configured so that

According to Figure 3, the two lighting devices 120 radiate light from different positions to the fish passing through the moving tube 100, but an observation part 110 with a wider width is provided on the opposite side to create a shadow. It can be created in the moving tube 100. The observation portion 110 with a wider width may be created with a width corresponding to both lighting devices 120 .

Figure 4 briefly shows a method for obtaining health abnormality information according to the present invention.

According to FIG. 4, the discrimination device 200 can recognize the movement of the fish based on the characteristic points of the fish passing through the movement passage. For example, if the fish is healthy, its head is likely to face one direction as it passes through the passage, and the fish can swim in one direction following the current. However, the health of the fish If an abnormality occurs, the direction of the fish's body may change differently from the direction of movement due to the flow of water as it passes through the passage. By detecting the direction of the body, the determination device 200 determines the fish's health condition. and may have learned big data in advance to determine health abnormalities.

Figure 5 schematically shows a discrimination device according to the present invention.

According to FIG. 5, the determination device 200 according to the present invention may include a processor 210, a memory 220, and a communication module 230.

The processor 210 may control other components by executing instructions stored in the memory 220. The processor 210 may execute instructions stored in the memory 220.

The processor 210 is a component that can perform calculations and control other devices. Mainly, it may mean a central processing unit (CPU), an application processor (AP), a graphics processing unit (GPU), etc. Additionally, the CPU, AP, or GPU may include one or more cores therein, and the CPU, AP, or GPU may operate using an operating voltage and clock signal. However, while a CPU or AP consists of a few cores optimized for serial processing, a GPU can consist of thousands of smaller, more efficient cores designed for parallel processing.

The processor 210 can provide or process appropriate information or functions to the user by processing signals, data, information, etc. input or output through the components discussed above, or by running an application program stored in the memory 220.

The memory 220 stores data supporting various functions of the determination device 200. The memory 220 may store a plurality of application programs (application programs or applications) running on the determination device 200, data for operating the determination device 200, and commands. At least some of these applications may be downloaded from an external server via wireless communication. Additionally, the application program may be stored in the memory 220, installed in the determination device 200, and driven by the processor 210 to perform the operation (or function) of the determination device 200.

The memory 220 is a flash memory type, a hard disk type, a solid state disk type, an SDD type (Silicon Disk Drive type), and a multimedia card micro type. ), card-type memory (e.g. SD or XD memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), EEPROM (electrically erasable programmable read) -only memory), PROM (programmable read-only memory), magnetic memory, magnetic disk, and optical disk may include at least one type of storage medium. Additionally, the memory 220 may include web storage that performs a storage function on the Internet.

The communication module 230 transmits and receives information to and from a base station or a camera including a communication function through an antenna. The communication module 230 may include a modulator, demodulator, signal processor, etc. Additionally, the communication module 230 may perform a wired communication function.

Wireless communication refers to communication using communication facilities already installed by communication companies and a wireless communication network that uses the frequencies of those communication facilities. At this time, the communication module 230 includes code division multiple access (CDMA), frequency division multiple access (FDMA), time division multiple access (TDMA), orthogonal frequency division multiple access (OFDMA), and single carrier frequency division multiple access (SC-FDMA). It can be used in various wireless communication systems such as access), and in addition, the communication module 230 can also be used in 3rd generation partnership project (3GPP) long term evolution (LTE), etc. In addition, not only 5G communication, which has recently been commercialized, but also 6G, which is scheduled for commercialization in the future, can be used. However, this specification can utilize a pre-installed communication network without being limited by such wireless communication method.

In addition, short range communication technologies include Bluetooth, BLE (Bluetooth Low Energy), Beacon, RFID (Radio Frequency Identification), NFC (Near Field Communication), and Infrared Data Association (IrDA). ), UWB (Ultra Wideband), ZigBee, etc. may be used.

Figure 6 shows the functional configuration included in the memory according to the present invention.

According to FIG. 6, the memory 220 according to the present invention may include a size information generation module 221 and a health abnormality information generation module 222.

The size information generation module 221 may have already performed big data learning on image data for each size according to the type of fish. The size information generation module 221 includes learning data that has previously learned the size information of the fish, and based on the pre-learned learning data, recognizes fish in the image data generated by the camera device 400, and recognizes the fish in the image data. size information can be generated.

The health abnormality information generation module 222 may have previously learned image big data including the movements of fish in which health abnormalities have occurred. The health abnormality information generation module 222 includes learning data obtained by pre-learning image data including the movement of fish in which health abnormalities have occurred, and images generated by the camera device 400 based on the pre-learned learning data. It is possible to recognize fish in the data and generate information on health abnormalities of the recognized fish.

At this time, the algorithm used by the size information generation module 221 and/or the health abnormality information generation module 222 to learn big data may be a CNN algorithm. The size information generation module and/or the health abnormality information generation module may perform learning by applying a CNN algorithm to big data consisting of pre-labeled image data or image data.

Figures 7 and 8 show a fish selection device including a discrimination device including a plurality of channels.

According to Figures 7 and 8, fish can be classified into four types. The present invention includes a plurality of water tanks, and the plurality of water tanks can each store four types of classified fish. Fish can be classified into unusual fish, small fish, medium-sized fish, and large fish. The plurality of tanks may include a first tank storing abnormal fish, a second tank storing small fish, a third tank storing medium-sized fish, and a fourth tank storing large fish.

According to FIG. 7, the determination device 200 can derive primary information about fish through the first channel. That is, the first channel can derive the primary size information and primary health abnormality information of the fish based on image data including the fish. That is, the first channel may be one unit including a first size information generation module and a second health abnormality information generation module.

According to FIG. 7 , the classification device 300 may include a first classification device corresponding to a first channel and a second classification device corresponding to a second channel. The first classification device may be controlled by a first channel, and the second classification device may be controlled by a second channel.

The first classification device can classify fish into four types. The second classification device can verify the suitability of classification for fish primarily classified by the first classification device. That is, the second classification device can perform secondary classification on the fish that has been classified as primary classification and verify the suitability of the primary classification based on whether the secondary classification result is the same as the primary classification result. The second classification device defines that classification has been performed appropriately if the secondary classification result is the same as the first classification result, and defines that classification has been performed inappropriately if the secondary classification result is different from the first classification result. You can.

The vision recognition algorithm used in the first channel may be adapted from an existing algorithm for animal recognition. Additionally, the vision recognition algorithm used in the second channel may be an existing algorithm applied with a filter to be described later.

According to FIG. 8, the determination device 200 may include first to third channels. Since the first channel and the second channel partially overlap with the content described in FIG. 5, overlapping descriptions will be omitted. The classification device 300 may further include a third classification device corresponding to the third channel. The third classification device can be controlled by the third channel. Additionally, the vision recognition algorithm used in the third channel may be an application filter that will be described later applied to the existing algorithm.

The third classification device can re-verify the suitability of classification for fish secondaryly classified by the second classification device. That is, the third classification device can perform third classification on the secondary classification fish and verify the suitability of the second classification based on whether the third classification result is the same as the second classification result. The third classification device defines that classification has been performed appropriately if the third classification result is the same as the second classification result, and that classification has been performed inappropriately if the third classification result is different from the second classification result. You can.

Figure 9 shows a camera device according to the present invention.

The camera device 400 according to the present invention can generate image data from an optical image. The camera module 1320 may include a housing including a through hole on a side wall, a lens 1321 installed in the through hole, and a driving unit 1323 that drives the lens 1321. The through hole may be formed in a size corresponding to the diameter of the lens 1321. Lens 1321 may be inserted into the through hole.

The driving unit 1323 may be configured to control the lens 1321 to move forward or backward. The lens 1321 and the driving unit 1323 are connected in a conventionally known manner, and the lens 1321 can be controlled by the driving unit 1323 in a conventionally known manner.

In order to obtain various image data, the lens 1321 needs to be exposed to the outside of the camera device 100 or the housing.

In particular, the camera device 400 according to the present invention needs to generate image data for fish in an environment with high humidity caused by moisture containing salt or impurities, so a lens with strong contamination resistance is required. Therefore, the present invention sought to solve this problem by proposing a coating layer for coating the lens.

Preferably, the surface of the lens 1321 may be coated with a coating composition containing a compound represented by the following formula (1), an organic solvent, an inorganic particle, and a dispersant.

[Formula 1]

here,

L ₁ is a single bond, a substituted or unsubstituted arylene group with 6 to 30 carbon atoms, a substituted or unsubstituted alkylene group with 1 to 20 carbon atoms, a substituted or unsubstituted cycloalkylene group with 3 to 20 carbon atoms, or a substituted or unsubstituted group. It is selected from the group consisting of an alkenylene group having 2 to 20 carbon atoms, a substituted or unsubstituted cycloalkenylene group having 3 to 20 carbon atoms, and a substituted or unsubstituted alkynylene group having 2 to 20 carbon atoms,

When L ₁ is substituted, hydrogen, nitro group, halogen group, hydroxy group, alkyl group of 1 to 30 carbon atoms, cycloalkyl group of 1 to 20 carbon atoms, alkenyl group of 2 to 30 carbon atoms, alkynyl group of 2 to 24 carbon atoms, carbon number It is substituted with one or more substituents selected from the group consisting of an aralkyl group with 7 to 30 carbon atoms, an aryl group with 6 to 30 carbon atoms, and an alkoxy group with 1 to 30 carbon atoms. When substituted with a plurality of substituents, they are the same or different from each other.

When the lens 1321 is coated with the coating composition, it can exhibit excellent water repellency and contamination resistance, so that clearer images or videos can be collected even if the lens 1321 installed around the road is exposed to a polluted environment for a long period of time.

The inorganic particles may be selected from the group consisting of silica, alumina, and mixtures thereof. The average diameter of the inorganic particles is 70 to 100 μm, but is not limited to the above example. After the inorganic particles are formed as a coating layer 1322 on the surface of the lens 1321, the physical strength can be improved and the moldability can be improved by maintaining the viscosity within a certain range.

The organic solvent is selected from the group consisting of methyl ethyl ketone (MEK), toluene, and mixtures thereof, preferably methyl ethyl ketone, but is not limited to the above examples.

As the dispersant, a polyester-based dispersant can be used. Specifically, a polyester-based dispersion stabilizer consisting of a copolymer of 2-methoxypropyl acetate and 1-methoxy-2-propyl acetate is TEGO-Disperse 670 (manufacturer) : EVONIK) can be used, but is not limited to the above example and any dispersant that is obvious to those skilled in the art can be used without limitation.

The coating composition may further include a stabilizer as other additives, and the stabilizer may include an ultraviolet absorber, an antioxidant, etc., but is not limited to the above examples and can be used without limitation.

More specifically, the coating composition for forming the coating layer 1322 may include a compound represented by Chemical Formula 1, an organic solvent, inorganic particles, and a dispersant.

The coating composition may include 40 to 60 parts by weight of the compound represented by Formula 1, 20 to 40 parts by weight of inorganic particles, and 5 to 15 parts by weight of a dispersant, based on 100 parts by weight of the organic solvent. If the range is within the above range, a synergistic effect of the water repellent effect due to the interaction of each component is expressed to a critical degree, and if it is outside the above range, the synergistic effect is rapidly reduced or almost non-existent.

More preferably, the viscosity of the coating composition is 1500 to 1800 cP. If the viscosity is less than 1500 cP, there is a problem that it flows down when applied to the surface of the lens 1321, making it difficult to form the coating layer 1322, and if the viscosity is less than 1800 cP, it flows down. In this case, there is a problem in that it is not easy to form a uniform coating layer 1322.

[Preparation Example 1: Preparation of coating layer]

1. Preparation of coating composition

A coating composition was prepared by mixing methyl ethyl ketone with a compound represented by the following formula (1), inorganic particles, and a dispersant:

[Formula 1]

here,

L ₁ is an unsubstituted alkylene group having 5 carbon atoms.

A more specific composition of the antistatic composition is shown in Table 1 below.

TX1 TX2 TX3 TX4 TX5 organic solvent 100 100 100 100 100 Compound represented by formula 1 30 40 50 60 70 inorganic particles 10 20 30 40 50 dispersant One 5 10 15 20

(unit weight)

2. Preparation of coating layer

The coating compositions DX1 to DX5 were applied to one surface of the lens 1321 and then cured to form a coating layer 1322.

[Experimental example]

1. Evaluation of surface appearance

Due to the difference in viscosity of the coating composition, after manufacturing the coating layer 1322, sensory evaluation was conducted to determine whether a uniform surface was formed. An evaluation was conducted on whether a uniform coating layer 1322 was formed, and the evaluation was conducted based on the following criteria.

○: Formation of uniform coating layer

×: Formation of uneven coating layer

TX1 TX2 TX3 TX4 TX5 Sensory evaluation Х ○ ○ ○ Х

When forming the coating layer 1322, if the viscosity is less than a certain level, flow occurs on the surface of the lens 1321, making it difficult to form a uniform coating layer 1322 after the curing process. Accordingly, the problem of lowering the production yield may occur. Additionally, even when the viscosity was too high, it was difficult to uniformly apply the composition, making it impossible to form a uniform coating layer 1322.

2. Measurement of water repellency angle

After forming the coating layer 1322 on the surface of the lens 1321, the results of measuring the water repellency angle are shown in Table 3 below.

Advancing contact angle (〃) Rest contact angle (〃) Receding contact angle (〃) TX1 117.1±2.9 112.1±4.1 < 10 TX2 132.4±1.5 131.5±2.7 141.7±3.4 TX3 138.9±3.0 138.9±2.7 139.8±3.7 TX4 136.9±2.0 135.6±2.6 140.4±3.4 TX5 116.9±0.7 115.4±3.0 < 10

As shown in Table 3, after forming the coating layer 1322 using the coating compositions TX1 to TX5, the results of measuring the contact angle were confirmed. TX1 and TX5 were measured to have receding contact angles of less than 10 degrees. In other words, it was confirmed that pinning of water droplets occurs when the coating composition falls outside the optimal range for manufacturing the coating composition. On the other hand, it was confirmed that no pinning phenomenon occurred in TX2 to 4, showing that excellent waterproofing effects can be achieved.

3. Contamination resistance evaluation

The lens 1321 on which the coating layer 1322 according to the above embodiment was formed around the tank was attached to the model camera and exposed to the environment around the tank containing seawater with a high salt concentration for 10 days. As a comparative example (Con), the same lens 1321 without the coating layer 1322 was used, and in each example, a model camera was installed at the same location around the water tank.

Afterwards, the degree of contamination of the lens 1321 before and after the experiment was evaluated, and for objective comparison, the results were compared with the comparative example in which the coating layer 1322 was not formed, and the results were evaluated with an index of 1 to 10, and are shown in Table 4 below. shown in As for the indices below, the lower the number, the better the contamination resistance.

Con TX1 TX2 TX3 TX4 TX5 Staining resistance 10 7 3 3 3 8

(Unit: index)

Referring to Table 4, when forming the coating layer 1322 on the lens 1321, even if the lens 1321 is exposed to the outside while installing the camera in the external environment, high contamination resistance is formed in a form that is easy to analyze for a long period of time. It can be seen that image data can be collected. In particular, in the case of TX2 to TX4, it can be confirmed that the contamination resistance by the coating layer 1322 is very excellent.

Figure 10 is a diagram showing an example of an image correction filter according to the present invention, and Figure 11 shows an HSV graph according to the present invention.

According to Figure 10, the types and functions of filters are shown. In other words, the CNN algorithm may be a learning algorithm that uses multiple layers. In addition, the CNN algorithm can automatically learn filters that maximize image classification accuracy, and by adding new layers called convolutional layers and polling layers before the fully connected layer, the filtered image is obtained after applying the filtering technique to the original image. A classification operation can be performed on . The CNN algorithm is configured to apply a filtering technique to the original image by adding a new layer called a convolutional layer and a pooling layer before the fully-connected layer, and then perform a classification operation on the filtered image. You can.

At this time, the calculation formula for the image filter using the CNN algorithm is as shown in Equation 1 below.

[Equation 1]

(step,

: Pixel of the ith row and jth column of the filtered image expressed as a matrix,

: filter,

: image,

: Height of filter (number of rows),

: Width (number of columns) of the filter. )

Preferably, the calculation equation for the image filter using the CNN algorithm is as shown in Equation 2 below.

[Equation 2]

(step,

: Pixel of the ith row and jth column of the filtered image expressed as a matrix,

: Application filter

: image,

: height of application filter (number of rows),

: Width (number of columns) of the application filter.)

Preferably, is an application filter that learns image data containing fish images and may be a filter applied to the image data to recognize the type of fish included in the image data. In particular, in the case of fish type recognition, recognition may be based on differences in shape and color, so an application filter may be necessary to effectively recognize shape and color. To meet these needs, application filters Can be calculated by Equation 3 below.

[Equation 3]

(step, : filter, : Coefficient, : application filter)

At this time, each The filter according to may be any one of the edge detection filter (Edge detection), sharpen filter (sharpen), and box blur filter (Box blur) matrix according to FIG. 10.

Preferably, The calculation formula for calculating is the same as Equation 4 below. At this time, can be interpreted as a variable used to increase the efficiency of the filter, and its unit can be ignored.

[Equation 4]

However, the diameter of the lens of the camera device 400 used to capture the image is in mm (however, when using multiple lenses, it is the average value), the f value of the lens is the aperture value (F number) of the lens, and the HSV average value. may mean the average value of the color coordinates according to the image converted to size values through an HSV graph. Specific details about the HSV value are as follows.

According to FIG. 11, the HSV graph according to the present invention may mean a color space that reflects perceptual characteristics. H(Hue, 0~360°) can mean hue, S(Saturation, 0~100%) can mean saturation, and V(Value, 0~100%) can mean brightness. . Hue, saturation, and brightness values can be referred to as HSV values, which can be easily extracted using graphic tools such as Adobe Illustrator CC 2019.

The HSV average value according to the present invention can be obtained through the HSV three-dimensional coordinates of FIG. 9. That is, the HSV average value according to the present invention can be calculated based on the HSV value obtained through a graphic tool. The distance value of the HSV coordinates measured with the origin coordinates on the HSV three-dimensional coordinates as the reference point can constitute the above-described HSV average value. That is, the image according to the present invention includes an image of a moving fish, and the HSV color coordinates of the image may be distributed in a certain area among the HSV three-dimensional coordinates. Therefore, the HSV average value according to the present invention may mean the distance value from the origin coordinate calculated based on the average coordinate using the average of the HSV coordinates for the color of a certain area.

[Experiment example]

The recognition accuracy of the type of moving fish when applying the application filter F' of the present invention to the image data according to the present invention is as follows. The table below is a numerical representation of the accuracy of recognition results based on whether or not filters are applied, based on requests from 10 experts in the relevant technology field.

No filter applied filter apply filter apply Average Accuracy (Score) 75 90 97

Table 5 above shows the accuracy scores evaluated by experts for each case. This experimental example investigates the accuracy of a learning module previously learned using the same big data, depending on whether or not a filter is applied.

In addition, this experimental example was experimented with image data including images of 120 different types of fish, and a survey was conducted from 10 experts on the accuracy of the fish species included in the image data regarding the recognition results.

As can be seen in Table 5, in the case where no filter was applied, there was a possibility of an error occurring in image recognition, so the accuracy was evaluated to be relatively low. In comparison, the accuracy was somewhat higher when the general CNN filter F was applied, but it was confirmed that the accuracy was significantly improved when the applied filter F' was applied.

The above-described present invention can be modeled as computer-readable code on a program-recorded medium. Computer-readable media includes all types of recording devices that store data that can be read by a computer system. Examples of computer-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SDD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. and also includes those modeled in the form of carrier waves (e.g., transmission over the Internet). Accordingly, the above detailed description should not be construed as restrictive in all respects and should be considered illustrative. The scope of this specification should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of this specification are included in the scope of this specification.

Any or other embodiments of the present invention described above are not exclusive or distinct from each other. In certain embodiments or other embodiments of the present invention described above, each configuration or function may be used in combination or combined.

The above detailed description should not be construed as restrictive in any respect and should be considered illustrative. The scope of the present invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the present invention are included in the scope of the present invention.

100: suction passage
200: Discrimination device
300: Sorting device
400: Camera device
500, 600: Water tank

Claims (5)

A moving tube for moving fish from the first tank to another tank;
At least one lighting device that radiates light to the fish passing through the moving tube;
at least one camera device that generates image data about a shadow generated by the at least one lighting device;
A determination device that derives size information and health information about the fish based on the image data; and
A classification device for classifying the fish into a plurality of predefined aquariums based on the size information and health abnormality information; including,
Fish selection system using artificial intelligence-based fish shadows.
According to paragraph 1,
The at least one lighting device includes a first lighting device and a second lighting device,
The first lighting device irradiates light to the fish at a first position, and the second lighting device irradiates light to the fish at a second position,
The at least one camera device,
Generating image data for a first shadow generated by the first lighting device and a second shadow generated by the second lighting device,
Fish selection system using artificial intelligence-based fish shadows.
According to paragraph 1,
The determination device is,
A first channel that derives primary size information and primary health abnormality information of the fish based on the image data, and
A second channel for deriving secondary size information and secondary health abnormality information of the fish by applying a filter to the image data,
Fish selection system using artificial intelligence-based fish shadows.
According to paragraph 3,
The determination device is,
Based on the primary size information and the secondary size information, tertiary size information of the fish is derived,
Fish selection system using artificial intelligence-based fish shadows.
According to paragraph 1,
The determination device is,
Based on the image data, generating health abnormality information about the fish and transmitting the health abnormality information to an external device,
Fish selection system using artificial intelligence-based fish shadows.