Fish quality determination system
Through machine learning technology, analyze the fish tail cross-section image data, crop the influence of noise, estimate the fish body area and combine fat and freshness analysis, the problem of determining the quality of tuna is solved, and the effect of being unskilled can also determine with high accuracy.
Patent Information
- Application Number
- CN202380016079.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-03-31
- Filing Date
- 2023-10-27
- Publication Date
- 2024-12-06
AI Technical Summary
It takes years of experience to determine the quality of tuna, especially in the non-frozen state, the fish tail section is rough and concave and convex, making it difficult to accurately determine the quality, especially for non-skilled people.
Machine learning technology is used to analyze the relationship between the fish tail section image data and the fish body area. By cropping the image data, the fish body area is estimated and cropped image data is generated. Combined with fat area and freshness analysis, the cropped image data is used to determine the fish's quality.
Even unskilled people can determine the quality of fish with high accuracy, including fat ratio and freshness, which improves the accuracy and ease of use of fish quality judgment.
Smart Images

Figure CN119096142A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a system for determining the quality of fish, in particular tuna. Background Art
[0002] Conventionally, a technique for evaluating the freshness of seafood has been proposed (for example, see Patent Document 1). In addition, conventionally, the quality of tuna has been determined by observing the cross section of the tuna tail in fish wholesale markets and the like. Prior Art Literature Patent Literature
[0003] Patent Document 1: Japanese Patent Application Publication No. 2010-286262 Summary of the invention Technical problem to be solved by the invention
[0004] In order to judge the quality of a fish (especially tuna) from the cross section of its tail, many years of experience are required, and it is difficult to judge the quality of tuna unless one is quite skilled. In particular, the cross section of a fish tail cut in a non-frozen (raw) state is rougher with many bumps and depressions than the cross section of a fish cut in a frozen state, so it is more difficult to judge the quality of tuna from the cross section. Therefore, it is desired to develop a system that allows anyone (even if not a skilled person) to easily judge the quality of non-frozen fish.
[0005] The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a fish quality determination system that allows even an unskilled person to easily determine the quality of fish. Technical solutions to solve problems
[0006] The fish quality determination system involved in the first method comprises: a first machine learning unit, which analyzes the relationship between image data of a cross-section of a fish tail and the area of a fish body in the image data through machine learning; a data acquisition unit, which acquires image data of a cross-section of the tail of a fish as a determination object from a user device; a first estimation unit, which estimates the area of the body of the fish as a determination object based on the relationship analyzed by the first machine learning unit, taking the image data of the cross-section of the tail of the fish as a determination object acquired by the data acquisition unit as input, and outputs an estimation result; a generation unit, which generates cropped image data obtained by cropping the area other than the fish body from the image data of the cross-section of the tail of the fish as a determination object based on the estimation result of the first estimation unit; and a quality determination unit, which uses the cropped image data to determine the quality of the fish as a determination object.
[0007] According to this structure, the fish body region is estimated using machine learning, and the quality of the fish to be determined is determined based on the cropped image data generated based on the estimation result, so that the influence of the region other than the fish body region (the region of the spine and the background region) in the image data of the cross section of the fish tail, which becomes noise when determining the quality, can be suppressed, and the quality of the fish can be determined with high accuracy. As a result, even an unskilled person can easily determine the quality of the fish.
[0008] According to the fish quality determination system involved in the first embodiment, the fish quality determination system involved in the second embodiment further includes: a second machine learning unit, which analyzes the relationship between the cropped image data and the area of fat in the cropped image data through machine learning; a second estimating unit, which takes the cropped image data as input, estimates the area of fat and outputs the estimation result based on the relationship analyzed by the second machine learning unit; and a fat determination unit, which determines the proportion of the fat area in the cropped image data based on the estimation result of the second estimating unit.
[0009] According to this configuration, the fat area of the fish is estimated by machine learning using the cropped image data as input data, and the proportion of the fat area in the cropped image data is determined based on the estimation result. Therefore, even when the image data of the cross section of the fish tail includes a concave-convex shape generated when cutting a non-frozen fish, the proportion of the fat area can be determined with high accuracy. As a result, even an unskilled person can easily determine the quality of the fish (the proportion of fat).
[0010] According to the fish quality determination system involved in the first or second embodiment, the fish quality determination system involved in the third embodiment further includes: a third machine learning unit, which analyzes the relationship between the cropped image data and freshness through machine learning; a third estimation unit, which takes the cropped image data as input, estimates the freshness and outputs the estimation result based on the relationship analyzed by the second machine learning unit; and a freshness determination unit, which determines the freshness based on the estimation result of the third estimation unit.
[0011] According to this configuration, the freshness of the fish is estimated by machine learning using the cropped image data as input data, and the freshness of the fish is determined based on the estimated result. Therefore, even when the image data of the cross section of the fish tail contains a concave-convex shape generated when a non-frozen fish is cut, the freshness of the fish can be determined with high accuracy. Thus, even an unskilled person can easily determine the quality (freshness) of the fish.
[0012] According to the fish quality determination system according to any one of the first to third aspects, the fish quality determination system according to a fourth aspect further includes a freshness determination unit that determines the freshness of the fish to be determined using color information of the cropped image data.
[0013] According to this configuration, the freshness of the fish to be determined is determined using the color information of the cropped image data, so that even when the image data of the cross section of the fish tail includes a concave-convex shape generated when a non-frozen fish is cut, the freshness of the fish can be determined with high accuracy. Thus, even an unskilled person can easily determine the quality (freshness) of the fish.
[0014] According to the fish quality judgment system involved in the third or fourth embodiment, in the fish quality judgment system involved in the fifth embodiment, the quality judgment unit judges the quality of the fish as the judgment object based on the fat ratio judged by the fat judgment unit and the freshness judged by the freshness judgment unit.
[0015] According to this structure, the quality of the fish to be judged can be judged based on the ratio of fat and freshness. Thus, even an unskilled person can easily judge the quality of the fish.
[0016] The method involved in the sixth embodiment is a method performed by a fish quality judgment system, and includes: a first machine learning step, analyzing the relationship between image data of a cross-section of a fish tail and the area of a fish body in the image data through machine learning; a data acquisition step, acquiring image data of a cross-section of the tail of a fish as a judgment object from a user device; a first estimation step, based on the relationship analyzed by the first machine learning step, using the image data of a cross-section of the tail of the fish as a judgment object acquired by the data acquisition step as input, estimating the area of the body of the fish as a judgment object and outputting an estimation result; a generation step, based on the estimation result of the first estimation step, generating cropped image data after cropping the area other than the fish body from the image data of the cross-section of the tail of the fish as a judgment object; and a quality judgment step, using the cropped image data to judge the quality of the fish as a judgment object.
[0017] According to this method, similarly to the above-mentioned system, the fish body region is estimated using machine learning, and the quality of the fish to be determined is determined based on the cropped image data generated based on the estimation result, so that the influence of the region other than the fish body region (the region of the spine, the background region) in the image data of the cross section of the fish tail, which becomes noise when determining the quality, can be suppressed, and the quality of the fish can be determined with high accuracy. As a result, even an unskilled person can easily determine the quality of the fish.
[0018] The program according to the seventh aspect is a program for causing a computer to execute the method according to the sixth aspect.
[0019] According to this program, similarly to the above-mentioned system, the fish body region is estimated using machine learning, and the quality of the fish to be determined is determined based on the cropped image data generated based on the estimation result, so that the influence of the region other than the fish body region (the region of the spine, the background region) in the image data of the cross section of the fish tail, which becomes noise when determining the quality, can be suppressed, and the quality of the fish can be determined with high accuracy. As a result, even an unskilled person can easily determine the quality of the fish.
[0020] The storage medium according to the eighth aspect is a storage medium readable by a computer storing the program according to the seventh aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 It is a block diagram of a quality determination system in an embodiment of the present invention. Figure 2 This is a flowchart showing an example of processing of the fish quality determination system in the embodiment of the present invention. Figure 3 This is an example of image data of a cross section of a non-frozen fish (tuna) tail in the embodiment of the present invention. Figure 4 This is an example of cropped image data in the embodiment of the present invention. DETAILED DESCRIPTION
[0022] Hereinafter, a fish quality determination system according to an embodiment of the present invention will be described using the drawings. In this embodiment, a tuna quality determination system is exemplified.
[0023] The configuration of a quality determination system according to an embodiment of the present invention will be described with reference to the drawings. Figure 1 is a block diagram showing the structure of the quality determination system of this embodiment. Figure 1 As shown, the quality determination system 1 is composed of a server device 2 and a user device 3. The server device 2 and the user device 3 are connected to each other via a network 4 so as to be able to communicate with each other. For example, the server device 2 is a cloud server owned by a provider of a quality determination service, and the user device 3 is a smartphone owned by a user of the quality determination service.
[0024] like Figure 1As shown, the server device 2 includes a first machine learning unit 201, a first input unit 202, a first estimation unit 203, a second machine learning unit 204, a second input unit 205, a second estimation unit 206, a third machine learning unit 207, a third input unit 208, a third estimation unit 209, a generation unit 210, a fat determination unit 211, a freshness determination unit 212, a quality determination unit 213, a data acquisition unit 214, and a storage unit 215.
[0025] The first machine learning unit 201 analyzes the image data of the cross section of the fish tail by machine learning (refer to Figure 3 ) and the relationship between the area of the fish body in the image data. In this machine learning, an arbitrary method such as deep learning based on a neural network is used.
[0026] For example, in the case of a neural network, the image data of a cross section of a tuna tail is input to an input layer, and data related to body regions in the image data is output from an output layer. Furthermore, the weight coefficients between neurons of the neural network are optimized by supervised learning using analysis data (teacher data) that associates the data input to the input layer with the data output from the output layer. As teacher data of the data related to body regions in the image data of a cross section of a tuna tail, for example, data to which an expert has added labels by region to the image data of the cross section of a tuna tail is used.
[0027] The image data of the cross section of the tail of the tuna as the determination target is input to the first input unit 202. The image data of the cross section of the tail of the tuna as the determination target is acquired by, for example, photographing the cross section of the tail of the tuna in various fishery factories.
[0028] The first estimation unit 203 takes the image data of the cross section of the tail of the tuna as the determination object input by the first input unit 202 as input, estimates and outputs the data related to the body region in the image data of the cross section of the tail of the tuna as the determination object. For example, in the case of the above-mentioned neural network, the image data of the cross section of the tail of the tuna as the determination object is input to the input layer, the data related to the body region in the image data of the tail of the tuna as the determination object is estimated, and the output (estimation result) is output from the output layer, thereby estimating the data related to the body region in the image data of the cross section of the tail of the tuna as the determination object.
[0029] The second machine learning unit 204 analyzes the cropped image data generated by the generation unit 210 described later by machine learning (see Figure 4 ) and the relationship between the area of fat in the cropped image data. In this machine learning, an arbitrary method such as deep learning based on a neural network is used.
[0030] For example, in the case of a neural network, the cropped image data is input to an input layer, and data related to fat areas in the image data is output from an output layer. Furthermore, the weighting coefficients between neurons of the neural network are optimized by supervised learning using analysis data (teacher data) that associates the data input to the input layer with the data output from the output layer. As teacher data for data related to fat areas in the cropped image data, for example, data to which an expert has added labels for each area of the cropped image data is used.
[0031] The cut-out image data of the tuna as the determination target is input to the second input unit 205. The cut-out image data of the tuna as the determination target is generated by the generation unit 210 described later.
[0032] The second estimation unit 206 takes the cropped image data of the tuna as the determination target inputted from the second input unit 205 as input, estimates and outputs the data related to the fat region in the cropped image data, based on the relationship analyzed by the second machine learning unit 204. For example, in the case of the above-mentioned neural network, the cropped image data of the tuna as the determination target is inputted into the input layer, the data related to the fat region in the cropped image data of the tuna as the determination target is estimated, and the output (estimation result) is outputted from the output layer, thereby estimating the data related to the fat region in the cropped image data of the tuna as the determination target.
[0033] The third machine learning unit 207 analyzes the cropped image data generated by the generation unit 210 described later by machine learning (see Figure 4 ) and the freshness of the fish. In this machine learning, an arbitrary method such as deep learning based on a neural network is used.
[0034] For example, in the case of a neural network, the cropped image data is input to an input layer, and data related to freshness corresponding to the cropped image data is output from an output layer. Furthermore, the weighting coefficients between neurons of the neural network are optimized by supervised learning using analysis data (teacher data) that associates the data input to the input layer with the data output from the output layer. As teacher data for data related to freshness corresponding to the cropped image data, for example, data in which a skilled person has attached a label of freshness (e.g., a scale of 1 to 5 (1 is the lowest freshness, 5 is the highest freshness)) to the cropped image data is used.
[0035] The cut-out image data of the tuna as the determination target is input to the third input unit 208. The cut-out image data of the tuna as the determination target is generated by the generation unit 210 described later.
[0036] The third estimation unit 209 takes the cropped image data of the tuna as the determination object inputted from the third input unit 208 as input, estimates and outputs the data related to the freshness corresponding to the cropped image data, based on the relationship analyzed by the third machine learning unit 207. For example, in the case of the above-mentioned neural network, the cropped image data of the tuna as the determination object is inputted into the input layer, the data related to the freshness corresponding to the cropped image data of the tuna as the determination object is estimated, and the output (estimation result) is outputted from the output layer, thereby estimating the data related to the freshness corresponding to the cropped image data of the tuna as the determination object.
[0037] The generation unit 210 generates cropped image data by cropping the area other than the fish body (the area of the spine and the background area) from the image data of the cross-section of the fish tail of the fish as the determination object, based on the estimation result of the first estimation unit 203 (data related to the area of the fish body in the image data of the cross-section of the fish tail). Figure 4 2 is a diagram showing an example of cropped image data generated by the generating unit 210. Figure 4 As shown, the generator 210 generates cropped image data by cropping the vertebral region and the background region located in the center of the cross section from the image data of the fishtail cross section. The cropped image data generated by the generator 210 is input to the second input unit 205 and the third input unit 208.
[0038] The fat determination unit 211 determines the ratio (content ratio) of the fat region in the cropped image data based on the estimation result (data related to the fat region in the cropped image data) of the second estimation unit 206. For example, the ratio of the area of the fat region to the total area of the cropped image data (image data of the fish body region) may be used as the fat content ratio.
[0039] The freshness determination unit 212 determines the freshness of the fish to be determined based on the estimation result (the freshness of the fish corresponding to the cropped image data) of the third estimation unit 209. Here, the freshness determination unit 212 may directly use the freshness estimated by the third estimation unit 209, or may determine the freshness by normalizing the value output by the third estimation unit 209 to a given numerical range.
[0040] It should be noted that the freshness determination unit 212 can determine the freshness based on the HSV color information of the cropped image data without using the estimation result of the third estimation unit 209. For example, the color of the cross section of the fish body becomes brighter red as the freshness increases, so the freshness determination unit 212 extracts red (Hue: 0 to 60 degrees and 300 to 360 degrees) from the cropped image data, and after calculating its chroma (0 to 100%), the calculated chroma is normalized to a given numerical range (e.g., 1 to 5) and the value is used as the freshness for determination.
[0041] The quality determination unit 213 determines the quality of the fish to be determined based on the fat ratio (content) determined by the fat determination unit 211 and the freshness determined by the freshness determination unit 212. For example, the quality is determined as follows: since the fresher and fatter tuna is, the higher its value is, the higher the freshness and the higher the fat ratio is, the higher the quality is.
[0042] The data acquisition unit 214 acquires the image data of the cross section of the tail of the tuna, which is the determination target, from the user device 3 .
[0043] The storage unit 215 stores image data acquired from the user device 3 and cropped image data generated by the generation unit 210. The storage unit 215 also stores estimation results output from the first estimation unit 203, the second estimation unit 206, and the third estimation unit 209.
[0044] Next, the user device 3 is described. Figure 1 As shown, the user device 3 includes a photographing unit 301 , a data input unit 302 , a display unit 303 , and a storage unit 304 .
[0045] The imaging unit 301 is implemented by the camera function of the user device 3, and generates image data of the cross section of the tuna tail by imaging the cross section of the tuna tail (see Figure 3 ). It should be noted that the generated image data of the cross section of the tuna tail may be associated with the data of the shooting date.
[0046] The data input unit 302 has a function of inputting various data. Information related to the tuna to be determined (for example, identification information of the fish to be determined, weight, fishing ground, fishing boat name, etc.) is input from the data input unit 302.
[0047] The display unit 303 has a function of displaying various data. At least one of the ratio (content) of fat determined by the fat determination unit 211, the freshness determined by the freshness determination unit 212, and the quality of the fish determined by the quality determination unit 213 is displayed on the display unit 303. It should be noted that when the display unit 303 has a touch panel function, the display unit 303 can also serve as the data input unit 302.
[0048] The storage unit 304 stores image data of tuna photographed by the photographing unit 301 and information input via the data input unit 302. The storage unit 304 also stores data output from the server device 2 on the ratio (content) of fat in the fish, freshness, and quality.
[0049] Regarding the quality determination system 1 configured as above, refer to Figure 2 The timing diagram of FIG. 1 illustrates its operation.
[0050] like Figure 2 As shown, in the quality determination system 1 of this embodiment, first, in the first machine learning unit 201 of the server device 2, the relationship between the image data of the cross section of the tuna tail and the body area in the image data is analyzed in advance by machine learning (S10).
[0051] In addition, the second machine learning unit 204 of the server device 2 analyzes the relationship between the cropped image data and the fat area in the cropped image data by machine learning in advance ( S11 ).
[0052] In addition, the third machine learning unit 207 of the server device 2 analyzes the relationship between the cropped image data and the freshness corresponding to the cropped image data in advance by machine learning ( S12 ).
[0053] When the image of the cross section of the tuna tail is captured in the user device 3 ( S13 ), the user device 3 transmits the image data of the cross section of the tuna tail to the server device 2 ( S14 ).
[0054] When receiving the image data of the cross section of the tuna tail from the user device 3 , the server device 2 estimates the area of the tuna body by the first estimating unit 203 ( S15 ).
[0055] Then, the server device 2 generates the cropped image data by the generating unit 210 using the estimation result (data related to the area of the tuna body) estimated by the first estimating unit 203 ( S16 ).
[0056] Next, the server device 2 uses the cropped image data generated in step S16 as input data, and estimates the fat area by the second estimating unit 206 ( S17 ).
[0057] Then, the server device 2 uses the estimation result estimated by the second estimation unit 206 (data on the fat region in the cropped image data) to determine the ratio (content ratio) of the fat region of the tuna by the fat determination unit 211 ( S18 ).
[0058] Next, the server device 2 uses the cutout image data generated in step S16 as input data, and estimates the freshness of the tuna to be determined by the third estimating unit 209 ( S19 ).
[0059] Then, the server device 2 uses the estimation result estimated by the third estimation unit 209 or the color information of the cropped image data to have the freshness determination unit 212 determine the freshness of the tuna to be determined ( S20 ).
[0060] Furthermore, the server device 2 determines the quality of the tuna to be determined by the quality determination unit 213 based on the ratio (content) of fat determined by the fat determination unit 211 and the freshness determined by the freshness determination unit 212 ( S21 ).
[0061] Then, the server device 2 transmits the determination result (at least one of the ratio of fat, freshness, and quality) to the user device 3 .
[0062] The quality determination system 1 according to the present embodiment includes: a first machine learning unit 201 that analyzes the relationship between the image data of the cross section of the fish tail and the area of the fish body in the image data by machine learning; a data acquisition unit 214 that acquires the image data of the cross section of the fish tail of the fish to be determined from the user device; a first estimation unit 203 that estimates the area of the fish body of the fish to be determined based on the relationship analyzed by the first machine learning unit, taking the image data of the cross section of the fish tail of the fish to be determined acquired by the data acquisition unit 214 as input, and outputs the estimation result; a generation unit 210 that generates cropped image data obtained by cropping the area other than the fish body from the image data of the cross section of the fish tail of the fish to be determined based on the estimation result of the first estimation unit 203; and a quality determination unit 213 that determines the quality of the fish to be determined using the cropped image data, thereby suppressing the influence of the area other than the fish body (the area of the spine and the background area) in the image data of the cross section of the fish tail, which becomes noise when determining the quality, and determining the quality of the fish with high accuracy. This makes it possible for even non-experts to easily determine the quality of fish.
[0063] In addition, according to the quality determination system 1 of the present embodiment, since machine learning is used in the determination of the ratio (content) of fat meat and the determination of freshness, even when the image data of the cross section of the fish tail includes the concavo-convex shape generated when the non-frozen fish is cut, the ratio (content) of fat meat and the freshness can be determined with high accuracy, and as a result, the quality of the fish can be determined with high accuracy. Therefore, even an unskilled person can more easily determine the quality of the fish.
[0064] The present embodiment is described above, but any part or all of the functional units described in this specification may be implemented by a program. The program mentioned in this specification may be released by recording it non-temporarily on a computer-readable recording medium, or by publishing it via a communication line such as the Internet (including wireless communication), or by installing it on any terminal.
[0065] Based on the above description, those skilled in the art may be able to think of additional effects and various modifications of the present invention, but the mode of the present invention is not limited to the above-mentioned embodiments. Various additions, changes and partial deletions can be made within the scope of the conceptual ideas and the main purpose of the present invention derived from the content specified in the scope of the patent claim and its equivalents.
[0066] For example, a device described as one device (or component, the same below) in this specification (including a device depicted as one device in the drawings) may also be implemented by multiple devices. Conversely, a device described as multiple devices in this specification (including a device depicted as multiple devices in the drawings) may also be implemented by one device. Alternatively, a unit or function included in a certain device (such as a server) may be included in part or all of another device (such as a user terminal).
[0067] In addition, not all matters described in this specification are essential conditions. In particular, matters described in this specification but not described in the scope of the patent claims can be regarded as arbitrary additional matters.
[0068] It should be noted that the applicant is only aware of the publicly known inventions in the literature recorded in the "Prior Art Literature" column of this specification, and the present invention is not necessarily intended to solve the problems in the publicly known inventions in the literature. The problem to be solved by the present invention should be determined by considering the entire specification. For example, in this specification, when there is a record that a given effect is achieved through a specific structure, it is also possible to solve problems opposite to the given effect. However, it does not mean that such a specific structure must be a necessary requirement. Industrial Applicability
[0069] As described above, the fish quality determination system according to the present invention has the effect that even an unskilled person can easily determine the quality of fish, and is useful as a tuna quality determination system and the like.
Claims
1. A fish quality determination system, comprising: A first machine learning unit that analyzes, by machine learning, a relationship between image data of a cross section of a fish tail and a region of a fish body in the image data; a data acquisition unit that acquires image data of a cross section of a fish tail of a fish to be determined from a user device; a first estimating unit that, based on the relationship analyzed by the first machine learning unit, takes the image data of the cross section of the tail of the fish to be determined obtained by the data acquisition unit as input, estimates the area of the body of the fish to be determined, and outputs an estimation result; a generating unit that generates cropped image data obtained by cropping a region other than a fish body from the image data of a cross section of a tail of the fish to be determined, based on the estimation result of the first estimating unit; as well as A quality determination unit determines the quality of the fish to be determined using the cropped image data.
2. The fish quality determination system according to claim 1, wherein: The fish quality determination system also has: a second machine learning unit, which analyzes the relationship between the cropped image data and the area of fat in the cropped image data by machine learning; a second estimating unit that estimates a fat area based on the relationship analyzed by the second machine learning unit and outputs an estimation result by taking the cropped image data as an input; as well as A fat meat determination unit determines a ratio of a fat meat area in the cropped image data based on the estimation result of the second estimation unit.
3. The fish quality determination system according to claim 2, wherein: The fish quality determination system also has: a third machine learning unit, which analyzes the relationship between the cropped image data and freshness through machine learning; a third estimating unit that takes the cropped image data as input, estimates freshness based on the relationship analyzed by the second machine learning unit, and outputs an estimation result; as well as The freshness determination unit determines the freshness based on the estimation result of the third estimation unit.
4. The fish quality determination system according to claim 2, wherein: The fish quality determination system further includes a freshness determination unit configured to determine the freshness of the fish to be determined using color information of the cropped image data.
5. The fish quality determination system according to claim 3 or 4, wherein: The quality determination unit determines the quality of the fish to be determined based on the fat ratio determined by the fat determination unit and the freshness determined by the freshness determination unit.
6. A method, which is a method performed by a fish quality determination system, and comprises: A first machine learning step is to analyze the relationship between the image data of the cross section of the fish tail and the area of the fish body in the image data by machine learning; A data acquisition step of acquiring image data of a cross section of a fish tail of a fish to be determined from a user device; A first estimation step, based on the relationship analyzed by the first machine learning step, using the image data of the cross section of the tail of the fish as the determination object acquired by the data acquisition step as input, estimating the area of the body of the fish as the determination object and outputting the estimation result; A generating step, based on the estimation result of the first estimating step, generating cropped image data obtained by cropping the area other than the body of the fish from the image data of the cross section of the tail of the fish as the determination object; as well as The quality determination step is to determine the quality of the fish to be determined using the cropped image data.
7. A program for causing a computer to execute the method according to claim 6. 8 . A storage medium readable by a computer storing the program according to claim 7 .
Citation Information
Patent Citations
Device and method for evaluation of freshness
JP2010286262A
