Device for detecting and classifying the incidence of yake (burnt flesh syndrome) in frozen fish pieces, and method for detecting and classifying the incidence of yake in frozen fish pieces using said device

Abstract

The invention relates to a device, and a method using the device, for the detection and classification of the incidence of Yake (lactic acid and its derivatives) in frozen fish intended for human consumption, in a non-invasive manner, without compromising the frozen state of the fish. For this purpose, the device comprises a contact sensor with retractable contacts connected to a switching and control board provided with an FPGA, as well as an impedance spectrophotometer, within the spectrum from 10 Hz to 1 MHz. When in contact with the frozen fish piece, the contact sensor generates electric fields which are used to determine the incidence of heterogeneous Yake at different depths, thanks to the algorithms programmed into the FPGA that can be used to obtain the α and β relaxation dispersions in the generated electric fields.

Description

[0001]

[0002] Device for detecting and classifying the incidence of Yake in frozen fish pieces and procedure for detecting and classifying the incidence of Yake in frozen fish pieces using said device.

[0003] OBJECT OF THE INVENTION

[0004] The present invention relates to a device and method for detecting and classifying the incidence of Yake—or burnt flesh syndrome caused by the presence of lactic acid—in frozen fish carcasses (pieces). Advantageously, the device and method of the present invention allow for the detection of Yake without the need to thaw the fish or alter its physical integrity to extract a sample and measure lactic acid, thus minimizing potential economic losses.

[0005] BACKGROUND OF THE INVENTION

[0006] Yake, or 'burned flesh syndrome' in fish, especially severe in tuna, is a growing concern in the fishing industry. This phenomenon refers to a set of physical and chemical alterations that occur in fish flesh during capture, particularly in large species like bluefin tuna, caused by the stress experienced by the animal. The stress associated with capture triggers the production of lactic acid in the muscle tissue, leading to various changes that result in severe losses in product quality. Yake, or 'burned flesh syndrome', is characterized by changes in the color, texture, and flavor of the flesh, negatively impacting its quality and commercial value. This phenomenon is particularly relevant to the fresh tuna industry, where product quality is crucial.However, this problem has spread to a wide variety of fish consumed in Europe, in addition to tuna, such as tilapia, grouper, swordfish, cod, sole or carp, among others.

[0007] Yake occurs due to the effects of stress on the animal, which causes oxygen depletion, leading to anaerobic respiration at the muscular level. This produces lactic acid and, consequently, a decrease in muscle pH. This decrease triggers biochemical reactions that alter the structure and composition of the meat. These changes can include protein denaturation, lipid oxidation, and the release of volatile compounds that affect the aroma and flavor of the fish. As a stress-induced syndrome, Yake is influenced by factors such as fishing and slaughter techniques.

[0008] Most fish caught in Europe arrives at fishing and processing facilities as frozen products. This means that, with current techniques, yake detection in fish fillets must be carried out only after the fish has been thawed. The major problem for the sector is that, once yake is detected in the thawed product, it is no longer possible to return it to the originating company, resulting in significant financial losses for the receiving companies.

[0009] Thus, two techniques involving the thawing of carcasses are currently used in practice. The first is a visual assessment by experts, where, once the samples are thawed, an expert describes the incidence of lactic acid (Yake) through visual evaluation, establishing the following classifications: A (no appearance), B+ (very slight appearance), B (slight appearance), B- (moderate appearance), and C (severe appearance, rendering the carcass or frozen fish piece unusable). The other technique, less common in the fishing industry but more widely used in other meat sectors affected by lactic acid, consists of measuring pH using a pH meter, yielding a classification similar to the previous one.

[0010] On the other hand, document ES1295393U discloses equipment for determining burnt flesh syndrome in fish. The equipment described in this document is designed to detect burns during the freezing process of fish and consists of a biosensor, or electrochemical sensor, that uses enzymes coupled to electrical conductivity meters. This biosensor is composed of metallic vapors whose conductivity changes depending on the reaction of the enzymes with lactic acid. This system requires taking a sample of the thawed product and mixing it with reagents before placing it in a container with the enzyme and the sensor. Therefore, the equipment disclosed in document ES1295393U involves a destructive measurement; that is, it is necessary to thaw and cut up the fish carcass. This means that the piece selected for measurement must be discarded, and it also prevents the measurement from being performed on all carcasses in the batch, if desired.Additionally, it has the drawback of requiring a medium and reagents to perform the measurement, which is not immediate.

[0011] The article 1975-03-24 JASON, AC et al., Journal of Physics E: Scientific Instruments, 1975,8(10): 826-830. DOI: 10.1088 / 0022-3735 / 8 / 10 / 011 'The development of an electronic fish freshness meter' discloses a fish freshness measuring device that measures electrical conductivity; however, this document does not disclose its application to frozen fish samples, nor a measurement principle that allows for the precise quantification of the presence of lactic acid.

[0012] Regarding other measurement techniques, the article "Measurements of the dielectric properties of frozen and defrosted meat and fish in the frequency range 10-200 MHz" (1963-02-15 BENGTSSON, NE et al., Journal of the Science of Food and Agriculture, 1963, 11, 33, 29557-29568. Doi: 10.1002 / jsfa.2740140812) is known. This article analyzes the dielectric properties of various animals and fish that have been previously frozen and thawed in the radiofrequency range. However, this article does not disclose a specific technical configuration for measuring yake in frozen fish carcasses, as the measurements were performed on unfrozen samples or on frozen and subsequently thawed samples, never on frozen samples.

[0013] Finally, patent ES2616150B1 from one of the applicants for the present invention discloses an apparatus and method for detecting damage caused by deep pectoral muscle myopathy in poultry. The apparatus of the invention is designed for detecting infarcts and angina in birds slaughtered after rigor mortis. It employs radiofrequency spectrophotometry, but requires tomographic capabilities to reach the internal pectoral muscle, near the heart, and adaptation to a curved and soft surface. The apparatus in patent ES2616150B1 uses large, rigid, flat sensors, enabling a system with high penetration capacity. It simultaneously measures tension and current, avoiding surface drift due to the absence of the surface tension present in frozen products.In other words, the device described in document ES2616150B1 is designed for use on soft surfaces, not hard surfaces like that of frozen fish. Therefore, this technology would not be applicable to the detection of Yake in frozen products, which have a different surface tension than unfrozen meat.

[0014] For all the above reasons, the applicants of the present invention recognize that there is a significant need to provide a device and procedure for the detection of yake in frozen fish carcasses that allows measurements to be taken without destroying and / or thawing the fish, providing a quick and accurate measurement of all pieces if desired, not just a sample, thus avoiding substantial economic losses resulting from the detection of yake at the destination once the product has been thawed, since this prevents its return to the origin or use.

[0015] DESCRIPTION OF THE INVENTION

[0016] The present invention discloses a device and method for detecting yake in frozen fish pieces by measuring the mean complex permittivity through the generation of pulsed electric fields on the fish piece, preferably in different areas. Permittivity is related to factors associated with the presence of yake, such as the appearance of lactate ions (related to scattering (a)) and the type of ice crystals formed in the fish tissue during freezing (type Ih ice, related to scattering ( )). All this information allows the fish to be classified according to the incidence of yake.

[0017] To this end, the device of the invention is based on a multi-contact sensor with at least 10 retractable contacts connected to an electronic system based on an FPGA (Field Programmable Gate Array). This FPGA has an embedded processor where the information is processed, the measurement channel is selected via a group of switches, and the measurement is executed. The processor is connected to an analog signal generator in the 10 Hz to 1 MHz range, with an analog-to-digital converter that sends the measurement information back to the processor where the signal is processed, obtaining the complex permittivity spectrum and applying a discriminating algorithm. The algorithm obtained from the complex permittivity of the tissue, preferably in different areas of the frozen fish, will allow discrimination between the different categories of Yake incidence.

[0018] Specifically, the device for detecting yake (lactic acid and its derivatives) in frozen fish carcasses comprises:

[0019] - a power supply;

[0020] - a multi-contact sensor provided with at least 10 retractable contacts made of a conductive material, such as a metal, one of the retractable contacts being for current measurement and the other two for voltage measurement. Preferably, the multi-contact sensor is provided with 10 retractable contacts, such that a group of 3 retractable contacts constitutes a measuring channel. Thus, between two measuring channels (5 retractable contacts in total), one of the retractable contacts is common to both channels and, together with one of the other contacts in the channel, is used to measure the voltage difference, while the remaining contact in that channel is used to measure the current intensity;

[0021] - an integrated switching and control board connected by, preferably, 4 coaxial connections to a processor embedded in an FPGA, the switching and control board being associated with the contact sensor;

[0022] - a spectrophotometer consisting of a digital-to-analog signal generator for obtaining the electromagnetic spectrum within the range of 10 Hz to 1 MHz with a digital-to-analog converter (emission converter), which allows its control; and

[0023] - an analog-to-digital converter (receiver converter) for capturing the signal with the information coming from the contact sensor;

[0024] where the FPGA is connected to the spectrophotometer signal generator, the emission digital-to-analog converter, the analog-to-digital converter, and the switching and control board.

[0025] It is worth noting that the processor embedded in the FPGA is programmed with an algorithm for obtaining the complex permittivity spectrum (where the real part corresponds to the dielectric constant and the imaginary part to the loss factor), an algorithm for obtaining the dispersions (a) and ( ), and a discrimination algorithm based on these properties. Thus, the retractable conductive contacts apply an electrical signal controlled by the switching board to the outer surface of the frozen fish piece, generating a voltage difference between the various contact points of the retractable conductive contacts with the frozen fish piece, which induces electric fields in the fish piece.Preferably, the device of the invention has a configuration that allows these electric fields to be generated in different areas of the fish piece, in order to avoid problems of heterogeneity in the surface measurements. The frozen fish piece distorts these electric fields (changes the path of the photons that make up the electric field) according to the complex permittivity, which in turn varies according to different physicochemical properties of the frozen piece. This variation is conditioned and processed by the FPGA processor for calculating the detection and quantification of the Yake level in the fish piece. Preferably, the retractable metal contacts are coated or integrated with gold, which advantageously allows the device to operate at the required low temperatures, on the order of -18°C. e C, which is the temperature for transporting and handling frozen fish.

[0026] Advantageously, since in frozen tissues the complex permeability (magnetic effect), with a perpendicular pole geometry as arranged in the invention, is ten thousand times smaller than the complex permittivity (electrical effect), the magnetic effect of the measurement can be disregarded. Thus, using the device and procedure of the invention, the electrical properties (complex permittivity) are isolated during the electromagnetic measurement.

[0027] The device of the invention obtains the complex permittivity spectrum in the radio frequency range, specifically from 10 Hz to 1 MHz, using the capabilities of the digital-to-analog signal generator and the algorithm processor to measure the permittivity spectrum. The resulting property, permittivity, is vectorial, and this equipment allows it to be obtained in complex or polar coordinates thanks to the novel algorithms programmed in the FPGA. When analyzed in complex form, three scatterings or interactions of photons with the medium appear in this portion of the spectrum:

[0028] - the first is alpha scattering (α), located in the range between 1Hz and 1 kHz, and which includes interactions with ionic chemical species of low molecular weight and high ionic strength, in which lactate is found;

[0029] - the second is beta scattering (£), located in the range between 1 kHz and 1 MHz, which includes interactions associated with high surface tension or the Maxwell-Wagner effect, very relevant in the formation of ice crystals Ih (at atmospheric pressure), and interactions associated with high molecular weight chemical species, such as proteins; and

[0030] - The third of these is a phenomenon called ionic conductivity. Permittivities (a) and ( ) affect the entire permittivity; however, ionic conductivity affects charged chemical species across the entire frequency range, but only affects the imaginary part of the complex permittivity.

[0031] Thus, the claimed device allows obtaining the complex permittivity in the range between 10 Hz and 1 MHz, a range that shows the two dispersions or relaxations (a) and (P), since the different levels of algorithm are programmed in the processor embedded in the FPGA, both for the control of the measurement system, for obtaining the permittivity of each of the relaxations (a) and ( ), as well as for the classification of the samples.

[0032] Additionally, the device, through its FPGA, can optionally be connected to a communication medium intended to be coupled to an interface - such as a screen, computer, tablet, smartphone, smartwatch, etc., where the information is displayed - or to a data processing system that allows this information to be incorporated into the monitoring of a conditioning plant.

[0033] The use of retractable contacts, made of metals that do not become rigid at low temperatures, is essential to maintain contact on a rigid and irregular surface such as that of frozen fish, which makes contact with conductive metal sensors difficult. Retractable contacts preferably have a rounded end, which advantageously ensures uniformity in measurements, even if the contact sensor is placed in different areas of the fish carcass.

[0034] Thus, in contrast to other spectrophotometry systems that require contact with a large surface area to ensure penetration and depth measurement for internal or tomographic measurements, the device of the invention does not require a bulky contact sensor. The retractable contact system employed ensures minimal contact on rough and rigid surfaces, and the need for contact with the fish sample is equally minimal, as the required measurements are superficial. This is because lactic acid and its derivatives, the chemical species responsible for yake, are distributed homogeneously throughout the depth of the fish carcass muscle, making the surface of the carcass a representative region of its entire volume.

[0035] On the other hand, the present invention also relates to the procedure for the detection of yake and classification of frozen fish carcasses using the described device, which has the following stages:

[0036] - Placement of the contact sensor on the surface of the frozen piece of fish, positioned so that all the conductive retractable contacts make continuous total contact with it, generating a voltage and current difference between the conductive retractable contacts;

[0037] - induction of low-intensity electric fields through the frozen fish pieces by the voltage difference between the conductive retractable contacts; - modulation of the electric fields generated as a result of their passage through the frozen fish piece;

[0038] - reception in the retractable conductive contacts of the modulated electric fields after passing through the frozen piece of fish;

[0039] - transmission of the signal corresponding to the modulated electric fields to the FPGA via coaxial cables from the switching board;

[0040] - conditioning and processing of the response signal by the FPGA, first obtaining the vector permittivity and subsequently the values ​​of the relaxation permittivities or scatterings (a) and ( ) by means of the algorithms programmed in the FPGA;

[0041] - Calculation, using a discrimination algorithm programmed on the FPGA, of the Yake incidence level based on the relaxation permittivities (a) and (|3) obtained in the previous stage. Preferably, the algorithm programmed on the FPGA is a neural network algorithm (Artificial Intelligence models).

[0042] Preferably, the contact sensor is placed on the frozen fish piece in the area of ​​maximum representativeness of the average lactic content, located at the top of the longitudinal area of ​​the fish piece, ensuring total contact of the retractable contacts.

[0043] Preferably, the generated electric fields will not exceed 250 mV / cm through the frozen pieces, in the range of 10 Hz to 1 MHz.

[0044] For all the above reasons, the device and procedure of the invention offer a significant competitive advantage to the fish industry, as it allows for the detection, with very high sensitivity and specificity, of the incidence of Yake online in distribution centers, far from fishing and freezing centers. This is because the frozen samples will not require thawing for detection and can therefore be returned to the centers of origin, where their purpose will be redirected, without generating losses for the distribution sector and without becoming waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] To complement the description that follows and to aid in a better understanding of the characteristics of the invention, according to a preferred embodiment thereof, a set of figures is included as an integral part of said description, in which, for illustrative and non-limiting purposes, the following has been represented:

[0046] Figure 1 shows a plan and profile view of the multi-contact sensor with retractable contacts, of the sensor's contact area, as well as the distribution of the different measuring channels according to a preferred embodiment of the object of the present invention, in which the contact sensor has 10 retractable contacts distributed in four measuring channels in groups of 3, with a retractable contactor shared between channels.

[0047] Figure 2 shows a diagram of the retractable contacts of the contact sensor, elevation, plan and profile, according to the preferred embodiment of the previous figure, as well as an electrical diagram of the measuring system, with the switching and the emission and reception of the signal.

[0048] Figure 3 shows a schematic of the device from the previous figures during the Yake measurement procedure and its impact on a piece of frozen fish, where the electric fields generated on it have been schematically represented.

[0049] Figure 4.- Shows a representation of the relaxation frequency of alpha (Hz) against two groups of qualities in which the Yake is classified, quality A (absence of Yake) (data series 0) and the rest of the qualities (data series 1), where the calculation has been carried out by means of a logistic modeling algorithm of the relaxations programmed in the FPGA.

[0050] Figure 5 shows a representation of the alpha relaxation frequency divided by the squared beta dielectric constant against all the quality types in which the Yake is classified (A (data series 0), B+ (data series 0.5), B (data series 1), B- (data series (1,5) and C (data series 2)), where the calculation has been performed using a logistic relaxation modeling algorithm programmed on the FPGA.

[0051] PREFERRED EMBODIMENT OF THE INVENTION

[0052] Figure 3 schematically details a preferred embodiment of the device for detecting and classifying the incidence of yake in frozen fish pieces (1) according to the object of the present invention during its use. The device comprises the following elements:

[0053] - an electrical power supply (10);

[0054] - a contact sensor (2) provided with a plurality of retractable contacts (3) of conductive material. In the embodiment illustrated in Figures 1, 2 and 3, the contact sensor (2) has 10 retractable contacts (3), capable of measuring in 4 channels;

[0055] - a switching and control board (4) integrated into the sensor (2) connected to the processor embedded in an FPGA (13), via four coaxial connections (9), wherein the processor embedded in the FPGA (13) is programmed with an algorithm for obtaining the complex permittivity spectrum, another algorithm for obtaining the properties of the scatterings (a) and (£), and an algorithm for discrimination based on said properties;

[0056] - a spectrophotometer consisting of a digital-to-analog signal generator and analog-to-digital converter (5) for obtaining the electromagnetic spectrum within the range of 10 Hz to 1 MHz, the spectrophotometer (5) being linked to the FPGA (13) and the switching and control board (4); and

[0057] - a screen (7) for manipulation and display of results.

[0058] Thus, Figure 1 shows the details of the contact sensor (2) according to an embodiment of the invention, which has 10 retractable contacts (3) distributed across 4 channels for four spectrophotometric measurements. The retractable contacts have a cylindrical body with a diameter of 1 mm and a height between 3 and 5 mm. As detailed above, each channel consists of 3 contacts (3), with one contact (3) shared between channels. The figure shows how the channels are formed by groups of three contacts, with each channel sharing one contact (3). Starting from the left end of Figure 1, the first channel has the first two contacts (3) separated by a distance (12) of 5 mm, while the second and third contacts (3) are separated by a distance (11) of 7 mm. The third contact (3) is shared with channel 2, and the measurements between contacts (3) are mirrored.For channels 3 and 4, the same separations between contacts are repeated (3), which allows minimizing interference between the current and voltage measurements, since measuring the current and voltage separately avoids the conductive effect of surface ice.

[0059] On the other hand, Figure 2 shows the elevation, plan, and profile details of the contact sensor (2), which has 10 retractable contacts (3). This allows for spectrophotometric measurement through four measurement channels, advantageously eliminating the effect of heterogeneity in the distribution of lactic acid on the surface of the frozen fish carcass (1). The electrical schematic of the measurement assembly is also shown in this figure, including the contacts (3), the switching system integrated into the switching and control board (4), and the digital-to-analog signal generator with the analog-to-digital conversion (5) for obtaining the electromagnetic spectrum.

[0060] In the implementation discussed, the Yake incidence detection and classification procedure has the following stages:

[0061] - placement of the contact sensor (2) on the surface of the frozen fish carcass (1), positioned so that all the retractable conductive contacts (3) make full contact with it, generating a voltage difference, and therefore an electric field (6) between the contacts (3) in each channel sequentially, in the figures described, in a frequency range between 10 Hz and 1 MHz; through the fish carcass (1);

[0062] - transmission of the response signal to the FPGA (13) by means of the switching board (4);

[0063] - conditioning and processing of the response signal by the FPGA, first obtaining the vector permittivity from the impedance of the medium calculated as a function of the measured voltage and current, and subsequently obtaining the values ​​of the relaxation permittivities or dispersions (a) and ( ) by means of the algorithms programmed in the FPGA;

[0064] - Calculation using an algorithm programmed on the FPGA of the Yake incidence level (lactic acid content and its derivatives) from the relaxation permittivities (a) and (p).

[0065] Thus, in the device and procedure of the present invention, the processing of the measured signals is carried out in two stages integrated into the developed software, and therefore into the FPGA (13).

[0066] According to one embodiment of the invention, the Yake incidence level can be calculated using an algorithm based on the logistic modeling of relaxations (physical models or logistic relaxation models), segregating based on the dispersions (a) and ( ). The result of the calculation with this type of algorithm is shown in Figure 4, where quality A (data series 0) is separated from the other qualities (1 ) by the relaxation frequency (a). Figure 5 shows the result of performing algorithms with different relaxation parameters, such as the dielectric constant in ( ) and the relaxation frequency in (a), which allows quality C (data series 2) to be separated from the other qualities.

[0067] In order to improve the calculation resolution and make the device and measurement procedure of the invention suitable for use in the fish processing line, the discrimination algorithm programmed in the FGPA is preferably a neural network model algorithm (a model that employs artificial intelligence). This allows for online discrimination of the presence and severity of Yake, with very high sensitivity and specificity. Thus, machine learning solutions based on neural networks can provide improved results once all the required information is entered and the networks are properly trained with data and measurements obtained by the invention and with known assessments of the presence or absence of Yake.

[0068] With the measurements taken and using artificial intelligence, it has been shown that it is possible, at a minimum, to classify into two classes, which we could label as absence of Yake (A) and generalized presence of Yake (C), with high accuracy. Thus, after training the model with only 200 measurements, an accuracy (cases in which the predictive model is correct versus the total number of cases) of 89.47% (± 6.85) was achieved in the classification, according to the confusion matrices represented in the tables below: Table 1. Confusion matrix for detection of the presence of Yake with a sensitivity of 90.9% and a specificity of 87.5%.

[0069]

[0070] Table 2. Confusion matrix for detecting the presence of Yake with a sensitivity of 83.33% and a specificity of 100%.

[0071]

[0072] In this context, sensitivity refers to the predictor's accuracy in detecting the yake-infected fish within the population of fish without yake, specifically tuna in the experiments conducted. Thus, high sensitivity means that a high percentage of tuna with yake are detected, which advantageously prevents them from reaching the end consumer of high-quality tuna (such as Japanese restaurants). Similarly, the specificity achieved is also high, avoiding false positives that would lead to many pieces of acceptable-quality fish being used for purposes requiring lower quality, resulting in economic losses for the producer.

[0073] To obtain these results, a topology was used that combines a radiofrequency spectrophotometry analyzer, a recurrent neural network, and a two-layer hidden-layer multilayer perceptron. Permittivity outputs from frequency sweeps between 10 Hz and 1 MHz were used as input to a recurrent network for the data series obtained at different frequencies, followed by a neural network specialized in classification (for example, a two-layer hidden-layer multilayer perceptron with an input that would be the pooled evolution of all the hidden neurons of the recurrent network and an output that would be an indicator of the presence or absence of Yake).

Claims

1. Device for the detection and classification of the incidence of yake in frozen fish pieces, characterized in that it comprises: - an electrical power supply (10); - a contact sensor (2) provided with at least 10 retractable contacts (3) made of a conductive material; - a switching and control board (4) connected via four coaxial connections (9) to a processor embedded in an FPGA (13), the switching and control board (4) being associated with the contact sensor (2); - a spectrophotometer consisting of a digital-to-analog signal generator (5) for obtaining the electromagnetic spectrum within the range of 10 Hz to 1 MHz with a digital-to-analog converter, the spectrophotometer being linked to the FPGA (13) and the switching and control board (4); and - an analog-to-digital converter for capturing the signal with the information coming from the contact sensor (2); where the processor embedded in the FPGA is programmed with an algorithm for obtaining the complex permittivity spectrum, an algorithm for obtaining the dispersions (a) and (P), and an algorithm for discrimination based on said properties;so that the retractable conductive contacts (3) apply a signal controlled by the switching board (4) on the external surface of the frozen piece of fish (1), generating a voltage difference between the different contact points of the retractable conductive contacts (3) with the frozen piece of fish (1), inducing electric fields (6) in the frozen piece of fish (1), which are modulated as they pass through the frozen piece of fish (1), obtaining the complex permittivity spectrum of the frozen piece of fish (1) and the dispersions (a) and (P), a signal that is conditioned and processed by the FPGA processor for the calculation of the detection and quantification of the level of incidence of Yake in the frozen piece of fish (1) from the dispersions (a) and (P).; 2. Device for detecting and classifying the incidence of Yake in frozen fish pieces, according to claim 1, characterized in that the retractable conductive contacts (3) have a rounded end.

3. Device for detecting and classifying the incidence of Yake in frozen fish pieces, according to claim 1, characterized in that the retractable conductive contacts (3) have a gold coating.

4. Device for the detection and classification of Yake incidence in frozen fish pieces, according to claim 1, characterized in that the retractable contacts (3) have a cylindrical body with a diameter of 1 mm and a height of between 3 and 5 mm.

5. Device for the detection and classification of Yake incidence in frozen fish pieces, according to claim 1, characterized in that the retractable contacts (3) are arranged with a separation between them of at least 5 mm.

6. Device for the detection and classification of Yake incidence in frozen fish pieces, according to claim 1, characterized in that the contact sensor (2) has 10 retractable contacts (3).

7. Device for the detection and classification of Yake incidence in frozen fish pieces, according to claim 1, characterized in that it is linked through the FPGA to an interface, computer, tablet, smartphone or smartwatch.

8. Procedure for detecting and classifying the incidence of Yake in frozen fish pieces using the device of any of the preceding claims, characterized in that it comprises the following steps: - placement of the contact sensor (2) on the surface of the frozen fish carcass (1), positioned so that all retractable contacts (3) make full contact with it, generating a voltage and current difference between the contacts (3); - induction of low-intensity electric fields (6) through the frozen fish carcasses (1) between the retractable contacts (3); - modulation of the electric fields (6) as they pass through the frozen piece of fish (1); - reception in the retractable conductive contacts (3) of the modulated electric fields (6) after passing through the frozen fish carcass (1); - transmission of the signal corresponding to the modulated electric fields (6) to the FPGA through the 4 coaxial cables from the switching and control board (4); - conditioning and processing of the response signal by the FPGA, first obtaining the vector permittivity and subsequently the values ​​of the dispersions (a) and ( ) by means of the algorithms programmed in the FPGA; - calculation using the discrimination algorithm programmed in the FPGA of the Yake incidence level from the dispersions (a) and ( ); where the calculation of the Yake incidence level using the discrimination algorithm programmed in the FPGA is carried out using a recurrent neural network algorithm employing artificial intelligence 9. Procedure for detecting and classifying the incidence of Yake in frozen fish pieces according to claim 8, characterized in that the generated electric fields have an intensity less than or equal to 250 mV / cm in the range of 10 Hz to 1 MHz.

Images