WO2023039663A1

WO2023039663A1 - Pond yield determination system and method of determining a yield of a pond containing marine organisms

Info

Publication number: WO2023039663A1
Application number: PCT/CA2022/051366
Authority: WO
Inventors: Yaroslav BABICH; Marnix FAES; Julien ROY; Vincent MARCEAU; Samuel COUTURE-BROCHU
Original assignee: Xpertsea Solutions Inc.
Priority date: 2021-09-15
Filing date: 2022-09-14
Publication date: 2023-03-23

Abstract

There is disclosed a pond yield determination system generally having: a computing device performing: accessing weight distribution data indicative of a weight distribution of marine organisms to be harvested from a pond by a producer; selecting a processing type indicative of a type of processing according to which the marine organisms of the pond are to be processed by a processor, including retrieving a processing yield function indicative of a marine organism fraction remaining after processing as a function of marine organism weight for the type of processing; accessing a transaction history bias factoring in prior transactions including initial weight distributions and associated actual pond yields for the selected type of processing; and determining a pond yield upon applying the processing yield function and the transaction history bias to the weight distribution data.

Description

POND YIELD DETERMINATION SYSTEM AND METHOD OF DETERMINING A YIELD OF A POND CONTAINING MARINE ORGANISMS

FIELD

[0001] The improvements generally relate to the field of marine organism production, and more specifically relate to assisting in transactions made between marine organism producers and processors.

BACKGROUND

[0002] Marine organisms, such as shrimp, can be grown in one or more bodies of water, typically named ponds, by a producer. Once it is deemed that a population of marine organisms growing in a pond has reached or is about to reach maturity, the producer can contemplate the selling of the marine organisms to a selected processor for the purpose of gross distribution. The marine organisms can be processed according to different processing types, and may include the removal of the head, the removal of the shell, the removal of the veins and the like. As schematically illustrated in Fig. 1 , once a producer-processor pair has agreed on a transaction for the marine organisms of a given pond, the marine organisms are harvested and transported towards the processor using a transport vehicle, such as a truck having one or more tanks, inside which the marine organisms are stored during transportation.

[0003] In an exemplary transaction occurring in the Ecuadorian market, the producer does not get paid for what the population of marine organisms is deemed to be worth at the time of harvest, but instead for what the remaining population of marine organisms is actually worth after the harvest, transportation and processing. The latter is typically referred to as a pond price which generally amounts to a pond yield multiplied or otherwise correlated to a price list. It is thus in the best interest of the producer to ensure that the marine organisms are carefully handled during the harvesting and well preserved during transportation. Accordingly, the marine organisms are preferably transported in an environment favouring well-being. In some circumstances, especially when the producer is far away from the processor, preservative solution may be added into the tanks receiving the marine organisms to ensure freshness and prevent discolouration during the transportation step. It is known that in practice, the storing of the marine organisms in such preservative solution may altertheir body water content, such that their weight is not exactly the same as it was measured at the pond when the marine organisms reach the processor. Losses can also occur during processing, should employees err in the initial weighting of the marine organisms, the selection of the processing type or trim of the marine organisms can be incorrect and may result in the marine organisms being trimmed too generously for instance.

[0004] It is only after the harvest, transportation and processing that a pond yield and price for the marine organisms can be determined, and that a payment to the producer is initiated by the processor. In practice, the processor may pay the producer weeks if not months after the time of harvest thereby negatively affecting the cash flow of the producer in the meantime. Although the existing pond yield and price determination techniques for such a transaction is satisfactory to a certain degree, there remains room for improvement.

SUMMARY

[0005] It was found that there is a need in the industry for accurately evaluating a pond yield and price prior to the harvest, transportation and processing of the marine organisms. For instance, in some embodiments, a third party may be interested in buying a pond’s marine organisms from a producer, swiftly proceeding with the payment based on an estimated pond yield, and then selling the marine organisms to a processor. For such a transaction to be advantageous for all parties, the third party should have sufficient funds to proceed with the swift payments to the producers and not bother too much with the processor’s delayed payments, in addition to generating a profit from the subsequent transaction with the processor. As such, as long as the producers are paid promptly and fairly, and that the processors are not overcharged in the process, the presence of such a broker has been found to be convenient. For any and all transactions, an inherent risk lies in wrongly evaluating a pond yield. For instance, should the broker pay too much for a pond’s marine organisms, the broker would not only advance funds forweeks, but also make a loss on the trade which would not be economically viable in the long term. Accordingly, there is an increased stress on accurately evaluating the pond yield to fairly price a pond’s marine organisms, especially as such a broker would want to scale this practice, and buy and sell a significant number of ponds from a same producer, or ponds from several producers, and generate profits on these trades in a sustainable manner over time. [0006] In high volume transactions, to satisfy economic and logistics considerations, such transactions can be performed in the framework of a computing device, which can include protocols designed to facilitate the various communications between the producer(s), the transporter(s), and the processor(s) as well as storing and updating mathematical models indicative of how each type of processing may affect the yield for instance as a function of the organisms’ weight. Prior transactions including initial weight distributions (e.g., in units per kg) and associated actual pond yields for all the types of processing may also be factored in the pond yield calculations. The prior transactions can be informative of a producer bias which pertains to, among other things, how well a producer evaluates the initial weight distribution of its ponds, how well a producer samples the population of its ponds, and how well a producer harvests and stores the marine organisms for transportation, to name a few examples. Prior transactions can also be informative of a processor bias which pertains to, but is not limited to, how much of each marine organism is discarded for each type of processing, how strict is the processor in the rejection of some less than perfect marine organisms and the like.

[0007] Applying the mathematical models and prior transactions to the initial weight distribution measured by the producers in a way that is statistically sound can require computational power which would exceed that of the mathematical mind, especially considering that the skill set of the broker typically lies in pond evaluation and sales rather than in arithmetic and statistics. Moreover, when a given pond has reached maturity, it was found that the pond should preferably be harvested within a short time frame, e.g., within one a few hours, to limit the losses and increase profitability, thus a quick agreement between the producer and the broker is often preferable. Due to physiological characteristic of the organisms, e.g., moulting, the harvesting window is often short. Accordingly, there would be no time for the broker to calculate a pond yield factoring in the mathematical model and all past prior transactions by hand and paper. Further, once the broker is present at a producer site, he/she will typically evaluate a significant number of ponds, thereby limiting the remaining amount of time for each set of calculations. In view of these practical limitations, it was found that the computational power of a computing device is of importance, especially if the pond yield calculation is to be made quickly, e.g., within about 10 minutes, preferably within about 1 minute and most preferably within a few seconds. [0008] In accordance with a first aspect of the present disclosure, there is provided a pond yield determination system comprising: a computing device having a microprocessor and a memory having stored thereon instructions that when executed by the microprocessor perform the steps of: accessing weight distribution data indicative of a weight distribution of marine organisms to be harvested from a pond by a producer, the weight distribution including a plurality of measured weight values associated to a sample of the marine organisms of the pond; selecting a processing type indicative of a type of processing according to which the marine organisms of the pond are to be processed by a processor, including retrieving a processing yield function indicative of a marine organism fraction remaining after processing as a function of marine organism weight for the type of processing; accessing a transaction history bias factoring in prior transactions including initial weight distributions and associated actual pond yields for the selected type of processing; and determining a pond yield upon applying the processing yield function and the transaction history bias to the weight distribution data, said determining being performed within a time window of about 10- minutes. In some embodiments, the pond yield can be indicative of a plurality of post-processing weight values. In some other embodiments, the pond yield includes an expected number of post-processed organisms in each one of a number of quality grades. The quality grades including one or more consumable quality grades and one or more rejected quality grades.

[0009] Further in accordance with the first aspect of the present disclosure, the pond yield determination system can for example further comprise generating a price quote for the marine organisms of the pond based on the pond yield and on a price list including price values for a plurality of weight ranges. In some embodiments, the pond yield determination system has a display displaying the price quote, and/or a communication module transmitting the price quote to an external device or network, for instance.

[0010] Still further in accordance with the first aspect of the present disclosure, said generating can for example further include generating a price probability curve indicative of a price distribution for the marine organisms of the pond at least based on the pond yield, on the price list and on the transaction history bias, and selecting the price quote from the price probability curve. [0011] Still further in accordance with the first aspect of the present disclosure, said selecting can for example further include selecting the price quote based on a pre-determined percentile value.

[0012] Still further in accordance with the first aspect of the present disclosure, the pond yield determination system can for example further comprise instructing harvesting of the marine organisms of the pond upon acceptance on the price quote.

[0013] Still further in accordance with the first aspect of the present disclosure, the pond yield determination system can for example further comprise receiving an actual pond yield for the pond after said processing and updating the transaction history bias based on the received actual pond yield.

[0014] Still further in accordance with the first aspect of the present disclosure, the transaction history bias can for example further include at least one of a producer transaction history bias indicative of prior weight distributions and associated actual pond yields cleared by the producer and a processor transaction history bias indicative of prior weight distributions and associated actual pond yields cleared by the processor.

[0015] Still further in accordance with the first aspect of the present disclosure, the sample can for example further include the weight values of at least 10 marine organisms, preferably at least 100 marine organisms, and most preferably at least 200 marine organisms.

[0016] Still further in accordance with the first aspect of the present disclosure, the transaction history bias can for example furtherfactor in at least 5 prior transactions, preferably at least 10 prior transactions and most preferably at least 40 prior transactions.

[0017] Still further in accordance with the first aspect of the present disclosure, said determining can for example further be performed within a time window below 30 minutes, preferably below 15 minutes and most preferably below 5 minutes.

[0018] Still further in accordance with the first aspect of the present disclosure, the pond yield determination system can for example further comprise selecting a marine organism type indicative of a type of the marine organisms, the processing yield function and the transaction history bias pertaining to the selected marine organism type.

[0019] Still further in accordance with the first aspect of the present disclosure, wherein the processing type can for example further be selected from a group consisting of: head-on processing, headless processing, shell-on processing, shell-off processing, and deveined processing.

[0020] Still further in accordance with the first aspect of the present disclosure, said determining a pond yield can for example further include performing Monte Carlo simulations for the pond yield at least based on the weight distribution data and on the transaction history bias.

[0021] Still further in accordance with the first aspect of the present disclosure, said determining the pond yield can for example further include outputting a plurality of postprocessed weight values.

[0022] Still further in accordance with the first aspect of the present disclosure, said determining the pond yield can for example further include outputting an expected number of post-processed organisms in each of a plurality of quality grades.

[0023] In accordance with a second aspect of the present disclosure, there is provided a method of determining a yield for a pond containing marine organisms, the method comprising: accessing weight distribution data indicative of a weight distribution of marine organisms to be harvested from the pond by a producer, the weight distribution including a plurality of measured weight values associated to a sample of the marine organisms of the pond; selecting a processing type indicative of a type of processing according to which the marine organisms of the pond are to be processed by a processor, including retrieving a processing yield function indicative of a marine organism fraction remaining after processing as a function of marine organism weight for the type of processing; accessing a transaction history bias factoring in prior transactions including initial weight distributions and associated actual pond yields for the selected type of processing; and determining a pond yield upon applying the processing yield function and the transaction history bias to the weight distribution data, said determining being performed within a time window of about 10 minutes.

[0024] Further in accordance with the second aspect of the present disclosure, the method can for example further comprise generating a price quote for the marine organisms of the pond based on the pond yield and on a price list including price values for a plurality of weight ranges.

[0025] Still further in accordance with the second aspect of the present disclosure, said generating can for example further include generating a price probability curve indicative of a price distribution for the marine organisms of the pond at least based on the pond yield, on the price list and on the transaction history bias, and selecting the price quote from the price probability curve.

[0026] Still further in accordance with the second aspect of the present disclosure, said selecting can for example further include selecting the price quote based on a pre-determined percentile value.

[0027] Still further in accordance with the second aspect of the present disclosure, the method can for example further comprise instructing harvesting of the marine organisms of the pond upon acceptance on the price quote.

[0028] Still further in accordance with the second aspect of the present disclosure, the method can for example further comprise receiving an actual pond yield for the pond and updating the transaction history bias based on the received actual pond yield.

[0029] Still further in accordance with the second aspect of the present disclosure, said determining the pond yield can for example further include outputting a plurality of postprocessed weight values.

[0030] Still further in accordance with the second aspect of the present disclosure, said determining the pond yield can for example further include outputting an expected number of post-processed organisms in each of a plurality of quality grades. [0031] It was found that the methods and systems described herein can be advantageous for the producers, as they allow a more accurate calculation of the final price for their pond’s population, thus improving their negotiation position and allowing them to select the best offer, on the basis of the price lists offered by the processors and also of their associated processing variations.

[0032] The methods and systems described herein can also be advantageous for the processor, as they allow the optimization of their purchases and fulfill orders more efficiently from local and international distributors, which are formulated on the basis of processed quality grades obtained after processing.

[0033] The methods and systems described herein can also be advantageous for the brokers, as they allow a more accurate calculation of the transactional value or distribution of possible values of each pond’s population they aim to finance, thus allowing them to give a bigger advance to the producer while quantifying and optimizing their risk position for overpayments and money losses.

[0034] Many further features and combinations thereof concerning the present improvements will appearto those skilled in the art following a reading of the instant disclosure.

DESCRIPTION OF THE FIGURES

[0035] In the figures,

[0036] Fig. 1 is a schematic view illustrating the path of a pond’s marine organisms from a producer to a processor;

[0037] Fig. 2 is a block diagram of an example of a pond yield determination system, in accordance with one or more embodiments;

[0038] Fig. 3 is a graph showing an example of an initial weight distribution indicative of a weight distribution of a pond’s marine organisms, in accordance with one or more embodiments; [0039] Fig. 4 is a graph showing an example of a processing yield function indicative of a marine organism fraction remaining after processing as a function of weight for a given type of processing, in accordance with one or more embodiments;

[0040] Fig. 5A is a graph showing examples transaction history biases, showing a producer transaction history bias and a processor transaction history bias, in accordance with one or more embodiments;

[0041] Fig. 5B is a graph showing examples of transaction history biases, showing processor transaction history biases for three consumable quality grades and two rejected quality grades, in accordance with one or more embodiments;

[0042] Fig. 6 is a graph showing a probability distribution of price quotes for a pond’s organisms, in accordance with one or more embodiments;

[0043] Fig. 7 is an example of a computing device implementing the pond yield determination system of Fig. 2, in accordance with one or more embodiments; and

[0044] Fig. 8 is a block diagram of another example of a pond yield determination system, involving a computer image analyzer module determining an initial weight distribution of a sample of marine organisms, in accordance with one or more embodiments.

DETAILED DESCRIPTION

[0045] Fig. 2 shows an example of a pond yield determination system 30. As depicted, the pond yield determination system 30 can be implemented by a combination of hardware and software components. For instance, the pond yield determination system 30 can be provided in the form of a computer, a smart phone, an electronic tablet, a cloud processing platform, or a combination thereof. The pond yield determination system 30 can be wiredly and/or wirelessly communicatively coupled to an external network such as the Internet. The pond yield determination system 30 can be communicatively coupled directly, or indirectly via the external network, to a producer module 32 associated to one or more producers 22 and/or to a processor module 34 associated to one or more processors 14. [0046] As shown, the pond yield determination system 30 has a pond yield determination module 36 which is stored on a memory of the pond yield determination system 30 and which has instructions executable to perform one or more computing steps. More specifically, the pond yield determination module 36 is communicatively coupled to one or more databases stored on an accessible memory system. The pond yield determination module 36 is configured to access weight distribution data 38 indicative of a weight distribution of marine organisms 10 to be harvested from a pond 12 by a producer 22. The weight distribution data 38 can include a plurality of measured weight values associated to one or more samples of the marine organisms 10 of the pond 12. In some embodiments, the weight distribution data 38 can be provided in the form of an array of weight values [w-i, W2, ... , w , with i being an integer corresponding to the number of sampled marine organisms 10. The number of sampled marine organisms 10 can vary from one sample to another, but may include at least about ten (10) marine organisms 10, preferably at least about one hundred (100) marine organisms 10, and most preferably at least about two hundred (200) marine organisms 10. An example of collected weight distribution data 38 is shown in Fig. 3.

[0047] As shown, the weight distribution data 38 can include weight values e.g., in grams) associated to more than one sample in some embodiments. In some other embodiments, the weight distribution data 38 is provided in the form of a matrix showing weight ranges and how many marine organisms 10 are associated to each weight range, e.g., [wi<w<W2, ; W2<w<W3, n2; ... ; w_n<w<w_n+i, n_n], with n being an integer corresponding to the number of weight ranges. The weight distribution data 38 can be received from a weight curve or otherwise formatted depending on the embodiment. Typically, the weight distribution data 38 is received from the producer module 32, for instance. However, in some other embodiments, the weight distribution data 38 can be inputted by the producer 22 or by the broker via a user interface communicatively coupled to the pond yield determination module 36. The weight values can also stem from analyzing an image showing a sample of marine organisms 10 using computer vision algorithms. Examples of such algorithms are described in PCT Published Patent Application No. WO 2019/210421 , the contents of which are hereby incorporated by reference.

[0048] Returning to Fig. 2, a processing type 40 is then selected for the pond’s marine organisms 10. The processing type 40 is indicative of a type of processing according to which the marine organisms 10 of the pond 12 are to be processed by the processor 14. The processing types 40 can differ for each type of marine organisms 10. For instance, for shrimp, the processing types 40 may include head-on processing, headless processing, shell-on processing, shell-off processing, deveined processing and the like, orany combination thereof. In some embodiments, the processing type 40 can be selected by the producer 22 or by the broker via a user interface communicatively coupled to the pond yield determination module 36. In some other embodiments, the processing type 40 may be automatically selected by the pond yield determination module 36 given a history between the marine organism producer 22 and the processor 14 for that transaction.

[0049] Once the processing type 40 has been selected, the pond yield determination module 36 retrieves a processing yield function 42 indicative of a marine organism fraction remaining after processing |i as a function of marine organism weight w for the selected type of processing. The processing yield function 42 can be a continuous function, e.g., |i = f(w), indicating the marine organism fraction |i remaining after processing as a function of the marine organism weight w for the selected processing type. An example of such a processing yield function 42 is shown in Fig. 4.

[0050] As shown, the processing yield function f(w) can be linear in this embodiment, but it is understood that in alternate embodiments, the function can also be non-linear without departing from the present disclosure. The type of function can depend on the processing type 40 selected and on the type of marine organisms 10 to be processed. The processing yield function f(w) may not necessarily be a curve but rather any suitable type of mathematical transfer function which when applied to the weight distribution data 38 outputs what is expected to be remaining of the marine organisms 10 after the selected processing. For instance, the processing yield function f(w) can also include an association between weight ranges and their corresponding marine organism fraction remaining after processing, e.g., [w-i<w<W2, pi; W2<w<W3, P2; ... ; w_n<w<w_n+i, p_n]. The processing yield function f(w) can be retrieved from a memory system accessible to the pond yield determination module 36.

[0051] The pond yield determination module 36 accesses a transaction history bias factoring in prior transactions including initial weight distributions and associated actual pond yields forthe selected processing type 40. The initial weight distributions are indicative of what the producer 22 or broker initially estimated the weight distribution of a pond 12 to be prior to harvest, transportation and processing. The actual pond yield refers to what a population of marine organisms 10 was actually worth after the harvest, transportation and processing according to a given processing type 40. The actual pond yield is a data point which is generally known only after the processing has occurred, and is therefore an unknown at the time a pond is to be harvested. The transaction history bias includes a non-negligible number of transactions. For instance, the transaction history bias may factor in at least 5 prior transactions, preferably at least 10 prior transactions and most preferably at least 40 prior transactions.

[0052] Referring back to Fig. 2, the transaction history bias can include a producer transaction history bias 44 indicative of prior weight distributions and associated actual pond yields cleared by the producer 22. For instance, the producer transaction history bias 44 can be indicative of how well a producer 22 evaluates the initial weight distribution of its ponds 12, how well a producer 22 samples the population of its ponds 12, and how well a producer 22 harvest and store the marine organisms 10 for transportation, to name a few examples. The producer transaction history bias 44 may factor in at least 5 prior transactions, preferably at least 10 prior transactions and most preferably at least 40 prior transactions for each given producer 22. The transaction history bias can also include a processor transaction history bias 46 indicative of prior weight distributions and associated actual pond yields cleared by the processor 14. The processor transaction history bias 46 can be indicative of, but is not limited to, how much of weight w of each type of marine organism 10 is discarded for each processing type 40, how strict is the processor 14 in the rejection of some less than perfect marine organisms 10, how well the marine organisms 10 are treated during the overall processing type 40, how ideal the environmental conditions are in the processing facility, to name a few examples. The processor transaction history bias 46 may factor in at least 5 prior transactions, preferably at least 10 prior transactions and most preferably at least 40 prior transactions for each given processor 14 and each processing type 40. Examples of the producer transaction history bias 44 and on the processor transaction history bias 46 are shown in Fig. 5A for understanding purposes. The transaction history bias can be stored on a memory of the pond yield determination system 30 and accessible by the pond yield determination module 36. By applying the processing yield function 42 and the transaction history bias, such as the producer transaction history bias 44 and/or the processor transaction history bias 46 for instance, to the weight distribution data 38, the pond yield determination module 36 can determine an expected pond yield 48 being indicative of a plurality of expected post-processed weight values. The post-processed weight values give an indication of what portion of the marine organisms 10, once processed by the processor 14, will remain and then be paid for. As the pond yield determination module 36 is computer-based, the determination can be performed within a time window of approximately 10 minutes. In some embodiments, the determination can be performed within a time window below 5 minutes, preferably below 1 minute and most preferably within a few seconds.

[0053] In some embodiments, the expected post-processed weight values can be classified into a number of quality grades each having categories or sub-categories. The categories can be defined using predefined weight ranges, for instance. The classification can be performed by a classification module in communication with the pond yield determination module 36. In these latter embodiments, the pond yield may be expressed in terms of an expected number of post-processed organisms in each one of the quality grades. It was found convenient to classify the post-processed weight values into the quality grades as price lists generally used in the industry include unitary prices per quality grades. As such, if a number N1 of postprocessed organisms are expected to be classified in a first quality grade, having a predefined weight range and a unitary price P1 , the price for these post-processed organisms can be swiftly estimated by multiplying the number N1 of post-processed organisms by the unitary price P1 of that first quality grade. A rejection process can be performed by a rejection module in communication with the pond yield determination module 36. In some embodiments, a given proportion of organisms in predefined weight ranges are expected to be rejected. For example, if the first quality grade discussed above has a rejection rate of R1 , then the expected number of remaining post-processed organisms in that quality grade can be given by N1 (1 -R1). The price for that quality grade would thereby be given by: P1x N1 (1 -R1). The predefined weight ranges for the quality grades, the unitary prices, the rejection rates are typically produced dependent and/or processor-dependent, and can be stored on an accessible memory system for easy access when desired. [0054] In some embodiments, the post-processed weight values need not to be calculated directly for such a classification process to be properly made. As depicted in Fig. 5B, a processing yield function 42 for the selected type of processing and a transaction history bias including information as to how the processor, for instance, has historically classified processed organisms into some quality grades can be applied to the weight distribution data 38. In the depicted embodiment, the transaction history bias includes processor transaction history biases 47 showing how likely the processor is to classify processed organisms of respective given weight ranges into one of the three consumable quality grades M1 , M2 and M3. Additionally, the transaction history bias includes processor transaction history bias biases 49 showing how likely the process is to classify processed organisms of respective given weight ranges into one of the two rejected quality grades R1 and R2. In these embodiments, the pond yield generated by the pond yield determination module is directly expressed in terms of an expected number of post-processed organisms in each one of the quality grades M1 , M2, M3, R1 and R2. The price for the pond yield can be swiftly calculated using a price list including unitary prices per quality grades. For instance, the price list can include a first unitary price for the consumable quality grade M1 , a second unitary price for the consumable quality grade M2, and so forth. In some other embodiments, the quality grades include quality grades A, B and C, with quality grade A being of greater quality (e.g., more expensive) than quality grade B which is in turn of greater quality than quality grade C. In some embodiments, the quality grades A and B both have the commercial weight grades U-5, U-7, U-10, U-12, U-15, 16-20, 21-25, 26-30, 31-35, 36-40, 41-50, 51-60, 61-70, 71-90, 91 -110, 1 11-130, 131 -150, and >=151 , with the numbers representing weight categories expressed in expected postprocessed organisms per weight unit (e.g., shrimp per kg) . However, in these embodiments, the quality grade C has qualitative commercial categories such as local, juvenile, basura, otros, rejected as well. It was found particularly useful to classify the post-processed organisms in this manner as historical data show that some processors tend to overgrade some of the processed organisms while some other processors tend to undergrade some of the processed organisms. By having transaction history biases factoring in these tendencies can thereby help better estimate the price of the organisms of a pond, which in turn reduce the risks of overestimating its price. [0055] In some embodiments, the pond yield determination module 36 also accounts for the type of the marine organisms 10 that are to be processed by the processor 14. In these embodiments, a selection of a marine organism type indicative of a type of the marine organism 10 is made. It is intended that in such embodiments, the processing yield function 42 and the transaction history biases, such as the producer transaction history bias 44 and/or the processor transaction history bias 46 for instance, may be selected to pertain to transactions on that type of marine organism 10 only, thereby ignoring or otherwise not using the bias data associated to the processing of other types of marine organisms, unrelated to the given transaction.

[0056] In some embodiments, the pond yield determination module 36 generates a price quote for the marine organisms 10 of the pond 12 based on the pond yield and on an accessible price list including price values for a plurality of weight ranges. Typically, larger marine organisms 10 are sold at a higher price per unit weight than smaller organisms. The price list can be producer- or processor-dependent depending on the embodiment. The price quote can account for the marine organisms of the whole pond’s population, or a portion thereof. The price quote can be expressed in terms of dollars or in terms of the number of dollars per unitary kilogram of post-processed weight. As shown in Fig. 6, the pond yield determination module 36 generate a price probability curve 50 indicative of a price distribution for the marine organisms 10 of the pond’s population at least based on the pond yield, on the price list and on the transaction history bias. In such embodiments, the broker is free to select a price quote based on the likeliness of each price quote of the price distribution. For instance, a price quote can be selected from the price distribution by identifying a likelier price quote associated to a maximal probability in the price distribution, and then selecting a price quote below the likelier price quote, as shown by the price selection line 52 in Fig. 6, and thereby increasing the chance of making a profit on a given transaction. At any point during negotiation between the producer and the broker, and preferably upon acceptance of a given price quote, the pond yield determination module 36 may send a virtual handshake to the producer 22 including an instruction to harvest the marine organisms 10 of the pond 12. The price selection line 52 can be selected based on a pre-determined percentile value factoring in the risk level that the broker is ready to tolerate when purchasing the pond’s organisms. The pre-determined percentile value is generally determined to reduce the risks of overpaying for the organisms 10 of the pond 12.

[0057] A given amount of time after a transaction, typically well after the processing of the marine organisms 10 of a pond 12 by the processor 14, the broker may receive a report indicating the actual pond yield for that pond 12. In such circumstances, the pond yield determination module 36 may update the transaction history bias, and more specifically the producer transaction history bias 44 and/or processor transaction history bias 46, based on that actual pond yield. By updating the transaction history biases 44, 46 over time, thereby increasing the number of transactions on which the bias is determined, the pond yield determination may be enhanced or refined.

[0058] The pond yield determination system 30 can be provided as a combination of hardware and software components. The hardware components can be implemented in the form of a computing device 70, an example of which is described with reference to Fig. 7. The software components of the pond yield determination system can be implemented in the form of a software application, a detailed example of which is described with reference to Fig. 8.

[0059] Referring to Fig. 7, the computing device 70 can have a processor 72, a memory 74, and I/O interface 76. Instructions 78 for determining the pond yield can be stored on the memory 74 and accessible by the processor 72.

[0060] The processor 72 can be, for example, a general-purpose microprocessor or microcontroller, a digital signal processing (DSP) processor, an integrated circuit, a field- programmable gate array (FPGA), a reconfigurable processor, a programmable read-only memory (PROM), or any combination thereof.

[0061 ] The memory 74 can include a suitable combination of any type of computer-readable memory that is located either internally or externally such as, for example, random-access memory (RAM), read-only memory (ROM), compact disc read-only memory (CDROM), electro-optical memory, magneto-optical memory, erasable programmable read-only memory (EPROM), and electrically erasable programmable read-only memory (EEPROM), Ferroelectric RAM (FRAM) or the like. [0062] Each I/O interface 76 enables the computing device 70 to interconnect with one or more input devices, such as a user interface, a mouse, a keyboard, a producer communication platform, or with one or more output devices such as a display, a memory system, an external network or a processor communication platform.

[0063] Each I/O interface 76 enables the pond yield determination system to communicate with other components, to exchange data with other components, to access and connect to network resources, to server applications, and perform other computing applications by connecting to a network (or multiple networks) capable of carrying data including the Internet, Ethernet, plain old telephone service (POTS) line, public switch telephone network (PSTN), integrated services digital network (ISDN), digital subscriber line (DSL), coaxial cable, fiber optics, satellite, mobile, wireless (e.g. Wi-Fi, WiMAX), SS7 signaling network, fixed line, local area network, wide area network, and others, including any combination of these.

[0064] Referring now to Fig. 8, the software application 80 is configured to receive the weight distribution data 82, the processing type 84, and to process them using the processing yield function, the producer transaction history bias 86, the processor transaction history bias 88 to determine and estimation of pond yield. In some embodiments, the software application 80 is stored on the memory 74 and accessible by the processor 72 of the computing device 70.

[0065] The computing device 70 and the software application 80 described above are meant to be examples only. Other suitable embodiments of the pond yield determination system can also be provided, as it will be apparent to the skilled reader.

[0066] For instance, in this specific embodiment, the software application 80 has a pond yield determination module which receives the weight distribution data 82, the selected processing type 84, and by accounting for producer transaction history bias 86 and for processor transaction history bias 88, determines a pond yield 90 indicative of post-processed weight values for the organisms of the pond. As shown, in this specific embodiment, the pond yield determination module can use images of samples of the marine organisms 92 of the pond in the determination. To acquire these images 92, the pond yield determination system can be equipped with one or more cameras communicatively coupled to the pond yield determination module, for instance. In a first aspect, the pond yield determination module can take these images into account for the determination of the pond yield for instance, by determining whether the initial weight distribution as estimated by the producer 22 are close to reality, and also by estimating a growth factor of these organisms based on their current growth level. Examples of such computing imaging techniques are described in PCT Published Patent Application No. WO 2020/124232, the contents of which are hereby incorporated by reference. The software application 80 may also include a rejection and classification module 94 which may be separate from the pond yield determination module, such as shown in Fig. 8, or part of the pond yield determination module in some other embodiments. The rejection and classification module 94 can evaluate, based on some images 92 of one or more samples of the pond’s marine organisms, on the transaction history bias 86, 88, what amount of marine organisms 10 are to be classified into each quality grades or even rejected based on some disease or other defects appearing on some marine organisms 10. Many of the transformations involved in the rejection and classification module 94 depend in part on visual morphological characteristics of the organisms that were harvested including, but not limited to, body-to-tail ratio and rejection on the basis of health-related characteristics, such as necrosis, black gills or red head, are examples of such characteristics. The image processing can provide a way to predict, for the transformation of a pond’s population of which one or more images of a sample were taken, more accurate estimates for the body-to-tail ratio distribution parameters and for the rejection distribution parameters. In the case of multiple images, note that these could have been taken at the same or at different moment of the lifetime of the population, in such a way that they would be indicated how the characteristics of the marine organisms evolved overtime. The determination of the pond yield can factor in the expected amount of marine organisms in the corresponding quality grades or the expected amount of rejected or otherwise discarded marine organisms to output a pond yield, or more specifically a fraction of biomass in each quality grade. For instance, Grade A can represent the best quality organisms, while Grades B and C can represent organisms of lower quality or commercial appeal. Grades B and C organisms are generally sold for a discounted price, depending on the markets. The pond yield determination can in some embodiments involves performing Monte Carlo simulations and/or Bayesian statistics for the pond yield at least based on the weight distribution data and on the transaction history bias, images of samples of the pond’s marine organisms, and the like. The pond yield can include a list of grades and the number of marine organisms in each grade.

[0067] Example - Pond Yield Determination Involving Bayesian Statistics

[0068] In the specific context of shrimp production, and in order to obtain more descriptive and more accurate results, the pond yield determination module can be configured to operate using the framework of Bayesian statistics. In this framework, each variable (e.g., the weight distribution, the processing yield function, the transaction history bias and the like) is characterized by an underlying probability distribution, and each calculation step yields the posteriori probability distribution of the output values.

[0069] It was found that variations around mean values obtained from limited sample size can often yield significantly different outcomes. Incorporating all these variations via probability distributions using a statistical approach is much more complex and rich. In the end, it gives results that are not only more accurate, but also allow for better decision-making. For example, when operating the engine using the framework of Bayesian statistics, one can obtain the probability distribution associated with the percentage of harvested shrimps in each class. When combined with a price list, this allows to calculate the probability distribution of the average price per unit of biomass of the pond. An example of which has been shown and described above with reference to Fig. 6. Such an output can give much more information than an average value to the on-site broker who is to negotiate with the producer for one or more ponds’ populations. For the producer, it indicates the range of prices they might get for his pond’s population, thus allowing to plan what money they will have available afterwards to start a new crop. For the processor, it indicates the range of prices they might need to pay out forthe crop, thus allowing to budget the expense more efficiently and plan for loans if needed. For the broker, which gets to pay the producer a cash advance for the crop in exchange of a small fee, this allows to calculate precisely the amount of money to advance while limiting the risk of overpaying and losing money. The following example shows how the different steps of the Bayesian calculation are carried out in one specific embodiment. However, it is understood that that this can be obtained by different methods without departing from the present disclosure. [0070] For example, if one measures the individual weights of a representative sample of harvested shrimp, the harvested weight distribution can be estimated using a kernel density estimator approach. Assuming a headless processing, one can assume that the processed weight w_p can be obtained from the harvest weight Wh by the following equation:

[0071] wp = w_h ■ f ■ h(w_K

[0072] It can be assumes that the weight change due to transport and the measurement bias at the processing plant are included in the same factor f, which follows a Normal distribution:

[0073] f ~ /V(n _0?)

[0074] an example representation of the processor transaction history bias. Parameters p_f and (j_f can be determined from empirical observations, expert knowledge and can be accessible from an accessible memory system. Moreover, from empirical evidence, it can be assumed that the tail-to-body weight ratio

is a function of the harvested weight and is governed by the following probability distribution:

[0075] h(w) ~ /V(n_h(w), <7_h)

[0076] with the mean following a linear relationship of the form:

[0077] n_h(w) = h_a + h_n w

[0078] Again, h_a, h_b and

can be determined empirically or from expert knowledge and can be accessible from an accessible memory system. As an example, for white leg shrimp, p_h can have a form such as the one shown in Fig. 4. With these assumptions, it is therefore possible to calculate the processed weight distribution. For given values of the distribution parameters, one may get a result that looks like the graph of Fig. 5A.

[0079] For the subsequent step, one can assume that the rejection rate from grade A to grade B, denoted r_B, is governed by a Bernouilli distribution and can be calculated by the following expression: [0080] r_B — p_B ■ 9_B + (1 p_B) ■ 9_B

[0081] where:

[0082] p_B ~ Bernoulli(p_p')

[0083] 9_B ~ Beta(a_B,p_B)

[0084] §B ~ Beta(a_B,PB)

[0085] Again, the parameters of all those distributions can be determined empirically orfrom expert knowledge. The parameters can also factor in the producer and/or processor biases by using the past transaction history to determine them empirically for a specific pair of parties involved in any given transaction.

[0086] Similar expressions for the rejection rates from grade A to the different categories of grade C can be assumed; note that in this case, a rate must be specified for each grade C category (such as juvenil, basura, etc.) because the latter do not only depend on the processed weight, like for grade A and B, but also other factors. It is noted that Grades A and B categories are determined by processing the weight (e.g. 21-30 units/kg). For Grade C, the categorization is more “qualitative” than “quantitative,” and can depend from one embodiment to another.

[0087] Now that all probability distributions have been specified, the full calculation can be carried out using the tools and formulas of Bayesian statistics. Another method is to perform Monte-Carlo simulations. While this does not give an exact result, it is often more convenient and provides an accurate enough estimation of the true output distributions. In this method, thousands of realizations of the process have been simulated, and for each realization, values from the specified probability distributions were randomly drawn. Then using all obtained final outputs, the pond yield determination module can calculate statistical quantities, such as the mean and standard deviation of the percentage of biomass observed in each shrimp category, preferably within a time window of about 10 minutes.

[0088] As can be understood, the examples described above and illustrated are intended to be exemplary only. The scope is indicated by the appended claims.

Claims

WHAT IS CLAIMED IS:

1 . A pond yield determination system comprising: a computing device having a microprocessor and a memory having stored thereon instructions that when executed by the microprocessor perform the steps of: accessing weight distribution data indicative of a weight distribution of marine organisms to be harvested from a pond by a producer, the weight distribution including a plurality of measured weight values associated to a sample of the marine organisms of the pond; selecting a processing type indicative of a type of processing according to which the marine organisms of the pond are to be processed by a processor, including retrieving a processing yield function indicative of a marine organism fraction remaining after processing as a function of marine organism weight for the type of processing; accessing a transaction history bias factoring in prior transactions including initial weight distributions and associated actual pond yields for the selected type of processing; and determining a pond yield upon applying the processing yield function and the transaction history bias to the weight distribution data, said determining being performed within a time window of about 10 minutes.

2. The pond yield determination system of claim 1 further comprising generating a price quote for the marine organisms of the pond based on the pond yield and on a price list including price values for a plurality of weight ranges.

3. The pond yield determination system of claim 2 wherein said generating includes generating a price probability curve indicative of a price distribution for the marine

22 organisms of the pond at least based on the pond yield, on the price list and on the transaction history bias, and selecting the price quote from the price probability curve.

4. The pond yield determination system of claim 3 wherein said selecting includes selecting the price quote based on a pre-determined percentile value.

5. The pond yield determination system of claim 2 further comprising instructing harvesting of the marine organisms of the pond upon acceptance on the price quote.

6. The pond yield determination system of claim 1 further comprising receiving an actual pond yield for the pond after said processing and updating the transaction history bias based on the received actual pond yield.

7. The pond yield determination system of claim 1 wherein the transaction history bias includes at least one of a producer transaction history bias indicative of prior weight distributions and associated actual pond yields cleared by the producer and a processor transaction history bias indicative of prior weight distributions and associated actual pond yields cleared by the processor.

8. The pond yield determination system of claim 1 wherein the sample includes the weight values of at least 10 marine organisms, preferably at least 100 marine organisms, and most preferably at least 200 marine organisms.

9. The pond yield determination system of claim 1 wherein the transaction history bias factors in at least 5 prior transactions, preferably at least 10 prior transactions and most preferably at least 40 prior transactions.

10. The pond yield determination system of claim 1 wherein said determining is performed within a time window below 30 minutes, preferably below 15 minutes and most preferably below 5 minutes.

1 1 . The pond yield determination system of claim 1 further comprising selecting a marine organism type indicative of a type of the marine organisms, the processing yield function and the transaction history bias pertaining to the selected marine organism type.

12. The pond yield determination system of claim 1 wherein the processing type is selected from a group consisting of: head-on processing, headless processing, shell-on processing, shell-off processing, and deveined processing.

13. The pond yield determination system of claim 1 wherein said determining a pond yield includes performing Monte Carlo simulations for the pond yield at least based on the weight distribution data and on the transaction history bias.

14. The pond yield determination system of claim 1 wherein said determining the pond yield includes outputting a plurality of post-processed weight values.

15. The pond yield determination system of claim 1 wherein said determining the pond yield includes outputting an expected number of post-processed organisms in each of a plurality of quality grades.

16. A method of determining a yield for a pond containing marine organisms, the method comprising: accessing weight distribution data indicative of a weight distribution of marine organisms to be harvested from the pond by a producer, the weight distribution including a plurality of measured weight values associated to a sample of the marine organisms of the pond; selecting a processing type indicative of a type of processing according to which the marine organisms of the pond are to be processed by a processor, including retrieving a processing yield function indicative of a marine organism fraction remaining after processing as a function of marine organism weight for the type of processing; accessing a transaction history bias factoring in prior transactions including initial weight distributions and associated actual pond yields for the selected type of processing; and determining a pond yield upon applying the processing yield function and the transaction history bias to the weight distribution data, said determining being performed within a time window of about 10 minutes.

17. The method of claim 16 further comprising generating a price quote for the marine organisms of the pond based on the pond yield and on a price list including price values for a plurality of weight ranges.

18. The method of claim 17 wherein said generating includes generating a price probability curve indicative of a price distribution for the marine organisms of the pond at least based on the pond yield, on the price list and on the transaction history bias, and selecting the price quote from the price probability curve.

19. The method of claim 18 wherein said selecting includes selecting the price quote based on a pre-determined percentile value.

20. The method of claim 17 further comprising instructing harvesting of the marine organisms of the pond upon acceptance on the price quote.

21 . The method of claim 16 further comprising receiving an actual pond yield for the pond and updating the transaction history bias based on the received actual pond yield.

22. The method of claim 16 wherein said determining the pond yield includes outputting a plurality of post-processed weight values.

23. The method of claim 16 wherein said determining the pond yield includes outputting an expected number of post-processed organisms in each of a plurality of quality grades.

25