Method for counting live mealworms or black soldier fly larvae
Patent Information
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-04-04
- Publication Date
- 2026-03-04
AI Technical Summary
Current methods for counting live Mealworm and Black Soldier Fly larvae are hindered by their small size and tendency to burrow into opaque organic substrates, making it difficult to accurately enumerate large quantities and distinguish them from food particles and exuviae, leading to errors in bioconversion yield monitoring and animal well-being assessment.
A process involving distributing larvae on a support in a thin layer, subjecting them to multiple light signals of different wavelengths, using a multispectral optical sensor to capture images, and employing image analysis software for reliable differentiation and counting, ensuring larvae remain visible and distinguishable from their substrate.
This method allows for precise and efficient counting of larvae with a reduced margin of error, enabling real-time monitoring of bioconversion yields and animal well-being, overcoming the limitations of existing techniques by differentiating larvae from their substrate and reducing errors in large quantity assessments.
Smart Images
Figure EP2024059240_31102024_PF_FP_ABST
Abstract
Description
[0001] DESCRIPTION
[0002] TITLE: Method for counting live larvae of mealworms or black soldier flies
[0003] GENERAL TECHNICAL FIELD
[0004] The present invention lies in the field of counting live insect larvae.
[0005] More specifically, it relates to a process of counting live larvae of mealworms or black soldier flies.
[0006] CONTEXT AND STATE OF THE ART
[0007] Entomology has allowed the description of the cycles of many insect species. Counting quickly became an essential method of evaluation for progress.
[0008] To date, the counting of animal and plant species is essential for the assessment of biodiversity within different environments.
[0009] Partial counting by counting represents a significant obstacle in entomology, due to the millimetric size of the individuals and the potentially large numbers, which most of the time requires the use of reduced samples, at the risk of their low representativeness or the use of optical magnification tools.
[0010] Before breeding, pest control saw the development of electronic counters in the 1960s. These include disadvantages related to population limits (less than 200), dead insects, and media (liquid). This device is not suitable for counting large numbers of live insects.
[0011] In 1997, document US5646404 highlighted the counting of larvae by infrared detection of flying insects passing through an air flow in order to control the infestation of agricultural commodity storage sites. This process has been taken up and improved on numerous occasions, notably by Wesley et al (2010).
[0012] For insect farming, the state of the art lists devices that result in the total or partial automation of a set of operations encountered (separation of the living environment (W02022 / 119442 or EP3979791 or W02019 / 053456 A1), sexing (W02022 / 070195), feeding for example). Among the millions of insect species, there are species with an aquatic larval stage, so that for their counting, it appears more accessible to adapt a technique used in aquaculture (Daoling et al, 2021).
[0013] For applications in entomology, biocontrol is used for the development of sorted populations (WO 2022 / 070195) or the breeding of edible insects (WO 2022 / 112770).
[0014] The aim of insect farming is to provide ingredients for livestock or domestic animal food chains, or even for human consumption.
[0015] Industries exploit species, two of which are reared at the larval stage, namely the Mealworm (Tenebrio Molitor, abbreviated as "TM") and the Black Soldier Fly larva (Hermeti hillucens, abbreviated as "MSN"). The millimetric size and high numbers of individuals make it impossible to assess in real time the numbers present in the breeding sites.
[0016] For convenience of language, these two species at the larval stage can be designated hereinafter TM and MSN.
[0017] Under these conditions, the breeding of such insects requires a tool to measure / evaluate the number of larvae in order to manage their activity, and particularly the bioconversion yields of the food resources made available to them.
[0018] The breeding of insect larvae requires the development of a new industry compared to the techniques used for livestock breeding known until then.
[0019] In fact, the breeding of these larvae takes the form of large production units (expressed in m 2 breeding area whose productivity is translated in terms of quantity of protein product) and numerous production sub-units (breeding tanks on average 1m 2 ).
[0020] Like any industrial system, performance measurement is essential to monitor process operations and validate equipment profitability. Monitoring larval growth is one of the primary levers for managing and estimating bioconversion (the latter being understood as the change in form of energy or matter (at the molecular level) involving a living being, which includes photosynthesis and cellular respiration). To date, industrial control measurements are carried out either by weighing production units at the end of growth and before processing “animal products” into proteins, or by sampling in breeding units (with a mass ratio that can be as low as 1 in 25,000).
[0021] On these samples of known masses, a count of individuals makes it possible to project the total biomass of larvae contained in a unit, a subset of unit or in the factory, but each time with a risk of error.
[0022] These locks do not allow the farm / factory to precisely monitor the bioconversion yields of organic resources allocated to feeding the larvae or to monitor the animal welfare of the larvae (mortality).
[0023] Among the techniques implemented to date, there are a number of drawbacks.
[0024] Several concern the physical support on which the insects and / or larvae are deliberately placed for the purpose of the counting operation. When the physical support is the living environment, it is almost exclusively aquatic in nature and therefore unsuitable for TM and MSN species which evolve in an opaque organic substrate.
[0025] Furthermore, the above-mentioned species, in the larval state, flee the light and bury themselves in the nourishing substrate.
[0026] By the expression "feeding substrate" is meant throughout the present application, including the claims, a solid, at least partly organic substrate on / in which the larvae are raised and on which they feed.
[0027] Furthermore, existing imaging solutions identify living insect forms recognized as larvae on liquid or solid substrates specifically designed to receive individuals for the purpose of counting. In any case, these substrates are different from a feeding substrate for rearing larvae.
[0028] This is explained by the fact that it is difficult to evaluate large quantities (greater than one kilogram) and especially to distinguish larvae without error in shape and / or color, compared to food particles and / or exuviae (which may have a similar shape and / or color).
[0029] There therefore remains an unresolved need for a technique that allows the counting of live TM and MSN larvae while they are growing on / in a feeding substrate, although these larvae tend, as indicated above, to flee from the light and burrow into said feeding substrate.
[0030] PRESENTATION OF THE INVENTION
[0031] To this end, the invention relates to a method for counting live larvae of Mealworms (Tenebrio Molitor) or Black Soldier Flies (Hermeti hillucens), characterized in that it comprises at least the following steps:
[0032] E1 / place a mass of Mealworm (Tenebrio Molitor) or Black Soldier Fly (Hermeti hillucens) larvae and nourishing substrate on a support;
[0033] E2 / distribute said mass in the form of a thin layer of such thickness that said larvae cannot burrow into said substrate and remain visible on the surface;
[0034] E3 / subjecting the same section of said thin layer to at least two light signals of different wavelengths, included in the range 200-1000 nanometers;
[0035] E4 / using a multispectral optical sensor, capture images of the said same section subjected to the said light signals;
[0036] E5 / reconstruct a single image by interlacing and assembling said images from step E4 / ;
[0037] E6 / count said larvae using a computer program for image analysis and recognition;
[0038] E7 / if necessary, repeat said steps E3 / to E6 / on a new section (SE).
[0039] Thanks to these characteristics, the exposure of the larvae present on the surface of the substrate (because they cannot bury themselves there) to light signals of different wavelengths allows, in return, the capture of reflectance spectra which, in turn, allow the larvae to be reliably differentiated from their substrate.
[0040] Thus, it is then possible to count said larvae using an image recognition computer program. This counting is carried out with a relatively small margin of error compared to the methods known from the prior art. According to other advantageous and non-limiting characteristics of the invention, taken alone or according to a technically compatible combination of at least two of them:
[0041] - the thickness of said thin layer is between 1 and 10 millimeters and preferably between 1 and 5 millimeters;
[0042] - said different wavelengths are respectively of the order of 560 and 780 nanometers;
[0043] - in step E4, a linear camera is used which integrates said sensor;
[0044] - in step E6, a program is used which works by automatic learning;
[0045] - in step E1, at least one mobile support is used.
[0046] - said at least one mobile support comprises at least one conveyor or one conveyor belt;
[0047] - in step E2, a single conveyor or rolling belt is used, the upper strand of which, which receives said mass, extends horizontally;
[0048] - in step E2, use is made of a pair of conveyors or conveyor belts whose upper strands which receive said mass extend horizontally, the downstream end of the first of said conveyors / conveyor belts extending vertically from the upstream end of the second of said conveyors / conveyor belts, and that said steps E3 and E4 are implemented in a zone of fall of said thin layer from the first to the second of said conveyors / conveyor belts;
[0049] - step E2, use is made of a pair of conveyors or rolling belts arranged in line with one another, the first having an upper strand which extends horizontally, while the second is inclined downwards;
[0050] - said step E2 is implemented in a transfer zone from the first to the second of said conveyors or conveyor belts, while said step E4 is implemented vertically above the second of said conveyors or conveyor belts.
[0051] DESCRIPTION OF FIGURES
[0052] Other characteristics and advantages of the invention will appear from the description which will now be given, with reference to the appended drawings, which represent, for informational but non-limiting purposes, a possible embodiment.
[0053] On these drawings:
[0054] [Fig. 1] is a flowchart which integrates the different stages of the method according to the invention;
[0055] [Fig. 2] is a schematic view of a first installation which is capable of being used for implementing the method according to the invention;
[0056] [Fig. 3] is a schematic view of a second installation which is capable of being used for implementing the method according to the invention;
[0057] [Fig. 4] is a schematic view of a third installation which is capable of being used for implementing the method according to the invention;
[0058] [Fig. 5] is a diagram illustrating how the images resulting from the implementation of the method according to the invention are processed.
[0059] DETAILED DESCRIPTION OF THE INVENTION
[0060] The method according to the invention, as it will be described according to a possible embodiment and with reference to the attached drawings, is applied to the living larvae of Mealworms (Tenebrio Molitor) or Black Soldier Flies (Hermeti hillucens) which will be referenced hereinafter TM and MSN.
[0061] These larvae have the particularity of seeing their weight multiplied by a factor of 6000 to 10000 in the space of a few weeks.
[0062] It is therefore understood that the number of larvae on their feeding substrate per breeding unit must be adapted to ensure the most efficient bioconversion possible.
[0063] It is in this context in particular that the method of the invention can be implemented.
[0064] It can also be used to estimate the total weight of larvae at the end of growth before possibly carrying out an actual weighing.
[0065] In the attached figures 2 to 4 is shown an installation I which can be used for implementing the method according to the present invention.
[0066] Step E1: depositing a mass of TM or TM larvae on a support
[0067] MSN and feed substrate. This first step of the method of the invention (see figure 1), which is illustrated in particular in figure 2, can be implemented manually or in a mechanized / automated manner, for example using a rotary distributor 1.
[0068] The mass M consists of a mixture of live larvae at a predetermined stage of growth and rearing and a nourishing substrate S, i.e. it serves not only as a living environment for the larvae, but also as food for them.
[0069] For purely indicative purposes, this substrate S may consist of organic matter from cereals and / or vegetables or vegetable food by-products (from fruits, vegetables, bakeries, dairies, starch factories, “ethanol factories” (i.e. ethanol production units for example) or animals (depending on the countries and regulations in force.
[0070] At the end of this step, the mass M is deposited in a thick and compact layer on a support constituted here by the upper strand 20 of a motorized conveyor belt or conveyor 2.
[0071] In a possible embodiment not shown in the attached figures, the support may be static, while the distributor 1 is moved along and above it.
[0072] In the embodiment of Figure 2, we are dealing with a single conveyor belt whose upper 20 and lower 22 strands are parallel and horizontal.
[0073] Step E2: distribution of mass in the form of a thin layer of such thickness that said larvae cannot bury themselves in said substrate and remain visible on the surface.
[0074] As previously stated, larvae have a natural tendency to flee from the light and bury themselves in the mass of substrate S that accommodates them.
[0075] This feature is in contradiction with the desire to implement a correct count of the number of larvae present in the substrate.
[0076] In order to overcome this difficulty, the present step aims to distribute the mass M in the form of a thin layer CM of such a thickness that the larvae cannot bury themselves in the substrate S, so that they remain visible on the surface. For purely indicative purposes, such a thin layer CM has a thickness of the order of 1 to 5 mm depending on the stage of development of the larvae. This adjustment can be extended to 10 millimeters depending on the type of larvae, the age or the physical presentation (particle size) of the substrate.
[0077] Before obtaining such a thin CM layer, use can be made of a static tool 3 in the form of a pallet as shown in FIG. 2, which does not constitute the heart of the invention.
[0078] In this alternative embodiment, not shown here, obtaining the CM thin layer can be carried out manually.
[0079] Step E3: subjecting the same section of said thin layer to at least two light signals of different wavelengths.
[0080] As can be seen in Figure 2, the installation I comprises, vertically above the upper strand 20 of the conveyor belt or conveyor 2, two lighting bars 4 and 4' which are oriented towards the same section SE of the thin layer CM.
[0081] By "same section" we mean the surface of the thin layer which is illuminated by the two bars 4 and 4'.
[0082] Using these bars 4 and 4', the SE section is subjected to at least two light signals of different wavelengths, in the range 200-1000 nanometers.
[0083] Advantageously, these signals are emitted alternately.
[0084] Preferably, wavelengths of the order of 560 to 860 nanometers are used.
[0085] As an indication, for this lighting phase, you can use Chromasens brand equipment, and more specifically the Corona II model LED bar.
[0086] Step E4: Capture images of the section subjected to light signals.
[0087] Between the bars 4 and 4' is positioned a multispectral optical sensor 5, for example integrated within a linear camera, which is shaped to capture and record the images reflected by the section SE which has been illuminated. These images translate the reflectance spectrum of the substrate S and the larvae at the wavelength considered.
[0088] These images are referenced ii and 1'2 in Figure 5.
[0089] As for images reflected from the SE section, because the light signals emitted in the previous step have different wavelengths, these images have significantly different characteristics, in terms of contrast of the larvae relative to the substrate, etc.
[0090] Step E5: Reconstruction of a single image by interlacing and assembling the images from step E4.
[0091] Using software such as that known as Aurora Vision from Zebra, the images ii and 1'2 obtained previously are then interlaced and assembled to obtain a single image 1'3 (see figure 5) in which the lines of images ii and 1'2 are arranged alternately.
[0092] Although this is not visible, we thus obtain a single image whose contrast and sharpness are a priori sufficient to allow us to discern and identify the larvae in relation to the substrate S.
[0093] Step 5a: Image processing to differentiate the larvae
[0094] The image processing method uses a pattern recognition learning algorithm. This algorithm was built and trained on a set of images labeled by business experts.
[0095] Step E5 ter: Sequencing of the relative displacement of the image to enable calculation.
[0096] This step corresponds to the segmentation of the linear image into several individually processed images with the aim of losing as little information as possible, particularly if one or more larvae overlap several images.
[0097] Step E6: counting larvae using image analysis and recognition software To implement this step, we will preferably use an image analysis algorithm (called "deep learning", which uses a multi-layer deep neural network) allowing, after learning, to identify areas labeled as "larvae", with a view to counting them.
[0098] Such a learning method, with "human" labeling, in "deep learning" makes it possible to improve the differentiation of larvae in relation to the substrate, whatever their size, whether they are clumped together or not, and whatever the direction of their arrangement in the substrate.
[0099] The evolutions of the neural network are dependent on the weighting values determined by learning. The initial and iterated values as the present technique is implemented are able to evolve according to the developments of new criteria that the present invention allows.
[0100] Among the new criteria, it may be possible to identify the shape, size, thickness, length, sexual dimorphism, color, and any other element of definitive or punctual dimorphism in particular linked to the appearance of pathology, relating to the larvae.
[0101] Step E7: repeat steps E3 to E6 on a new section.
[0102] If necessary, the steps just described are implemented again on a new section of the CM thin layer.
[0103] Step E8: Observation results
[0104] The output data varies depending on the criterion. For counting, this is the number of larvae present per image. For size, thickness, or length, individual results are expressed in millimeters. For sex, this is the number of larvae per sexual type per image.
[0105] Step E9: Exit Information
[0106] All output information is expressed per batch. For counting, this refers to the total number of larvae and / or the number by sexual type. For size, thickness, or length, individual results are expressed using traditional descriptive indicators (mean, median, standard deviation, etc.).
[0107] The installation I of Figure 3 differs from that of Figure 2 in that use is made of a pair of conveyors or conveyor belts 2 and 2', the upper strands 20, respectively 20', which receive said mass M, extend horizontally. The downstream end of the first 2 of these conveyors / conveyor belts extends vertically from the upstream end of the second 2'. In addition, the steps E3 and E4 described above are implemented in the zone of fall of the thin layer from the first to the second of said conveyors / conveyor belts.
[0108] For this purpose, materials 4, 4' and 5 are arranged appropriately.
[0109] Such an arrangement can be used for reasons of compactness, for example in a room where the available surface area is limited.
[0110] The installation of Figure 4 differs from that of Figure 2 in that a pair of conveyors or rolling belts 2 and 2' are used, which are arranged in line with one another. The first 2 has an upper strand 20 which extends horizontally. The second 2' is inclined downwards. The aforementioned step E2 is implemented in the transition zone from the first to the second rolling belt / conveyor. As for the steps E3 and E4 described above, they are implemented in the central region of the second rolling belt / conveyor 2'.
[0111] For this purpose, materials 4, 4' and 5 are arranged appropriately.
[0112] Such an arrangement can be used for reasons of compactness, for example in a room where the collection of the larvae which have undergone counting cannot be done in the same room as the deposit of the mass M on the conveyor belt / roller2, but in another room at a lower level.
[0113] Thus, the present invention aims to carry out a larval count by analyzing still or dynamic images taken by a multi-spectral sensor / camera after alternating illumination at two wavelengths of a mixture of organic matter composed of insect larvae and its substrate (feeding and habitat). The difficulties of counting larvae in imaging, as with the human eye, are based both on the effectiveness of shape recognition attributable to the object sought but also in our application on the differentiation between the larvae and the living / feeding substrate itself, the heterogeneity of larval sizes (linked among other things to differences in growth speed), and the mobility of the larvae which tend to flee into the substrate to avoid the light.
[0114] The method according to the invention not only makes it possible to differentiate larvae in relation to the substrate formed by food particles, but also in relation to foreign bodies and / or dead particles (exuviae). It also offers the possibility of using the information captured to evaluate other key performance indicators of larval rearing and / or their morphological details (length, width, thickness, arrangement of morphological details, posture curvature, etc.), both on average values and their dispersions. This method also makes it possible to count larvae separated from their living environment, upstream (i.e. from hatching) and downstream (at the end of growth or before slaughter).
Claims
CLAIMS 1. Method for counting live larvae of Mealworms (Tenebrio Molitor) or Black Soldier Flies (Hermeti hillucens), characterized in that it comprises at least the following steps: E1 / place on a support a mass (M) of Mealworm (Tenebrio Molitor) or Black Soldier Fly (Hermeti hillucens) larvae and nourishing substrate (S); E2 / distribute said mass (M) in the form of a thin layer (CM) of such thickness that said larvae cannot bury themselves in said substrate (S) and remain visible on the surface; E3 / subjecting the same section (SE) of said thin layer (CM) to at least two light signals of different wavelengths, included in the range 200-1000 nanometers; E4 / using a multispectral optical sensor (5), capture the images (ii, 1'2) of said same section (SE) subjected to said light signals; E5 / reconstruct a single image (1'3) by interlacing and assembling said images (11, 12) from step E4 / ; E6 / count said larvae using a computer program for image analysis and recognition; E7 / if necessary, repeat said steps E3 / to E6 / on a new section (SE).
2. Method according to claim 1, characterized in that said different wavelengths are respectively of the order of 560 and 780 nanometers.
3. Method according to claim 1 or 2, characterized in that the thickness of said thin layer (CM) is between 1 and 10 millimeters and preferably between 1 and 5 millimeters.
4. Method according to one of claims 1 to 3, characterized in that, in step E4, use is made of a linear camera which integrates said sensor (5).
5. Method according to one of the preceding claims, characterized in that, in step E6, use is made of a program which operates by automatic learning.
6. Method according to one of the preceding claims, characterized in that, in step E1, at least one mobile support is used.
7. Method according to claim 6, characterized in that said at least one mobile support comprises at least one conveyor or one conveyor belt (2, 2').
8. Method according to claim 7, characterized in that, in step E2, a single conveyor or rolling belt (2) is used, the upper strand (20) of which, which receives said mass (M), extends horizontally.
9. Method according to claim 7, characterized in that, in step E2, use is made of a pair of conveyors or conveyor belts (2, 2') whose upper strands (20, 20') which receive said mass (M) extend horizontally, the downstream end of the first (2) of said conveyors / conveyor belts extending vertically from the upstream end of the second (2') of said conveyors / conveyor belts, and that said steps E3 and E4 are implemented in a zone of fall of said thin layer (CM) from the first (2) to the second (2') of said conveyors / conveyor belts.
10. Method according to claim 7, characterized in that, in step E2, use is made of a pair of conveyors or rolling belts (2, 2') arranged in the extension of one another, the first (2) comprising an upper strand (20) which extends horizontally, while the second (2') is inclined downwards.
11. Method according to claim 10, characterized in that said step E2 is implemented in a transfer zone from the first (2) to the second (2') of said conveyors or conveyor belts, while said step E4 is implemented vertically above the second (2') of said conveyors or conveyor belts.