Silkworm maturing state detection system based on improved YOLOv8

Abstract

The invention relates to the technical field of silkworm breeding, and discloses an improved YOLOv8-based silkworm maturity state detection system, an acquisition unit acquires image data of silkworms, and a model unit performs silkworm recognition in the image data by adopting an image recognition model of instance segmentation; the iteration unit is adopted to update and iterate an image recognition model before image recognition, the algorithm unit provides algorithm support for the model unit, and the processing unit cuts a silkworm image recognized by the model unit and calculates a volume change rate TJ and a color change rate YS; and the central unit receives the volume change rate TJ and the color change rate YS sent by the processing unit and calculates a decision value P. According to the invention, silkworm recognition in image data is carried out by improving an image recognition model of YOLOv8 instance segmentation, the characteristics of mature silkworms and immature silkworms are concluded and learned, and the number of various silkworms is rapidly and accurately classified and counted.

Description

Technical Field

[0001] The present invention relates to the technical field of silkworm breeding, and more specifically to a silkworm maturity state detection system based on improved YOLOv8. Background Art

[0002] Silkworms are insects of the genus Bombyx mori, family Bombyx mori, order Lepidoptera. The silk they produce is an important textile raw material. Silkworms produce silk cocoons to produce silk, which is mostly white and soft to the touch and can be used to weave high-grade silk. Silkworm skins and dead silkworms are valuable traditional Chinese medicines. Silkworm feces and excrement are also high-quality feed for livestock, and can be used to extract chlorophyll, copper sodium salt and other industrial raw materials. Silkworm pupae can be fried and eaten, and pupa protein and pupa oil can also be extracted for use in medicine.

[0003] The automated breeding method of silkworms is a popular research direction at present. At present, the determination of the maturity of silkworms is still mainly based on the subjective judgment of breeders, which has the problems of strong subjectivity, low efficiency, low accuracy, and easy to cause visual fatigue.

[0004] When silkworms are being bred, they will stop eating after they reach maturity. The determination of the maturity of silkworms can also help silkworm farmers reduce feeding pressure and reduce the amount of mulberry leaves picked in time. Good mulberry feeding is the basis. If the silkworms are ripened and clustered too early, the quality of the cocoons will inevitably decline. If the silkworms are clustered too late, it will cause a waste of silk and directly affect the cocoon yield. In order to promote the uniform clustering of mature silkworms, molting hormones are added when the amount of mature silkworms reaches about 5%. The use of molting hormones at this time not only has no adverse effect on the quality of the cocoon products, but on the contrary can increase the square cluster hole rate, reduce hairy cocoons, improve cocoon quality, and is beneficial to clustering operations. However, it is currently impossible to accurately judge the maturity of silkworms. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the implementation regulations of the present invention provide a silkworm maturity state detection system based on improved YOLOv8 to solve the technical problems raised in the background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a mulberry silkworm maturity state detection system based on improved YOLOv8, comprising a collection unit, a model unit, an iteration unit, an algorithm unit, a processing unit, a central unit, a display unit, a verification unit and a feeding unit, wherein the collection unit collects image data of mulberry silkworms, the model unit adopts an instance segmentation image recognition model to perform mulberry silkworm recognition in the image data, the iteration unit adopts the iteration unit to perform image recognition model update iteration before performing image recognition, the algorithm unit provides algorithm support to the model unit, the processing unit cuts the mulberry silkworm image recognized by the model unit and calculates the volume change rate TJ and the color change rate YS, the central unit receives the volume change rate TJ and the color change rate YS sent by the processing unit and calculates the judgment value P, the display unit receives the maturity instruction and the feeding instruction sent by the central unit and displays them, the verification unit calculates the real-time correction value T in the judgment value P, and the feeding unit performs the feeding of mulberry silkworms;

[0007] The iteration unit includes a training module, a test module and a verification module. The training module receives the image collected by the image acquisition unit and performs image annotation. The training module annotates the image by instance segmentation. The training module inputs the annotated image into the verification module for verification. After the verification result of the training module meets the requirements, the test module receives the image in the image acquisition unit for testing. After the use result of the test module meets the requirements, the iteration unit sends the latest parameters to the model unit, and the model unit uses the latest parameters to identify silkworms.

[0008] In a preferred embodiment, a CARAFE sampling operator is introduced into the model unit, the CARAFE sampling operator upsamples the low-resolution feature map and generates a feature map of the same size as the high-resolution feature map, the CARAFE sampling operator divides the low-resolution feature map into several small blocks, and then interacts each small block with the high-resolution feature map to obtain contextual information required for reconstruction, the CARAFE operator performs a convolution operation on the interacted feature map to obtain a sampling result, and performs silkworm boundary recognition.

[0009] In a preferred embodiment, the algorithm unit adopts the YOLOv8 backbone network, and the algorithm unit adds a DAT attention mechanism to the YOLOv8 backbone network, the last two C2F modules of the YOLOv8 backbone network in the algorithm unit are replaced by C2f_DCNv2 modules, the C2f_DCNv2 module combines the C2f module and the DCNv2 module, and the Conv operation in the C2f module Bottleneck is replaced by DCNv2 to form a new DBottleneck, and each C2f-DCNv2 module consists of two DCNv2 and n DBottleneck modules.

[0010] In a preferred embodiment, the processing unit receives the image recognized by the model unit, and the processing unit uses a bitmap cutting method to cut the silkworm image recognized by the model unit. The processing unit performs bitmap cutting every ten minutes, and the processing unit compares the image after the latest image cutting with the image after the first bitmap cutting of the day. The processing unit compares the volume change rate TJ and color change rate YS of the silkworm itself, and the processing unit sends the calculated results to the central unit.

[0011] In a preferred embodiment, when the processing unit calculates the volume change rate, the processing unit outputs the latest volume ZT of each silkworm after the latest bitmap cutting and the previous volume ST after the previous side bitmap cutting. The calculation formula of the volume change rate TJ is: The processing unit performs grayscale processing on the image of each silkworm, and the calculation formula for grayscale processing is W=0.3R+0.58G+0.12B, where W is the grayscale value after processing, R is a red image, G is a green image, and B is a blue image, and R, G, and B are all within the grayscale value range of 0-255. The processing unit outputs the grayscale value after the latest side bitmap is cut as WZ, and the processing unit outputs the grayscale value after the previous bitmap is cut as WS. The calculation formula of the color change rate YS is:

[0012]

[0013] In a preferred embodiment, the central unit receives the volume change rate TJ and the color change rate YS and calculates the judgment value P. The calculation formula of the judgment value P is: Where k1 and k2 are weights, and 0≤k1≤1, 0≤k2≤1, TS is the current feeding days of silkworms, BZ is the standard number of days required for silkworms to mature, and T is the real-time correction value.

[0014] In a preferred embodiment, the central unit compares the calculated judgment value P with its internal judgment threshold Y. When the judgment value P≥threshold Y, the central unit sends a maturity instruction to the display unit. When the judgment value P<threshold Y, the central unit sends a feeding instruction to the display unit. The display unit displays the maturity instruction and the feeding instruction, and when the feeding instruction in the display unit becomes a maturity instruction, the display unit emits an alarm to prompt.

[0015] In a preferred embodiment, the inspection unit goes to the location of the silkworms for inspection after the central unit issues a maturity instruction, and the inspection unit inspects the silkworm maturity CS on site. The inspection unit calculates the real-time correction value T using the formula T=[1+(CS-5%)]. When the inspection unit conducts the inspection, the silkworm maturity CS is input, and when the silkworm maturity CS is not input during the inspection, the real-time correction value T does not change.

[0016] In a preferred embodiment, the feeding unit feeds silkworms, and when the central unit does not send a maturity instruction, the feeding unit adds mulberry leaves, and when the central unit sends a maturity instruction, the feeding unit adds ecdysone to the mulberry leaves added by the feeding unit, and the feeding unit reduces the amount of mulberry leaves added after adding the ecdysone.

[0017] Technical effects and advantages of the present invention:

[0018] 1. The present invention uses an image recognition model based on improved YOLOv8 instance segmentation to identify silkworms in image data, summarizes and learns the characteristics of mature silkworms and immature silkworms, quickly and accurately classifies and counts the number of various types of silkworms, annotates the collected silkworm images in the actual breeding process, and performs iterative training, so that the model unit is more accurate in silkworm identification;

[0019] 2. The present invention can improve the model's ability to capture details and silkworm body boundaries by introducing the CARAFE upsampling operator in the model unit, which is conducive to distinguishing silkworm individuals, making the model prediction more reliable, and thus enabling more accurate identification of silkworms. The DAT attention mechanism is added to the backbone network of YOLOv8 to help the network better capture the detailed features of silkworms, making feature fusion more accurate and targeted;

[0020] 3. The present invention ensures the accuracy of the volume change rate TJ and the color change rate YS, and combines the feeding days TS to calculate the judgment value P that can accurately express the changes of the silkworms. When the judgment value P ≥ threshold value Y, the amount of mature silkworms reaches 5%. Therefore, when the feeding process needs to be adjusted, adding molting hormone can increase the grid cluster penetration rate, reduce hairy cocoons, improve cocoon quality, and facilitate clustering operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the overall system composition of the present invention.

[0022] Figure 2 Schematic diagram of the C2f_DCNv2 module of the present invention.

[0023] Figure 3 This is a schematic diagram of the labeling operation before the model unit of the present invention is identified.

[0024] Figure 4 This is a schematic diagram of the results after identification by the present invention. DETAILED DESCRIPTION

[0025] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are only examples. The silkworm maturity state detection system based on improved YOLOv8 involved in the present invention is not limited to the various structures recorded in the following embodiments. All other implementations obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0026] The invention provides a mulberry silkworm maturity state detection system based on improved YOLOv8, comprising a collection unit, a model unit, an iteration unit, an algorithm unit, a processing unit, a central unit, a display unit, a verification unit and a feeding unit, wherein the collection unit collects image data of mulberry silkworms, the model unit adopts an instance segmentation image recognition model to perform mulberry silkworm recognition in the image data, the iteration unit adopts the iteration unit to perform image recognition model update iteration before performing image recognition, the algorithm unit gives algorithm support to the model unit, the processing unit cuts the mulberry silkworm image recognized by the model unit and calculates a volume change rate TJ and a color change rate YS, the central unit receives the volume change rate TJ and the color change rate YS sent by the processing unit and calculates a judgment value P, the display unit receives a maturity instruction and a feeding instruction sent by the central unit and displays them, the verification unit calculates a real-time correction value T in the judgment value P, and the feeding unit feeds the silkworms.

[0027] In the embodiments of the present application, the present application adopts an image recognition model based on improved YOLOv8 instance segmentation to identify silkworms in image data, summarizes and learns the characteristics of mature silkworms and immature silkworms, and quickly and accurately classifies and counts the number of various types of silkworms, which helps silkworm farmers to add molting hormones in time, cluster on time, and reduce silk waste. At the same time, if immature silkworms are clustered in advance, their behavior and excrement will cause contamination of cocoons and cluster tools, thereby increasing the cocoon output rate.

[0028] Referring to the figure, the iteration unit includes a training module, a testing module and a verification module. The training module receives the image captured by the image acquisition unit and performs image annotation. The training module annotates the image by instance segmentation. The training module inputs the annotated image into the verification module for verification. After the verification result of the training module meets the requirements, the testing module receives the image in the image acquisition unit for testing. After the use result of the testing module meets the requirements, the iteration unit sends the latest parameters to the model unit.

[0029] In the embodiment of the present application, based on the instance segmentation method, the present application performs annotation operations on the collected silkworm images in the actual breeding process, creates a data set, and divides it into a training module, a test module, and a verification module, performs iterative training, keeps the model unit in an updated state at all times, and promptly corrects any problems it has, ultimately making the model unit more accurate in silkworm identification.

[0030] Referring to the figure, the CARAFE sampling operator is introduced into the model unit. The CARAFE sampling operator upsamples the low-resolution feature map and generates a feature map of the same size as the high-resolution feature map. The CARAFE sampling operator divides the low-resolution feature map into several small blocks, and then interacts each small block with the high-resolution feature map to obtain the context information required for reconstruction. The CARAFE operator performs a convolution operation on the interacted feature map to obtain a sampling result for silkworm boundary recognition.

[0031] In the embodiments of the present application, in the actual breeding environment of mulberry silkworms, the silkworm bodies are densely stacked and the boundaries are unclear. Therefore, the present application introduces the CARAFE upsampling operator in the model unit to improve the model's ability to capture details and silkworm body boundaries, which is conducive to distinguishing individual silkworms, making the model prediction more reliable, and thus enabling more accurate identification of silkworms.

[0032] Referring to the figure, the algorithm unit adopts the YOLOv8 backbone network, and the algorithm unit adds the DAT attention mechanism to the YOLOv8 backbone network. The last two C2F modules of the YOLOv8 backbone network in the algorithm unit are replaced by C2f_DCNv2 modules. The C2f_DCNv2 module combines the C2f module and the DCNv2 module. The Conv operation in the Bottleneck of the C2f module is replaced by DCNv2 to form a new DBottleneck. Each C2f-DCNv2 module consists of two DCNv2 and n DBottleneck modules.

[0033] In the embodiment of the present application, the attention mechanism is a computational model that imitates the way human attention is allocated. The allocation of attention can be adjusted according to the relevance and importance of the input, so that the model can process information more effectively. In view of the problems of dense silkworm bodies, mutual occlusion, and similar characteristics of immature silkworms and mature silkworms in the actual breeding environment, the present application adds a DAT attention mechanism to the backbone network of YOLOv8 to help the network better capture the detailed characteristics of silkworms, making feature fusion more accurate and targeted, thereby improving the detection accuracy and the characterization ability of the detection network. The C2f_DCNv2 module is designed by combining the C2f module and the DCNv2 module, and the last two C2F modules in the backbone network are replaced by C2f_DCNv2 modules. The Conv operation in the Bottleneck of the C2f module is replaced by DCNv2 to form a new DBottleneck. Each C2f-DCNv2 module consists of two DCNv2 and n DBottleneck modules, which can better extract features from different silkworm bodies.

[0034] Referring to the figure, the processing unit receives the image recognized by the model unit, and the processing unit cuts the silkworm image recognized by the model unit by a bitmap cutting method. The processing unit performs bitmap cutting every ten minutes, and the processing unit compares the image after the latest image cutting with the image after the first bitmap cutting of the day. The processing unit compares the volume change rate TJ and the color change rate YS of the silkworm itself, and the processing unit sends the calculated value to the central unit. When the processing unit calculates the volume change rate, the processing unit outputs the latest volume ZT of each silkworm after the latest bitmap cutting and the previous volume ST after the bitmap cutting on the upper side. The calculation formula of the volume change rate TJ is: The processing unit performs grayscale processing on the image of each silkworm, and the calculation formula for grayscale processing is W=0.3R+0.58G+0.12B, where W is the grayscale value after processing, R is a red image, G is a green image, and B is a blue image, and R, G, and B are all within the grayscale value range of 0-255. The processing unit outputs the grayscale value after the latest side bitmap is cut as WZ, and the processing unit outputs the grayscale value after the previous bitmap is cut as WS. The calculation formula of the color change rate YS is:

[0035] In the embodiment of the present application, after the mulberry silkworm is accurately identified by the model unit, the image of the mulberry silkworm identified by the model unit is cut every five minutes using a bitmap cutting method. At this time, the images before and after are compared to understand the state changes of the mulberry silkworm itself. The mulberry silkworm will undergo two changes when it matures. The first part is that its body becomes shorter after maturity, and the second is that the chest of the silkworm gradually becomes translucent, and then the abdomen gradually becomes translucent, so the color changes. The present application calculates the volume change rate TJ and the color change rate YS to quickly understand whether the mulberry silkworm itself is mature. The present application adopts a bitmap cutting method to ensure the accuracy of the volume change rate TJ and the color change rate YS calculation. After the first bitmap cutting of the mulberry silkworm is performed every day, subsequent changes are based on this. This is because the maturity of the silkworm is in a constantly changing state. Therefore, the first collection of the day is based on the subsequent changes to ensure the accuracy of the detection.

[0036] Referring to the figure, the central unit receives the volume change rate TJ and the color change rate YS and calculates the judgment value P. The calculation formula of the judgment value P is: Wherein k1 and k2 are weights, and 0≤k1≤1, 0≤k2≤1, TS is the current feeding days of mulberry silkworms, BZ is the standard number of days required for mulberry silkworms to be fed to maturity, T is a real-time correction value, and the central unit compares the calculated judgment value P with its internal judgment threshold Y. When the judgment value P≥threshold Y, the central unit sends a maturity instruction to the display unit. When the judgment value P<threshold Y, the central unit sends a feeding instruction to the display unit. The display unit displays the maturity instruction and the feeding instruction, and when the feeding instruction in the display unit becomes a maturity instruction, the display unit emits an alarm to prompt.

[0037] In the embodiment of the present application, based on the calculated volume change rate TJ and color change rate YS, that is, the two and the number of days of feeding the silkworms, the calculated judgment value P can accurately express the changes in the silkworms. When the judgment value P ≥ threshold value Y, the amount of mature silkworms reaches 5%, so the feeding process needs to be adjusted. The present application can make adjustments in time, and when the display unit receives the maturity instruction, it will sound an alarm to remind the feeding personnel to make timely adjustments.

[0038] Referring to the figure, the inspection unit goes to the location of the silkworms for inspection after the central unit issues a maturity instruction, and the inspection unit inspects the silkworm maturity CS on site. The calculation formula of the real-time correction value T performed by the inspection unit is T=[1+(CS-5%)]. When the inspection unit performs the inspection, the silkworm maturity CS is input, and when the silkworm maturity CS is not input during the inspection, the real-time correction value T does not change. The feeding unit feeds the silkworms. When the central unit does not send a maturity instruction, the feeding unit adds mulberry leaves. When the central unit sends a maturity instruction, the feeding unit adds molting hormone to the mulberry leaves added by the feeding unit, and the feeding unit reduces the amount of mulberry leaves added after adding the molting hormone.

[0039] In the embodiment of the present application, after the maturity changes are carried out through the system, the inspection unit goes to the location of the silkworms to inspect the silkworm maturity CS, and determines whether the silkworm maturity CS is 5%. If it fluctuates around 5%, the real-time correction value T will be adjusted at this time, and then the judgment value P will be adjusted, so that the judgment value P is continuously adjusted and more accurate. When the silkworm maturity reaches 5%, the molting hormone is added in time, which not only has no adverse effect on the quality of the silkworm cocoon product, but on the contrary can increase the square cluster entry rate, reduce hairy cocoons, improve the cocoon quality, and is beneficial to the cluster operation.

[0040] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The units and algorithm steps of each example described in the embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0041] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0042] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

[0043] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A silkworm maturity state detection system based on improved YOLOv8, characterized in that: It includes a collection unit, a model unit, an iteration unit, an algorithm unit, a processing unit, a central unit, a display unit, a testing unit and a feeding unit, wherein the collection unit collects image data of mulberry silkworms, the model unit adopts an instance segmentation image recognition model to identify mulberry silkworms in the image data, the iteration unit adopts the iteration unit to update and iterate the image recognition model before image recognition, the algorithm unit provides algorithm support to the model unit, the processing unit cuts the mulberry silkworm image identified by the model unit and calculates the volume change rate TJ and the color change rate YS, the central unit receives the volume change rate TJ and the color change rate YS sent by the processing unit and calculates the judgment value P, the display unit receives the maturation instruction and the feeding instruction sent by the central unit and displays them, the testing unit calculates the real-time correction value T in the judgment value P, and the feeding unit feeds the silkworms; The iteration unit includes a training module, a test module and a verification module. The training module receives the image collected by the image acquisition unit and performs image annotation. The training module annotates the image by instance segmentation. The training module inputs the annotated image into the verification module for verification. After the verification result of the training module meets the requirements, the test module receives the image in the image acquisition unit for testing. After the use result of the test module meets the requirements, the iteration unit sends the latest parameters to the model unit, and the model unit uses the latest parameters to identify silkworms.

2. The silkworm maturity state detection system based on improved YOLOv8 according to claim 1, characterized in that: The CARAFE sampling operator is introduced into the model unit. The CARAFE sampling operator upsamples the low-resolution feature map and generates a feature map of the same size as the high-resolution feature map. The CARAFE sampling operator divides the low-resolution feature map into several small blocks, and then interacts each small block with the high-resolution feature map to obtain context information required for reconstruction. The CARAFE operator performs a convolution operation on the interacted feature map to obtain a sampling result for silkworm boundary recognition.

3. The silkworm maturity state detection system based on improved YOLOv8 according to claim 1, characterized in that: The algorithm unit adopts the YOLOv8 backbone network, and the algorithm unit adds a DAT attention mechanism to the YOLOv8 backbone network. The last two C2F modules of the YOLOv8 backbone network in the algorithm unit are replaced by C2f_DCNv2 modules. The C2f_DCNv2 module combines the C2f module and the DCNv2 module. The Conv operation in the Bottleneck of the C2f module is replaced by DCNv2 to form a new DBottleneck. Each C2f-DCNv2 module consists of two DCNv2 and n DBottleneck modules.

4. The silkworm maturity state detection system based on improved YOLOv8 according to claim 1, characterized in that: The processing unit receives the image recognized by the model unit, and the processing unit uses a bitmap cutting method to cut the silkworm image recognized by the model unit. The processing unit performs bitmap cutting every ten minutes, and the processing unit compares the image after the latest image cutting with the image after the first bitmap cutting of the day. The processing unit compares the volume change rate TJ and color change rate YS of the silkworm itself, and the processing unit sends the calculated results to the central unit.

5. The silkworm maturity state detection system based on improved YOLOv8 according to claim 4, characterized in that: When the processing unit calculates the volume change rate, the processing unit outputs the latest volume ZT of each silkworm after the latest bitmap cutting and the previous volume ST after the previous side bitmap cutting. The calculation formula of the volume change rate TJ is: The processing unit performs grayscale processing on the image of each silkworm, and the calculation formula for grayscale processing is W=0.3R+0.58G+0.12B, where W is the grayscale value after processing, R is a red image, G is a green image, and B is a blue image, and R, G, and B are all within the grayscale value range of 0-255. The processing unit outputs the grayscale value after the latest side bitmap is cut as WZ, and the processing unit outputs the grayscale value after the previous bitmap is cut as WS. The calculation formula of the color change rate YS is:

6. The silkworm maturity state detection system based on improved YOLOv8 according to claim 5, characterized in that: The central unit receives the volume change rate TJ and the color change rate YS and calculates the judgment value P. The calculation formula of the judgment value P is: Where k1 and k2 are weights, and 0≤k1≤1, 0≤k2≤1, TS is the current feeding days of silkworms, BZ is the standard number of days required for silkworms to mature, and T is the real-time correction value.

7. The silkworm maturity state detection system based on improved YOLOv8 according to claim 6, characterized in that: The central unit compares the calculated judgment value P with its internal judgment threshold value Y. When the judgment value P≥threshold value Y, the central unit sends a maturity instruction to the display unit. When the judgment value P<threshold value Y, the central unit sends a feeding instruction to the display unit. The display unit displays the maturity instruction and the feeding instruction. When the feeding instruction in the display unit becomes a maturity instruction, the display unit emits an alarm to prompt.

8. The silkworm maturity state detection system based on improved YOLOv8 according to claim 1, characterized in that: After the central unit issues a maturity instruction, the inspection unit goes to the location of the silkworms for inspection, and the inspection unit inspects the silkworm maturity CS on site. The inspection unit calculates the real-time correction value T using the formula T=[1+(CS-5%)]. When the inspection unit conducts the inspection, the silkworm maturity CS is input, and when the silkworm maturity CS is not input during the inspection, the real-time correction value T does not change.

9. The silkworm maturity state detection system based on improved YOLOv8 according to claim 1, characterized in that: The feeding unit feeds the silkworms. When the central unit does not send a maturity instruction, the feeding unit adds mulberry leaves. When the central unit sends a maturity instruction, the feeding unit adds ecdysone to the mulberry leaves added by the feeding unit, and the feeding unit reduces the amount of mulberry leaves added after adding the ecdysone.

Images