Method for estimating length of underwater organism and underwater organism length estimation system

Abstract

A method and system for estimating the length of an underwater organism. The system includes a memory and a processor. The processor is electrically connected to the memory to execute instructions stored in the memory to perform the method. The method includes receiving an underwater image, the underwater image being captured by an image capturing device and including an organism pattern of a target organism; identifying the organism pattern to obtain organism brightness data corresponding to the target organism; and calculating a distance between the target organism and the image capturing device based on the organism brightness data. In some embodiments, the identifying step further obtains organism structure data of the target organism, and the length of the target organism is calculated based on the distance and the organism structure data.

Description

Technical Field

[0001] This invention relates to a method and system for estimating the length of underwater organisms. Background Technology

[0002] In aquaculture, management procedures are frequently performed, such as assessing the growth status of aquatic organisms to facilitate their cultivation. Traditionally, this assessment is mostly done manually. For example, a few aquatic organisms are randomly caught from the pond and their size is measured to determine their growth status. However, this method of catching aquatic organisms and then assessing their growth status can easily harm them, and may even lead to their death.

[0003] Therefore, a method and system for predicting the length of underwater organisms are needed to solve the above problems. Summary of the Invention

[0004] The embodiments of the present invention propose an underwater organism length estimation method and system, which can obtain the length of aquatic organisms in the aquaculture pond without harvesting them, thereby allowing users to understand the growth status of the aquatic organisms.

[0005] According to an embodiment of the present invention, the above-described underwater organism length estimation method includes: receiving an underwater image, wherein the underwater image is captured using an image capturing device, and the underwater image contains a biological pattern of a target organism; identifying the biological pattern to obtain biological brightness data corresponding to the target organism; and calculating a biological distance between the target organism and the image capturing device based on the biological brightness data.

[0006] In some embodiments, the step of identifying the biopattern further obtains biological structure data of the target organism, and the above-mentioned biological length estimation method further includes: calculating the biological length of the target organism based on the biological distance and biological structure data.

[0007] According to another embodiment of the present invention, the above-described underwater organism length estimation method includes: a biological structure model establishment step and an online measurement step. The biological structure model establishment step includes: providing multiple training images, wherein each training image includes at least one training biological pattern and multiple training coordinate values ​​of multiple biological feature points of the at least one training biological pattern; and training a computer model using the training images and training coordinate values ​​to obtain a biological structure model, wherein the computer model is a neural network model, a mathematical model, or a statistical model. The online measurement steps include: receiving multiple underwater images of a breeding pond within a preset time period, wherein the underwater images are captured using an image capturing device, and these underwater images contain multiple biological patterns of multiple organisms; using a biological structure model to identify the biological patterns in the underwater images to obtain multiple biological structure data and multiple biological brightness data corresponding to these organisms; determining a reference biological pattern from the aforementioned biological patterns based on the biological structure data; calculating a biological distance between each organism and the image capturing device based on the biological brightness data of each biological pattern and the biological brightness data of the reference biological pattern; and calculating a biological length of each organism based on the biological distance between each organism and the image capturing device and the biological structure data of each organism.

[0008] In some embodiments, the step of using a biological structure model to identify biological patterns in underwater images includes: inputting each underwater image into a biological structure model to obtain multiple coordinate values ​​of feature points for each biological pattern.

[0009] In some embodiments, the step of calculating the biological distance between each organism and the image capturing device based on the biological brightness data of each biological pattern and the biological brightness data of a reference biological pattern includes: calculating a brightness difference between the biological brightness data of each biological pattern and the biological brightness data of the reference biological pattern; and calculating the biological distance between each organism and the image capturing device based on the brightness difference of each biological pattern and a reference distance of the aquaculture pond.

[0010] In some embodiments, the step of calculating the biological length of each organism based on the biological distance between each organism and the image capturing device and the biological structure data of each organism includes: determining the length of at least one part of each biological pattern based on the coordinate values ​​of feature points of each biological pattern; calculating a biological observation length of each organism based on the length of at least one part of each biological pattern; calculating a length adjustment factor based on the biological distance between each organism and the image capturing device; and calculating the biological length of each organism based on the biological observation length and the length adjustment factor.

[0011] In some embodiments, the organism is a shrimp, and the characteristic features include the shrimp's eyes, head, internal organs, body, and tail.

[0012] In some embodiments, the training images and underwater images are infrared images, and the image capturing device is an infrared camera.

[0013] In some embodiments, the preset time is one day, and the underwater images are obtained by sampling an underwater video at a sampling frequency of one image every 3 seconds.

[0014] In some embodiments, the above-described underwater organism length estimation method further includes: calculating an average length of the organism based on the length of each organism; and storing the average length of the organism in a database.

[0015] In some embodiments, underwater images are captured using only one image capturing device.

[0016] According to another embodiment of the present invention, the above-described underwater organism length estimation system includes a memory and a processor. The memory stores multiple instructions, and the processor is electrically connected to the memory to load the instructions and perform the following steps: receiving multiple underwater images of a breeding pond within a preset time period, wherein the underwater images are captured using an image capturing device, and the underwater images contain multiple biological patterns of multiple organisms; identifying the biological patterns in the underwater images using a biological structure model to obtain multiple biological structure data and multiple biological brightness data corresponding to the organisms; determining a reference biological pattern from the biological patterns based on the biological structure data; calculating a biological distance between each organism and the image capturing device based on the biological brightness data of each biological pattern and the biological brightness data of the reference biological pattern; and calculating a biological length of each organism based on the biological distance between each organism and the image capturing device and the biological structure data of each organism.

[0017] In some embodiments, when the processor uses a biological structure model to identify biological patterns in underwater images, the processor performs the following: inputting each underwater image into the biological structure model to obtain multiple coordinate values ​​of feature points for each biological pattern.

[0018] In some embodiments, when the processor calculates the biological distance between each organism and the image capturing device based on the biological brightness data of each biological pattern and the biological brightness data of a reference biological pattern, the processor performs the following: calculating a brightness difference between the biological brightness data of each biological pattern and the biological brightness data of the reference biological pattern; and calculating the biological distance between each organism and the image capturing device based on the brightness difference of each biological pattern and a reference distance of the aquaculture pond.

[0019] In some embodiments, when the processor calculates the biological length of each organism based on the biological distance between each organism and the image capturing device and the biological structure data of each organism, the processor performs the following: determining the length of at least one part of each biological pattern based on the coordinate values ​​of the feature points of each biological pattern; calculating a biological observation length of each organism based on the length of at least one part of each biological pattern; calculating a length adjustment factor based on the biological distance between each organism and the image capturing device; and calculating the biological length of each organism based on the biological observation length and the length adjustment factor.

[0020] In some embodiments, the organism is a shrimp, and the characteristic features include the shrimp's eyes, head, viscera, body, and tail.

[0021] In some embodiments, the training images and underwater images are infrared images, and the image capturing device is an infrared camera.

[0022] In some embodiments, the preset time is one day, and the underwater images are obtained by sampling an underwater video at a sampling frequency of one image every 3 seconds.

[0023] In some embodiments, after the processor loads the instructions, the processor further performs the following: calculates an average length of the organism based on the length of each organism; and stores the average length of the organism into a database.

[0024] In some embodiments, underwater images are captured using only one image capturing device.

[0025] To make the above features and advantages of the present invention more apparent and understandable, specific embodiments are described below in conjunction with the accompanying drawings. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating an underwater organism length estimation method according to an embodiment of the present invention;

[0027] Figure 2 This is a flowchart illustrating the steps for establishing a biological structural model according to an embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram illustrating training images according to an embodiment of the present invention;

[0029] Figure 4A This is a schematic diagram illustrating the predicted branch merging according to an embodiment of the present invention;

[0030] Figure 4B This is a schematic diagram illustrating a convolutional block according to an embodiment of the present invention;

[0031] Figure 5 This is a schematic diagram illustrating the characteristic points of a shrimp according to an embodiment of the present invention;

[0032] Figure 6 This is a schematic diagram illustrating an underwater organism length estimation system according to an embodiment of the present invention;

[0033] Figure 7 This is a flowchart illustrating the steps of online measurement according to an embodiment of the present invention;

[0034] Figure 8 This is a schematic diagram illustrating underwater images according to an embodiment of the present invention;

[0035] Figure 9 This is a schematic diagram illustrating the conversion of actual biological length according to an embodiment of the present invention.

[0036] [Symbol Explanation]

[0037] 1-8: Feature points

[0038] 100: Methods for estimating the length of underwater organisms

[0039] 110-120: Steps

[0040] 111-112: Steps

[0041] 121-125: Steps

[0042] 610: Memory

[0043] 620: Processor

[0044] M23, M78: Midpoint

[0045] P1: Interocular distance

[0046] P2: Tail section

[0047] P3: From the eyes to the internal organs

[0048] P4: Internal organs

[0049] P5: Head section

[0050] TI: Training Images Detailed Implementation

[0051] The following is a detailed description of the embodiments in conjunction with the accompanying drawings. However, the embodiments provided are not intended to limit the scope of the invention, and the description of the structural operation is not intended to limit the order of execution. Any structure resulting from the recombination of elements and producing a device with equivalent functionality is within the scope of the invention. Furthermore, the accompanying drawings are for illustrative purposes only and are not drawn to their original dimensions.

[0052] Please refer to Figure 1 This is a flowchart illustrating an underwater organism length estimation method 100 according to an embodiment of the present invention. In the underwater organism length estimation method 100, step 110 is first performed to establish a biological structure model of the underwater organism. Step 110 involves training a computer model using multiple training images of the underwater organism to obtain the aforementioned biological structure model. In this embodiment, the underwater organism is a shrimp, such as a white shrimp, but the embodiments of the present invention are not limited to this. Next, step 120 is performed to perform online measurement using the aforementioned biological structure model. In this embodiment, step 120 of online measurement involves capturing images of the underwater organism in a culture pond and using the aforementioned biological structure model to determine the size of the aquatic organism. Furthermore, the aforementioned computer model can be a neural network model, a mathematical model, or a statistical model.

[0053] To overcome the drawback of conventional techniques that require harvesting underwater organisms to assess their growth status, the underwater organism length estimation method 100 of this invention establishes a biological structure model of the underwater organisms and uses an underwater image acquisition device in the aquaculture pond to capture images of the underwater organisms in the pond, along with the aforementioned biological structure model, to determine the growth status of the underwater organisms in the aquaculture pond. This eliminates the need to harvest the underwater organisms in the pond, allowing the determination of their growth status (i.e., size). Steps 110 and 120 will be described in detail below.

[0054] Please refer to Figure 2 This is a flowchart illustrating step 110 of establishing a biological structural model according to an embodiment of the present invention. In step 110, step 111 is first performed to provide multiple training images TI, such as... Figure 3 As shown. In this embodiment, the training image TI is an infrared image, but the embodiments of the present invention are not limited to this.

[0055] Each training image TI contains at least one training biopattern and multiple training coordinate values ​​for multiple biometric points of the training biopattern. For example, Figure 3 The training biological pattern is a shrimp pattern, and this embodiment uses eight biometric points 1 to 8, where feature point 1 is the shrimp's head; feature points 2 and 3 are the shrimp's left and right eyes; feature point 4 is the shrimp's internal organs; feature point 5 is the shrimp's body; and feature points 6 to 8 are the shrimp's tail, where feature point 6 is the main body of the tail, and feature points 7 and 8 are the left / right tail forks. Thus, each training image TI contains at least eight biometric points of a shrimp and their coordinate values ​​(hereinafter referred to as training coordinate values).

[0056] In some embodiments, the number of biomarkers may be increased or decreased. For example, only biomarkers 2-3 from the shrimp's eyes and biomarker 4 from its internal organs may be used.

[0057] Following step 111, step 112 is performed to train a computer model using the aforementioned training image TI to obtain the aforementioned biological structure model. In this embodiment, the computer model is a neural network-like model, such as an openpose model, but the embodiments of the present invention are not limited to this. In some embodiments, to reduce the computational load of the openpose algorithm corresponding to the openpose model, the openpose algorithm is lightweighted. For example, branches are merged, and the refinement stage is modified into a convolutional block. For example, two prediction branches are merged into a single prediction branch because the first few convolutional layers of the openpose stage are mainly used to extract features, while the part confidence map (PCM) and the part affinity field (PAF) are related to the image features they focus on. Thus, the first few convolutional layers are merged, and the last two convolutional layers generate PCM and PAF respectively to achieve branch merging, as shown below. Figure 4A As shown. For example, the numerous 7x7 convolutional layers used in the refinement stage can be modified into convolutional blocks, such as... Figure 4B As shown, this reduces the amount of computation while maintaining a 7x7 field of view.

[0058] Furthermore, considering the architecture of Convolutional Neural Networks (CNNs) in the OpenPose model, in order to effectively fuse deep and shallow features, some embodiments employ a Feature Pyramid Network architecture. This allows shallow features to be added to deep features using smaller convolutions (i.e., convolution 1x1), and during upsampling, each feature tensor is input into the subsequent prediction network. In this way, low-level, high-semantic-information high-level features and high-level, low-semantic-information low-level features are self-preserved, ensuring that features at all sizes possess rich semantic information, enabling the network to achieve high accuracy in detecting large, medium, and small objects simultaneously.

[0059] In some embodiments, YOLOv3 can be applied to assist in object recognition and / or the attention mechanism of Self-Attention Generative Adversarial Network (SG-GAN) can be used to find the correlation between all similar features on the image features, so that the network training does not just see local images.

[0060] It is worth mentioning that, because step 112 of this embodiment uses multiple structural feature points of an organism (e.g., a shrimp) to establish a biological structure model, even if the biological pattern in the training image does not correspond to a complete organism, step 112 can still use this incomplete biological pattern for model training. Thus, the biological structure model established in step 112 of this embodiment can also identify incomplete biological patterns in the image and obtain, for example, the length of the biological portion (e.g., the shrimp portion) in the image. In other words, the biological structure model of this embodiment can identify individual parts of an organism.

[0061] Specifically, if a shrimp only shows its head and body in an image, the biological structure model of this embodiment can still identify the shrimp's head and body feature points in the image, and thus obtain the distance between the shrimp's head and body. In this embodiment, the shrimp portion includes: the distance between its eyes, its tail, the area from its eyes to its internal organs, its internal organs, and its head, such as... Figure 5 As shown. Figure 5 This is an illustration of various shrimp portions according to an embodiment of the present invention, wherein Figure 5 Therefore Figure 3 The shrimp structure formed by feature points 1 to 8 is used as an example to illustrate this. For example... Figure 5 As shown, the distance between the eyes, P1, is the distance between feature point 2 and feature point 3; the distance between the tail, P2, is the distance between feature point 6 and the midpoint M78 of feature point 7 and feature point 8; the distance between the eyes and internal organs, P3, is the distance between feature point 4 and the midpoint M23 of feature point 2 and feature point 3; the distance between the internal organs, P4, is the distance between feature point 4 and feature point 5; and the distance between the head, P5, is the distance between feature point 1 and feature point 4.

[0062] After establishing the biological structure model of the underwater organism in step 112, the aforementioned online measurement step 120 is then performed to capture images of the underwater organism in the aquaculture pond, and the size of the aquatic organism is determined using the aforementioned biological structure model.

[0063] Please refer to the following at the same time Figure 6 and Figure 7 , Figure 6 This is a schematic diagram illustrating an underwater organism length estimation system 600 according to an embodiment of the present invention. Figure 7 This is a flowchart illustrating step 120 of online measurement according to an embodiment of the present invention. The underwater organism length estimation system 600 includes a memory 610 and a processor 620. The memory 610 is used to store multiple instructions, and the processor 620 is electrically connected to the memory 610 to load the instructions in the memory 610 to perform the aforementioned online measurement step 120. In some embodiments, after the processor 620 loads the instructions in the memory 610, it can perform the aforementioned biological structure model building step 110.

[0064] In step 120 of the online measurement, step 121 is first performed to receive multiple underwater images 800 of the aquaculture pond within a preset time period, such as... Figure 8 As shown. These underwater images are captured using an image capturing device located underwater in the aquaculture pond. In this embodiment, the aforementioned preset time is 24 hours (i.e., the images captured each day), the image capturing device is an infrared camera, and the underwater images are provided by only this one image capturing device. In some embodiments, multiple image capturing devices can be installed in the aquaculture pond to capture underwater images.

[0065] To facilitate observation of organisms in the aquaculture pond, in this embodiment, the image capturing device is mounted on a feeding tray to observe and capture images of the organisms on the tray. Thus, the feeding tray serves as the background for the underwater image in this embodiment. In some embodiments, the image capturing device observes and captures images of organisms on the bottom net of the aquaculture pond. Furthermore, the image capturing device in this embodiment samples underwater video at a frequency of one frame every 3 seconds, but the embodiments of the present invention are not limited to this.

[0066] Each underwater image contains multiple biological patterns of multiple organisms. In this embodiment, each underwater image contains shrimp patterns corresponding to multiple shrimp. The underwater image data includes the brightness values ​​of pixels. In this embodiment, the brightness values ​​can range from a minimum of 0 to a maximum of 255. However, the embodiments of the present invention are not limited thereto.

[0067] Then, step 122 is performed to identify biological patterns in underwater images using the aforementioned biological structure model, thereby obtaining multiple biological structure data and multiple biological brightness data corresponding to the organisms in the underwater images. As previously described, the biological structure model of this embodiment can identify the biological structure data of the organisms corresponding to biological patterns in underwater images and obtain the coordinate values ​​of the feature points of the biological patterns. In some embodiments, the results of the aforementioned biological structure model in identifying biological patterns in underwater images include complete biological structures (e.g., a whole shrimp) and / or partial biological structures (e.g., the aforementioned shrimp portion).

[0068] Thus, step 122 can obtain the corresponding brightness data from the coordinate values ​​of the feature points corresponding to the underwater image based on the biological structure data. In this embodiment, the brightness data corresponding to the organism includes the brightness values ​​of its structural feature points. For example, in this embodiment, when a shrimp (or part of a shrimp) is identified in the underwater image, its corresponding biological structure data will include at least two of the aforementioned feature points 1 to 8, and its corresponding biological brightness data will include the brightness value of one of the aforementioned feature points 1 to 8. In this embodiment, the biological brightness data is feature point 4, i.e., the shrimp's viscera, but the embodiments of the present invention are not limited to this. In some embodiments, the biological brightness data may be feature point 2 or feature point 3, i.e., the shrimp's left eye or right eye.

[0069] Then, step 123 is performed to determine a reference biological pattern from the biological patterns in the underwater image based on the aforementioned biological structure data. In embodiments of the present invention, in order to calculate the conversion ratio between the pixels of the underwater image and the actual length of the organism, a reference organism needs to be selected as the reference organism for conversion. This reference organism can be considered as the organism furthest from the lens of the image capturing device. In embodiments of the present invention, this reference organism (reference biological pattern) is determined based on a brightness threshold. In this embodiment, the brightness threshold is 10%, so the reference organism is a shrimp whose brightness is the 10% darkest during the aforementioned preset time (i.e., 24 hours). For example, if the underwater image in 24 hours contains a total of 100 complete shrimp patterns, and these shrimp patterns correspond to 100 different brightness values ​​from 1 to 100, then the shrimp pattern with a brightness value of 10 is the reference biological pattern. In some embodiments, the brightness threshold can be changed as needed, for example, changed to 15%.

[0070] In addition, in underwater images, reference biological patterns with brightness below the brightness threshold are ignored and will no longer be treated as organisms for length estimation.

[0071] Next, step 124 is performed to calculate a bio-distance between each organism and the image capturing device based on the bio-brightness data of each bio-pattern and the bio-brightness data of a reference bio-pattern. In this embodiment, the bio-distance d is calculated as follows:

[0072] d=D-(R*C) (1)

[0073] Where D is the distance between the lens of the image capturing device and the feed tray; R is the difference between the biological brightness data (brightness value) of the target biological pattern being processed and the biological brightness data (brightness value) of the reference biological pattern and the background brightness ratio; C is an environmental coefficient, and in this embodiment, the value of the environmental coefficient C is 5.

[0074] In this embodiment, the difference between the bioluminance value of the target bio-pattern and the bioluminance value of the reference bio-pattern can be calculated first, and then this difference can be divided by the luminance value of the feed tray to obtain the ratio R. In some embodiments, the ratio of the bioluminance value of the target bio-pattern to the luminance value of the feed tray and the ratio of the bioluminance value of the reference bio-pattern to the luminance value of the feed tray can be calculated first, and then the difference between the two ratios can be calculated to obtain the ratio R.

[0075] Then, step 125 is performed to calculate the biological length of each organism based on the biological distance between each organism and the image capturing device, as well as the biological structure data of each organism. In step 125, the length of a biological pattern is first determined based on the coordinate values ​​of the feature points of each biological pattern. Then, the biological observation length of each organism is calculated based on the length of the portion of each biological pattern, where the biological observation length is the length of the biological pattern in the underwater image, and its unit is pixels. In the aforementioned step 122, if the identified biological structure data is an incomplete biological pattern (e.g., a biological pattern with only one shrimp portion), step 125 multiplies the length of this shrimp portion by the corresponding portion ratio to calculate the biological observation length. Conversely, if the identified biological structure data is a complete biological pattern (e.g., a complete shrimp pattern), step 125 selects one portion from the multiple biological portions of this complete biological pattern and multiplies it by the corresponding portion ratio to calculate the biological observation length.

[0076] In this embodiment, the ratio of the portion corresponding to the eyes to the internal organs is 3.95; the ratio of the portion corresponding to the distance between the eyes is 8.42; the ratio of the portion corresponding to the tail is 6.61; the ratio of the portion corresponding to the internal organs is 5.17; and the ratio of the portion corresponding to the head is 2.49.

[0077] Then, a length adjustment factor is calculated based on the biological distance between each organism and the image capturing device. In embodiments of the present invention, since the biological distance to the capturing device affects the length of the organism's bio-pattern in the image, embodiments of the present invention provide a length adjustment factor for corresponding length adjustment. In the aforementioned step 124, the biological distance d between each organism and the image capturing device has been calculated, and step 125 divides the biological distance d of each organism by the distance D between the lens of the image capturing device and the feed tray to obtain the length adjustment factor T.

[0078] Next, the actual biological length of each organism is calculated based on its observed biological length and length adjustment factor. In embodiments of the present invention, such as... Figure 9As shown, by first measuring the shooting angle θ of the underwater camera and the distance D between the image capturing device and the feeding tray, the length and width of each pixel in the image captured by the image capturing device at distance D can be calculated in centimeters. Here, L is the actual length of the biological part (e.g., in centimeters), and FP is the actual length of the feeding tray (e.g., in centimeters). In this way, the length of the biological part (e.g., the shrimp parts mentioned above) in the underwater image can be converted into the corresponding length in centimeters at distance D. Combining this with the biological distance d mentioned above, we use the principle of similar triangles to convert objects at different distances. The conversion formula is as follows:

[0079]

[0080] Where Pw is the width of the bio-part at a distance D (e.g., in centimeters), and Ph is the height of the bio-part at a distance D (e.g., in centimeters).

[0081] In some embodiments, if the identified biological structure data is a complete biological pattern (e.g., a complete shrimp pattern), step 125 may not calculate the length of the organism by means of parts. For example, the actual length of each biological part can be calculated using formulas (1) and (2) above, and then summed to obtain the total length. In some embodiments, the length from feature point 1 (shrimp head) to midpoint M78 (e.g.) can be calculated. Figure 5 The actual length of the shrimp is obtained by measuring the distance shown in the figure.

[0082] In some embodiments, the average length of organisms can be calculated based on the biological lengths of all organisms in the underwater images, and this average length value can then be stored in a database. In this way, users can know the daily length of the organisms in the aquaculture pond, and thus determine the growth status of the organisms in the pond.

[0083] As described above, the underwater organism length estimation method 100 and underwater organism length estimation system 600 of this embodiment utilize an image capturing device located underwater in the aquaculture pond to capture images of the organisms in the pond, and analyze the images to calculate the length of the organisms. Thus, the growth status of the organisms in the aquaculture pond can be determined without harvesting them. Furthermore, the underwater organism length estimation method 100 and underwater organism length estimation system 600 of this embodiment also consider changes in the water quality of the aquaculture pond. Therefore, even in aquaculture ponds with relatively turbid water, the underwater organism length estimation method 100 and underwater organism length estimation system 600 of this embodiment can be applied to estimate and calculate the length of underwater organisms.

[0084] Although the present invention has been disclosed above by way of embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the scope defined in the appended claims.

Claims

1. A method for estimating the length of underwater organisms, characterized in that, Include: The steps for establishing a biological structural model include: Provide multiple training images, wherein each training image includes at least one training biological pattern and multiple training coordinate values ​​of multiple feature points of the at least one training biological pattern; as well as A computer model is trained using the multiple training images and the multiple training coordinate values ​​to obtain a biological structure model. The online measurement steps include: Receive multiple underwater images of a breeding pond within a preset time period, wherein the multiple underwater images are captured using an image capturing device, and the underwater images contain multiple biological patterns of multiple organisms. The biological structure model is used to identify the multiple biological patterns in the multiple underwater images to obtain multiple biological structure data and multiple biological brightness data corresponding to the multiple organisms. A reference biological pattern is determined from the multiple biological patterns based on the multiple biological structure data; A biological distance between each of the said biological patterns and the image capturing device is calculated based on the biological brightness data of each of the said biological patterns and the biological brightness data of the reference biological pattern. as well as The biological length of each organism is calculated based on the biological distance between each organism and the image capturing device and the biological structure data of each organism. The multiple organisms mentioned are shrimp, and the multiple feature points include the shrimp's eyes, head, internal organs, body, and tail. The step of using this biological structure model to identify the multiple biological patterns in the multiple underwater images includes: Each of the underwater images is input into the biological structure model to obtain the coordinate values ​​of the plurality of feature points of each biological pattern; The step of calculating the biological distance between each biological entity and the image capturing device based on the biological brightness data of each biological pattern and the biological brightness data of the reference biological pattern includes: Calculate a brightness difference between the bioluminance data of each of the bio-patterns and the bioluminance data of the reference bio-pattern; and The distance between each organism and the image capturing device is calculated based on the brightness difference of each organism pattern and the distance to a reference object in the aquaculture pond.

2. The underwater organism length estimation method according to claim 1, characterized in that, The step of calculating the length of each organism based on the distance between each organism and the image capturing device and the biological structure data of each organism includes: The length of at least one part of each of the biological patterns is determined based on the coordinate values ​​of the plurality of feature points of each of the biological patterns; A bioobservation length for each of the organisms is calculated based on the length of at least one part of each of the biopatterns; A length adjustment factor is calculated based on the distance between each organism and the image capturing device; as well as The length of each organism is calculated based on its observed length and the length adjustment factor.

3. The method for estimating the length of underwater organisms according to claim 1, characterized in that, The multiple training images and the multiple underwater images are infrared images, and the image acquisition device is an infrared camera.

4. The method for estimating the length of underwater organisms according to claim 1, characterized in that, The preset time is one day, and the multiple underwater images are obtained by sampling an underwater video. The sampling frequency of the underwater video is one image every 3 seconds.

5. The method for estimating the length of underwater organisms according to claim 1, characterized in that, Also includes: Calculate an average length of the plurality of organisms based on the length of each of the said organisms; and The average length of the multiple organisms is stored in a database.

6. The method for estimating the length of underwater organisms according to claim 1, characterized in that, The multiple underwater images were captured using only one image capturing device.

7. A system for predicting the length of underwater organisms, characterized in that, Include: A memory module used to store multiple instructions; and A processor, electrically connected to the memory, loads the plurality of instructions to perform the following: Receive multiple underwater images of a breeding pond within a preset time period, wherein the multiple underwater images are captured using an image capturing device, and the underwater images contain multiple biological patterns of multiple organisms. A biological structure model is used to identify multiple biological patterns in the multiple underwater images to obtain multiple biological structure data and multiple biological brightness data corresponding to the multiple organisms. A reference biological pattern is determined from the multiple biological patterns based on the multiple biological structure data; A biological distance between each of the said biological patterns and the image capturing device is calculated based on the biological brightness data of each of the said biological patterns and the biological brightness data of the reference biological pattern. as well as The biological length of each organism is calculated based on the biological distance between each organism and the image capturing device and the biological structure data of each organism. When the processor uses the biological structure model to identify the multiple biological patterns in the multiple underwater images, the processor performs the following steps: inputting each underwater image into the biological structure model to obtain the coordinate values ​​of multiple feature points of each biological pattern. Among them, the multiple organisms are shrimp, and the multiple feature points include the shrimp's eyes, head, internal organs, body, and tail. Specifically, when the processor calculates the biological distance between each biological entity and the image capturing device based on the biological brightness data of each biological pattern and the biological brightness data of the reference biological pattern, the processor performs the following: Calculate a brightness difference between the bioluminance data of each of the bio-patterns and the bioluminance data of the reference bio-pattern; and The distance between each organism and the image capturing device is calculated based on the brightness difference of each organism pattern and the distance to a reference object in the aquaculture pond.

8. The underwater organism length estimation system according to claim 7, characterized in that, When the processor calculates the length of each organism based on the distance between each organism and the image capturing device and the biological structure data of each organism, the processor performs: The length of at least one part of each of the biological patterns is determined based on the coordinate values ​​of the plurality of feature points of each of the biological patterns; A bioobservation length for each of the organisms is calculated based on the length of at least one part of each of the biopatterns; A length adjustment factor is calculated based on the distance between each organism and the image capturing device; as well as The length of each organism is calculated based on its observed length and the length adjustment factor.

9. The underwater organism length estimation system according to claim 7, characterized in that, The underwater images are infrared images, and the image acquisition device is an infrared camera.

10. The underwater organism length estimation system according to claim 7, characterized in that, The preset time is one day, and the multiple underwater images are obtained by sampling an underwater video. The sampling frequency of the underwater video is one image every 3 seconds.

11. The underwater organism length estimation system according to claim 7, characterized in that, After the processor loads the aforementioned instructions, the processor further performs: Calculate an average length of the plurality of organisms based on the length of each of the said organisms; and The average length of the multiple organisms is stored in a database.

12. The underwater organism length estimation system according to claim 7, characterized in that, The multiple underwater images were captured using only one image capturing device.

Images