CN-121999353-A - Intelligent monitoring method, system and program product for aquaculture

CN121999353ACN 121999353 ACN121999353 ACN 121999353ACN-121999353-A

Abstract

The invention belongs to the technical field of aquaculture, and particularly discloses an intelligent aquaculture monitoring method, an intelligent aquaculture monitoring system and a program product. The method has the advantages that complicated depth measurement is simplified into proportion operation of a two-dimensional plane through high-robustness non-contact measurement, physical damage to fish bodies is avoided, meanwhile, efficient and accurate measurement of the fish bodies can be achieved, user intention is recognized through a cognitive man-machine interaction mode, a user can directly control bottom hardware logic through natural language instructions, complexity of system control and cultivation threshold combination are reduced, intelligent aquaculture can be achieved through combination of growth trend judgment, and fish cultivation progress and fish monomer economic value can be effectively mined through multi-mode data fusion, so that the user can intuitively and accurately grasp aquaculture conditions.

Inventors

LI ZHONGXIN
CAI JIA
HUANG JUNJIE
YU DAPENG

Assignees

深圳市天衍智擎科技有限公司
广东海洋大学深圳研究院

Dates

Publication Date: 20260508
Application Date: 20260205

Claims (10)

1. An intelligent monitoring method for aquaculture is characterized by comprising the following steps: each fish body monitoring image is collected, the fish body monitoring image comprises a fish body area and two diffuse reflection light spots projected on the fish body area, and the two diffuse reflection light spots are generated by projecting two parallel collimated laser beams with fixed physical distance to the fish body area, wherein the two parallel collimated laser beams are emitted by a laser emitter; Performing laser feature extraction on each fish body monitoring image to obtain centroid coordinates of two diffuse reflection light spots in each fish body monitoring image, and calculating a depth scale of each fish body monitoring image based on the centroid coordinates of the two diffuse reflection light spots in each fish body monitoring image and the physical distance between two parallel collimated laser beams of the laser transmitters; performing target detection on each fish body monitoring image to obtain a fish body area boundary frame of each fish body monitoring image, and determining the length of a pixel main shaft of the fish body area boundary frame in each fish body monitoring image; calculating the physical body length of each fish body monitoring image by using the depth scale of each fish body monitoring image and the main axis length of the pixels of the boundary frame of the fish body area; determining fish seed growth parameters according to the current fish seed monitoring mode, and calculating the fish body weight of each fish body monitoring image by utilizing the fish seed growth parameters and the fish body physical body length of each fish body monitoring image; Determining the average weight of the current fish school based on the weight of the fish in each fish monitoring image, and judging whether the growth trend of the fish school is slow or not according to the average weight of the current fish school; When the growth trend of the fish shoal is determined to be slow, the feeding amount or the feeding frequency is determined according to the average weight of the current fish shoal, a feeding control strategy is generated according to the feeding amount or the feeding frequency, and the feeding control strategy is sent to the intelligent feeding device.
2. An aquaculture intelligent monitoring method according to claim 1, characterized in that said method further comprises: Acquiring a natural language instruction of a user, and performing natural language processing on the natural language instruction of the user based on a large language model to acquire monitoring instruction content; When the monitoring instruction content comprises a fingerling switching monitoring mode instruction, determining a fingerling monitoring mode after switching according to the fingerling switching monitoring mode instruction, and updating fingerling growth parameters according to the fingerling monitoring mode after switching; When the monitoring instruction content contains a growth parameter correction instruction, determining a corrected fingerling growth parameter according to the growth parameter correction instruction, and replacing the current fingerling growth parameter by using the corrected fingerling growth parameter.
3. An aquaculture intelligent monitoring method according to claim 2, wherein when the monitoring instruction content comprises an optimized feeding instruction, the method further comprises: Determining the average weight of the current fish school based on the fish body weight of each fish body monitoring image; directly determining supplementary feeding amount or feeding frequency according to the average weight of the current fish school, and generating a feeding control strategy according to the supplementary feeding amount or the feeding frequency; and sending the feeding control strategy to the intelligent feeding device.
4. An aquaculture intelligent monitoring method according to claim 2, wherein when the monitoring instruction content contains an economic value assessment instruction, the method further comprises: inquiring market unit price of fingerlings corresponding to the current fingerling monitoring mode according to the economic value evaluation instruction; And calculating the monomer body prices of the fingerlings according to the average weight of the current fingerlings and the market price of the fingerlings corresponding to the current fingerlings monitoring mode, and outputting the monomer body prices of the fingerlings.
5. The method of claim 1, wherein the calculating the depth scale of each fish-body monitor image based on the centroid coordinates of two diffuse reflection light spots in each fish-body monitor image and the physical distance between two parallel collimated laser beams of the laser emitter comprises: Substituting the barycenter coordinates of two diffuse reflection light spots in the fish body monitoring image and the physical distance between two parallel collimated laser beams of the laser transmitter into a depth scale formula to calculate, so as to obtain a depth scale of the fish body monitoring image, wherein the depth scale formula is as follows: Wherein λ is a depth scale, D real is a physical distance between two parallel collimated laser beams of the laser emitter, and centroid coordinates of two diffuse reflection light spots in the fish body monitoring image are (x 1 ,y 1 ) and (x 2 ,y 2 ) respectively.
6. The method of claim 5, wherein calculating the physical body length of each fish body of the fish body monitoring image using the depth scale of each fish body monitoring image and the main axis length of the pixels of the boundary box of the fish body area comprises: Substituting the depth scale of the fish body monitoring image and the main axis length of the pixel of the boundary frame of the fish body area into a fish body physical length calculation formula to calculate so as to obtain the fish body physical length of the fish body monitoring image, wherein the fish body physical length calculation formula is as follows: Wherein, L real is the physical body length of the fish body, and L pix is the main axis length of the pixel of the boundary frame of the fish body region.
7. The intelligent monitoring method according to claim 6, wherein the fish growth parameters include a condition factor and a different-speed growth index, the calculating the fish weight of each fish monitoring image using the fish growth parameters and the fish physical body length of each fish monitoring image includes: Substituting the physical body length of the fish body monitoring image and the fish seed growth parameter into a preset fish body weight calculation formula to calculate so as to obtain the fish body weight of the fish body monitoring image, wherein the fish body weight calculation formula is as follows: Wherein W is fish body weight, alpha is a conditional factor, and beta is an abnormal growth index.
8. The utility model provides an aquaculture intelligent monitoring system which characterized in that, includes laser emitter, camera, central controller, intelligent feeding device and user terminal, wherein: the laser emitter is used for emitting two parallel collimated laser beams with fixed physical distance to be projected onto the fish body area; the camera is used for collecting each fish body monitoring image and transmitting each fish body monitoring image to the central controller, wherein the fish body monitoring image comprises a fish body area and two diffuse reflection light spots projected on the fish body area; The central controller is used for extracting laser characteristics of each fish body monitoring image to obtain barycenter coordinates of two diffuse reflection light spots in each fish body monitoring image, calculating depth scale of each fish body monitoring image based on barycenter coordinates of two diffuse reflection light spots in each fish body monitoring image and physical distance between two parallel collimation laser beams of a laser emitter, carrying out target detection on each fish body monitoring image to obtain a fish body area boundary frame of each fish body monitoring image, determining the length of a pixel main shaft of the fish body area boundary frame in each fish body monitoring image, calculating the physical body length of each fish body monitoring image by utilizing the depth scale of each fish body monitoring image and the pixel main shaft length of each fish body area boundary frame, determining fish body growth parameters according to a current fish species monitoring mode, calculating the fish body weight of each fish body monitoring image by utilizing the fish species growth parameters and the fish body physical body length of each fish body monitoring image, determining the average weight of each fish body based on the fish body weight of each fish body monitoring image, judging whether the fish body growth trend is slow or not on average, when the fish body growth trend is judged, determining a current feed trend, or a feed rate control strategy is switched to be used for a natural language control mode, and a natural language control command is generated based on a natural language control command, and a natural language control command is obtained after the mode is switched to a natural language control command is obtained, updating fish seed growth parameters according to the switched fish seed monitoring mode; when the monitoring instruction content contains a growth parameter correction instruction, determining a corrected fingerling growth parameter according to the growth parameter correction instruction, and replacing the current fingerling growth parameter by using the corrected fingerling growth parameter; when the monitoring instruction content contains an economic value assessment instruction, inquiring market price of the fingerlings corresponding to the current fingerling monitoring mode according to the economic value assessment instruction, calculating monomer body prices of the fingerlings according to the average weight of the current fingerlings and the market price of the fingerlings corresponding to the current fingerling monitoring mode, and sending the monomer body prices of the fingerlings to a user terminal; the intelligent feeding device is used for executing a feeding control strategy sent by the central controller to feed the shoal of fish; the user terminal is used for displaying the fish body monitoring image, the current fish species monitoring mode, the current average weight of the fish shoal and/or the individual body price of the fish species sent by the central controller and sending natural language instructions of the user to the central controller.
9. An aquaculture intelligent monitoring system, characterized by comprising: A memory for storing instructions; A processor for reading the instructions stored in the memory and executing the intelligent aquaculture monitoring method according to the instructions.
10. A computer program product, characterized in that the method for intelligent monitoring of aquaculture according to any of claims 1-7 is performed when said computer program product is run on a computer.

Description

Intelligent monitoring method, system and program product for aquaculture Technical Field The invention belongs to the technical field of aquaculture, and particularly relates to an intelligent aquaculture monitoring method, an intelligent aquaculture monitoring system and a program product. Background In the traditional aquaculture process, monitoring of fish growth indexes (body length and weight) mainly depends on manual sampling and fishing, namely, aquaculture personnel regularly weigh and measure the fish fishing water surface through a net tool, and then perform adaptive bait throwing control according to the observation condition of sampling and fishing. The method has the advantages of limited sampling times, very labor consumption, dependence on manual experience judgment, and insufficient fine and accurate control of the growth trend of the fish shoals. In recent years, an underwater monitoring technology based on computer vision appears, but most of the underwater monitoring technology adopts a common monocular camera, and the monocular camera cannot accurately calculate the actual physical size and weight of a fish body due to the perspective projection problem, so that the accuracy of fish swarm growth trend judgment and the adaptation degree of bait feeding control are affected. Disclosure of Invention The invention aims to provide an aquaculture intelligent monitoring method, system and program product, which are used for solving the problems in the prior art. In order to achieve the above purpose, the present invention adopts the following technical scheme: in a first aspect, an aquaculture intelligent monitoring method is provided, comprising: each fish body monitoring image is collected, the fish body monitoring image comprises a fish body area and two diffuse reflection light spots projected on the fish body area, and the two diffuse reflection light spots are generated by projecting two parallel collimated laser beams with fixed physical distance to the fish body area, wherein the two parallel collimated laser beams are emitted by a laser emitter; Performing laser feature extraction on each fish body monitoring image to obtain centroid coordinates of two diffuse reflection light spots in each fish body monitoring image, and calculating a depth scale of each fish body monitoring image based on the centroid coordinates of the two diffuse reflection light spots in each fish body monitoring image and the physical distance between two parallel collimated laser beams of the laser transmitters; performing target detection on each fish body monitoring image to obtain a fish body area boundary frame of each fish body monitoring image, and determining the length of a pixel main shaft of the fish body area boundary frame in each fish body monitoring image; calculating the physical body length of each fish body monitoring image by using the depth scale of each fish body monitoring image and the main axis length of the pixels of the boundary frame of the fish body area; determining fish seed growth parameters according to the current fish seed monitoring mode, and calculating the fish body weight of each fish body monitoring image by utilizing the fish seed growth parameters and the fish body physical body length of each fish body monitoring image; Determining the average weight of the current fish school based on the weight of the fish in each fish monitoring image, and judging whether the growth trend of the fish school is slow or not according to the average weight of the current fish school; When the growth trend of the fish shoal is determined to be slow, the feeding amount or the feeding frequency is determined according to the average weight of the current fish shoal, a feeding control strategy is generated according to the feeding amount or the feeding frequency, and the feeding control strategy is sent to the intelligent feeding device. In one possible design, the method further comprises: Acquiring a natural language instruction of a user, and performing natural language processing on the natural language instruction of the user based on a large language model to acquire monitoring instruction content; When the monitoring instruction content comprises a fingerling switching monitoring mode instruction, determining a fingerling monitoring mode after switching according to the fingerling switching monitoring mode instruction, and updating fingerling growth parameters according to the fingerling monitoring mode after switching; When the monitoring instruction content contains a growth parameter correction instruction, determining a corrected fingerling growth parameter according to the growth parameter correction instruction, and replacing the current fingerling growth parameter by using the corrected fingerling growth parameter. In one possible design, when the monitoring instruction content includes an optimized feeding instruction, the method further includes: Determining the average weigh