CN211048177U

CN211048177U - Intelligent breeding system

Info

Publication number: CN211048177U
Application number: CN201921293439.2U
Authority: CN
Inventors: 黄英哲; 洪庆章; 张雲南
Original assignee: National Sun Yat Sen University
Current assignee: Sun Yat Sen University; National Sun Yat Sen University
Priority date: 2019-08-12
Filing date: 2019-08-12
Publication date: 2020-07-21
Anticipated expiration: 2029-08-12

Abstract

The utility model provides an intelligent culture system, including breed pond, feeding machine, camera and controller. The camera is arranged in the culture pond and used for capturing underwater images. The controller calculates the residual amount of the feed according to the underwater image and controls the feeding machine to feed the feed into the culture pond according to the residual amount of the feed. Therefore, the waste of the feed can be reduced.

Description

Intelligent breeding system

Technical Field

The utility model relates to an intelligent breeding system, which can automatically feed.

Background

In the current aquaculture technology, although there is a dedicated feeding device, it has only simple functions of starting, closing, and controlling flow rate, and the conventional feeding device still has to rely on personnel to start feeding, stop, or adjust the flow rate of feed. The greatest challenge when feeding by human power is turbid pond water quality. The growth of the shrimps depends on the amount and time point of the fed feed, but the turbid water quality makes fishermen almost unable to judge the health condition of the shrimps by naked eyes, so that excessive feeding of shrimp groups is easily caused, excessive residual baits and excrement are left in the water, and the water quality is further deteriorated. In addition, poor water quality will proliferate bacteria, which in turn will cause the shrimp to be sick, resulting in a total pool being covered.

According to the survey, 76% of taiwan interviewees have planned to start or adopt more Artificial Intelligence (AI) solutions in the future, but 83% of interviewees do not know where to start. To date, most taiwan enterprises have not established an AI database, especially in the traditional industry, and most of them rely on the experience of technicians to make decisions on site. In order to make a transformation breakthrough in the traditional industry, important data hidden in enterprises and rich in value must be found, which depends on a stable information technology architecture as a foundation. Therefore, how to automatically feed using artificial intelligence is an issue of concern to those skilled in the art.

SUMMERY OF THE UTILITY MODEL

An embodiment of the utility model provides an intelligent farming systems, including breed pond, feeding machine, camera and controller. The camera is arranged in the culture pond and used for capturing underwater images. The controller is connected to the camera and the bait casting machine in a communication mode and used for obtaining underwater images. The controller calculates the residual amount of the feed according to the underwater image or transmits the underwater image to the server to calculate the residual amount of the feed by the server. The controller controls the bait casting machine to feed into the culture pond according to the residual amount of the feed.

In some embodiments, the intelligent farming system further comprises a stand disposed in the farming pond, the camera being disposed on the stand. The feeder deposits at least part of the feed into the support.

In some embodiments, the support comprises a mesh plane and a collection tray. The camera has a camera direction towards the mesh plane and the collection tray extends outwardly from the mesh plane.

In some embodiments, the frame further comprises a support portion disposed below the mesh plane, thereby spacing the mesh plane a distance from the bottom of the culture pond.

In some embodiments, the intelligent farming system further comprises a washer to spray water or air bubbles to the rack. The controller executes an image segmentation algorithm on the underwater image to obtain a support image area, and judges whether to start the cleaning machine according to the area of the support image area.

In some embodiments, the controller performs an image segmentation algorithm on the underwater image to obtain a feed image region, determines whether an area of the feed image region is smaller than a threshold value, and controls the feeding machine to feed the feed into the culture pond if the area of the feed image region is smaller than the threshold value.

In some embodiments, the camera end of the camera is planar or convex.

In the system, the feed waste can be reduced by automatically feeding.

In order to make the aforementioned and other features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.

Drawings

FIG. 1 is a schematic diagram illustrating a smart farming system according to an embodiment;

FIG. 2 is a schematic diagram illustrating a camera mounted on a stand according to one embodiment;

FIG. 3 is a schematic diagram illustrating an underwater image according to one embodiment.

[ notation ] to show

100: intelligent breeding system

110: culture pond

112: camera with a lens having a plurality of lenses

114: support frame

116: bait throwing machine

117: bait casting tube

118: lighting device

120: water quality sensor

122: controller

124: router

126: server

128: sewage discharging motor

130: user's hand

201: collecting tray

202: mesh plane

203: supporting part

210: feed stuff

220: direction of photography

230: cleaning machine

241 to 243: camera terminal

310: underwater image

320: feed image area

330: fouling material

Detailed Description

FIG. 1 is a schematic diagram illustrating a smart farming system, according to an embodiment. Referring to fig. 1, the intelligent farming system 100 includes a farming pond 110, a camera 112, a support 114, a bait casting machine 116, a bait casting pipe 117, a lighting device 118, a water quality sensor 120, a controller 122, a router 124, a server 126, and a sewage motor 128.

In this embodiment, the shape of the cultivation pond 110 is a circle in which shrimps are cultivated, but in other embodiments, the shape of the cultivation pond 110 may be a square or any other shape, and in other embodiments, fish, crabs or other aquatic creatures may be cultivated, which is not limited herein. The blowdown motor 128 is used to draw water from the bottom of the culture pond 110 for blowdown. The feeder 116 may include a motor or other suitable mechanical device, and the feeder 116 is configured to feed the feed through the feeder tube 117 into the support 114, although the feeder 116 may have other openings to feed the feed directly into the aquarium 110.

The camera 112 is disposed below the water surface of the culture pond 110 for continuously capturing underwater images. The camera 112 may include a Charge-coupled Device (CCD) sensor, a Complementary Metal-Oxide Semiconductor (Complementary Metal-Oxide Semiconductor) sensor, or other suitable light sensor, which may be a visible light sensor or a non-visible light sensor, but the invention is not limited thereto. In some embodiments, the camera 112 has an infrared sensor and a visible light sensor for capturing infrared images and visible light images, respectively. In some embodiments, the camera 112 may also include a plurality of visible light sensors, and a three-dimensional scene may be calculated from images captured by the sensors. The illumination device 118 may include one or more light emitting diodes, but the invention is not limited to the wavelength of these light emitting diodes. The router 124 may be a wired router or a wireless router.

The water quality sensor 120 is used to sense the dissolved oxygen, temperature or other values related to water quality. The controller 122 is communicatively connected to the water quality sensor 120, the camera 112, the bait casting machine 116 and the sewage motor 128, wherein the communication connection may be a wired connection or a wireless connection, such as through the internet, a local area network, a wide area network, a near field communication, an infrared communication, bluetooth, WiFi, etc., so that signals and data can be transmitted between the devices. The controller 122 may be a central processing unit, a microprocessor, a microcontroller, a digital signal processor, an image processing chip, an asic, or the like. The controller 122 may take the captured underwater images from the camera 112 and collect the relevant data from the water quality sensor 120, and these collected data may be transmitted to the server 126 through the router 124, and the user 130 may log into the server 126 to access these data.

Referring to fig. 2, fig. 2 is a schematic diagram illustrating a camera being mounted on a stand according to an embodiment. In some embodiments, the support 114 has a mesh plane 202, a collection pan 201, and a support 203, which may be formed of metal or other suitable material. The feed 210 is thrown onto the mesh plane 202, but since the water flow in the cultivation pond may be large, the feed 210 may drift out of the mesh plane 202, and therefore the collecting tray 201 is arranged to extend outward from the mesh plane 202 to catch the feed, the complete collecting tray 201 is not shown in fig. 2 for simplicity, and the spreading angle of the collecting tray 201 may be 360 degrees in some embodiments, so as to catch more feed. The camera 112 may be disposed on the upper portion of the stand 114, and the photographing direction 220 is toward the mesh plane 202, that is, the main content of the captured underwater image is the mesh plane 202 and the feed thereon. In addition, the bottom of the culture pond is usually turbid, and the supporting part 203 is disposed below the mesh plane 202, so that the mesh plane 202 is separated from the bottom of the culture pond by a certain distance, and thus the captured underwater images can be clear. However, fig. 2 is only an example, and those skilled in the art can design other shapes and sizes of the bracket 114 according to the above disclosure, and the invention is not limited thereto. In addition, the shape of the mesh plane 202 may be rectangular, other polygonal shapes, or any shape in other embodiments.

Fig. 2 also shows an enlarged view of the photographing end of the camera 112, wherein the photographing end 241 is a plane in one embodiment, and the photographing end 242 is a convex surface in another embodiment. The photographing ends 241, 242 may be provided with a lens or a protective cover plate, but the present invention is not limited thereto. Compared to the image capturing end 243 of the prior art, the image capturing end 243 has a retaining wall, so that bubbles are formed, which affects the captured image. In contrast, no air bubbles are formed on the photographing ends 241, 242.

Referring to fig. 1 and 2, after the underwater image is obtained, the remaining amount of the feed can be calculated according to the underwater image, and the feeding machine 116 can be controlled to feed the cultivation pond 110 according to the remaining amount of the feed. For example, when the amount of the remaining feed is greater than a threshold value, the controller 122 does not control the feeding machine 116 to feed, and when the amount of the remaining feed is less than the threshold value, the controller 122 controls the feeding machine 116 to feed. Therefore, the feed can be put in when needed, and the waste of the feed is reduced. The above calculation of the remaining amount of the fodder can be performed by the controller 122, or the controller 122 can also transmit the underwater image to the server 126, and the server 126 calculates the remaining amount of the fodder in a cloud manner and then transmits the result back to the controller 122. The following will describe how to calculate the remaining amount of the feed by way of example.

In this embodiment, the remaining amount of the feedstuff is calculated by machine learning, that is, the underwater image is inputted to a machine learning model, and the machine learning model outputs a number representing the remaining amount of the feedstuff. The machine learning model may be a convolutional neural network, a support vector machine, or other suitable model, which is not limited herein. In some embodiments, the machine learning model is used to perform image segmentation (image segmentation) to cut out the feed image region in the underwater image, for example, referring to fig. 3, the feed image region 320 can be cut out from the underwater image 310. In some embodiments, the underwater image 310 is an infrared image, which can clearly identify the feedstuff, but in other embodiments, the underwater image 310 may also be a visible light image, which is not limited herein. Taking a convolutional neural network as an example, in a training stage, the captured underwater image is input into the convolutional neural network, and the output of the convolutional neural network is an artificially cut binary image (which is used for indicating that the underwater image is feed and the underwater image is not feed); in the testing stage, the underwater image shot in real time can be input into the trained convolutional neural network, and the convolutional neural network can output the cut binary image. The area of the feed image area 320 is proportional to the remaining amount of the feed, so that when the area of the feed image area 320 is smaller than a threshold value, it indicates that the feed is not enough, and the feeding machine 116 can be controlled to feed the cultivation pond 110. In other embodiments, the weight of the feed may be measured in practice and the machine learning model trained on this weight, so that the output at the time of the testing phase will be the predicted feed weight. In other embodiments, the machine learning model may also be trained according to the volume, quantity, etc. of the fodder, that is, the fodder remaining amount may be the weight, volume, quantity, or area of the fodder image area 320, which is not limited herein.

Referring to fig. 2 and 3, in some embodiments, a cleaning machine 230 is further provided, the cleaning machine 230 is used for spraying water or bubbles to clean the rack 114, and the cleaning machine 230 may include a motor, a compressor or other suitable mechanical devices. The timing of cleaning may also be determined by machine learning, specifically, an image segmentation algorithm may be performed on the underwater image captured by the camera 112 to obtain a stent image region, which refers to a region belonging to the stent 114 in the image, and the region belonging to the stent 114 may be marked by a person during the training phase to be input to the convolutional neural network. When the rack 114 has the dirt 330, the dirt 330 covers the mesh plane, so that the area of the rack image area is reduced, and when the area of the rack image area is smaller than a threshold value, it indicates that there is much dirt on the rack 114, so the controller 122 activates the cleaning machine 230 to clean the rack 114.

In some embodiments, to accommodate a variety of different culture ponds, the mesh-like planar portions in the underwater imagery may be combined with other culture pond backgrounds to create more images for training, whereby a more robust convolutional neural network may be trained. For example, a mesh-shaped plane portion can be manually cut out and attached to the background image of another culture pond, and the boundary between the two can be edited by the poisson equation (poisson equation), so that a plurality of training underwater images can be additionally generated.

In some embodiments, the controller 122 or the server 126 may also perform a defogging algorithm on the underwater image, wherein the underwater image is a visible light image. In particular, the defogging algorithm may also be performed by a convolutional neural network, which has an input of a foggy underwater image and an output of a clear underwater image during the training phase. Clear underwater images may be obtained through defogging algorithms, and in some embodiments an object detection algorithm may be performed on the clear underwater images to detect underwater animals/plants. In some embodiments, the demisted underwater images are transmitted to the server 126 for storage.

In the intelligent culture system, the bait feeding amount is controlled by using an image processing technology, so that the waste of feed can be reduced. In addition, because the camera is erected on the bracket, most of the underwater images shot relate to the mesh plane and the feed, so that the system can be suitable for different culture ponds and is less susceptible to the background change of the culture ponds.

Although the present invention has been described with reference to the above embodiments, it is not intended to limit the present invention, and those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention.

Claims

1. An intelligent farming system, comprising:

a culture pond;

a bait casting machine;

a camera arranged in the culture pond and used for capturing an underwater image; and

a controller, which is connected to the camera and the bait casting machine in a communication way and is used for obtaining the underwater image,

the controller calculates the residual amount of the feed according to the underwater image or transmits the underwater image to a server, and the server calculates the residual amount of the feed, and the controller controls the bait casting machine to feed the breeding pond according to the residual amount of the feed.

2. The intelligent farming system of claim 1, further comprising:

a support, set up in this breed pond, wherein this camera sets up on this support, and this machine of feeding throws at least part of this fodder into this support.

3. The intelligent farming system of claim 2 wherein the frame comprises a mesh plane and a collection pan, the camera direction of the camera being oriented toward the mesh plane, the collection pan extending outwardly from the mesh plane.

4. The intelligent farming system of claim 3 wherein the support further comprises a support portion disposed below the mesh plane, thereby spacing the mesh plane a distance from the bottom of the farming pond.

5. The intelligent farming system of claim 2 further comprising a washer for spraying water or air bubbles to the rack,

the controller executes an image cutting algorithm on the underwater image to obtain a support image area, and judges whether to start the cleaning machine according to the area of the support image area.

6. The intelligent farming system of claim 1 wherein the controller performs an image segmentation algorithm on the underwater image to obtain a feed image area, determines whether the area of the feed image area is less than a threshold, and controls the feeding machine to feed the farming pond if the area of the feed image area is less than the threshold.

7. The intelligent farming system of claim 1 wherein a photographing end of the camera is planar or convex.