KR20210102002A - Factory-type insect breeding system for mass production based on image recognition and breeding cage module - Google Patents

Abstract

The present invention relates to an insect breeding system and, more specifically, to an insect breeding factory system for mass production based on image recognition and a breeding cage module, in which an actuator, a sensor unit, and a camera module are installed in an insect breeding cage, and images of a camera are used to train a you-only-look-once (YOLO) neural network to identify growth, food source status, and excrement of insects. According to the present invention, a method for operating the insect breeding factory system comprises: a step (P1) of building an insect growth activity database with the actuator (30), the sensor unit (40), and the camera module (34)(P1); a step (P2) of training a YOLO neural network with the database after the database building step (P1); a step (P3) of identifying insects, food, and excrement after the YOLO neural network training step (P2); an image recognition module step (P4) after the identification step (P3); and an image discrimination step (P5) of after the image recognition step (P4). According to the present invention, the size and number of insects and the state of food sources and excrement can be checked by applying a YOLO technology which is a single-step object detection algorithm, thereby reducing labor cost by 50% compared to an existing box type and acquiring products more than three times a year by optimizing light source and growth activity on the basis of an insect database.

Description

Factory-type insect breeding system for mass production based on image recognition and breeding cage module}

The present invention relates to an insect breeding system, and in more detail, an actuator, a sensor unit and a camera module are installed in an insect breeding cage, and the image of the camera is YOLO neural network learning to learn about insect growth, insect food source status, and excrement. It relates to a factory-type insect breeding system for mass production based on image recognition and breeding cage module, characterized in that it identifies such things.

Insect breeding is divided into outdoor breeding in a natural environment, living box breeding in plastic cages on a small scale, and large-capacity breeding equipment for mass production.

Most domestic insect farms are living box-type and the container body is covered with a cover, so it is difficult to provide growth conditions for insect breeding, and adults and larvae coexist.

The reality of domestic insect farms is that insect breeders are unsanitary, pre-modern production methods of insect production, and most of them do not have a multi-variety mass production system.

In order to meet the demand for insects, it is necessary to develop a flexible production system suitable for mass production of multiple species or mass production of small species.

Advanced companies are in the stage of establishing a multi-variety mass production system by combining the optimal database of insect growth and ICT (Information & Communication Technology) technology.

Currently, the data base for photoperiod management of temperature, humidity and light source required for insect growth activity is incomplete.

Therefore, in order to build a database necessary for insect growth, it is necessary to install a camera in the breeding gauge, and to make a DB on the insect growth status, number of individuals, and feeding activity based on the camera image information, and use the artificial intelligence neural network for learning data. The need for its use is emerging.

In addition, it is expected that labor costs can be reduced by 50% compared to the existing manual method by building a web-based scalable insect growth management system to which IoT is applied. It is expected that it will be able to produce more than three times.

1. KR 10-1986764 B1, 'Composite breeding apparatus using a mobile bogie system and its remote control method', Registration date: 2019.05.31. 2. KR 10-1946839 B1, 'Useful insect breeding system', registration date: 2019.02.01. 3. KR 10-2011-0061954 A, 'A portable device providing information on animals, plants, and insects and a method for providing the information', Publication date: 2011.06.10.

The present invention was devised to solve the above problems, by installing an actuator, a sensor unit and a camera module in an insect breeding cage, and learning the image of the camera by YOLO neural network learning, insect growth, insect food source status, and excrement, etc. It is an object of the present invention to disclose a factory-type insect breeding system for mass production based on image recognition and breeding cage module that can identify .

In order to achieve the object of the present invention, the present invention is an insect breeding cage with a receiving part capable of loading multiple layers, an entrance on one side, and a humidification module 31, a heating module 32, and a light source module in the insect breeding cage. (33) and the actuator 30 of the camera module 34, the sensor unit 40 of the temperature sensor 41 and the humidity sensor 42, and a movable storage case 10 having a wheel 15 at the lower end thereof Wow; A data collection unit 51 and an actuator control unit 53 for collecting control and information of the actuator 30, a data storage unit for storing information of the sensor unit 40, and a power supply unit 55 and an insect growth environment collecting device 50 for controlling an insect growth environment by having a communication unit 54 with; In the breeding cage module-based mass production factory type insect breeding system including; ), building an insect growth activity database with the sensor unit 40 and the camera module 34 (P1);

After the data building step (P1), the step of learning the database YOLO neural network (P2); After the YOLO neural network learning step (P2), the environment identification step (P3) of insects, food, and excrement; After the identification step (P3), the image recognition module step (P4); After the image recognition step (P4), the image identification step (P5); to disclose an image recognition and breeding cage module-based factory type insect breeding system for mass production, characterized in that it includes.

In addition, the present invention intends to disclose a factory-type insect breeding system for mass production based on image recognition and breeding cage module, characterized in that it further comprises; visualizing or graphing the insect growth environment and growth information.

In addition, the present invention is an image recognition and breeding cage module-based factory-type insect breeding for mass production, characterized in that it further comprises; transmitting the environmental status of insect growth, abnormal symptoms, and the appropriate shipment time to the mobile web of the administrator or user. We want to start the system.

Due to the above solution, the present invention has the following effects.

First, it is possible to check the size, number of insects, food sources and the status of excreta by applying YOLO technology, a single-step method of object detection algorithm, by using deep learning technology for camera images individually installed on insect breeding gauges.

Second, by building a web-based, scalable insect growth management system applying IoT, it is possible to reduce labor costs by 50% compared to the existing box type.

Third, it is possible to produce more than 3 times a year by optimizing the light source and growth activity using the insect date base (previously, once a year)

1 is a flowchart of a preferred embodiment of the present invention.
2 is a flowchart of an image recognition and processing process.
3 is a flowchart of a mobile or web-based visualization module.
4 is a block diagram of a breeding gauge.
5 is a schematic diagram of an emission automation system and a flow chart of a central control system.
6 is a configuration diagram of the movable storage case 10 .
7 is a configuration diagram of the automatic excretion discharge.
8 is a picture of insect, food source, and excrement object detection using the YOLO algorithm.
10 is an embodiment of the actuator 30 applied to the breeding gauge.

Hereinafter, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those of ordinary skill in the art to which the present invention pertains can easily implement them. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein.

In order to clearly explain the present invention, parts irrelevant to the description (such as electric circuits and other necessary parts) are omitted, and the same or similar components are assigned the same reference numerals throughout the specification.

In addition, since the size, color, and thickness of each component shown in the drawings are arbitrarily indicated for convenience of description, the present invention is not necessarily limited to the bars shown in the drawings. In order to clearly express various parts and regions in the drawings, the colors for each component are expressed as in the drawings, but it is obvious that the patent rights range includes all of the various colors.

The thickness and size of the drawings and components are also enlarged for clarity of expression. In addition, in the drawings, the thickness of some parts and regions is exaggerated for convenience of explanation.

The present invention relates to an insect breeding system, and in more detail, an actuator, a sensor unit and a camera module are installed in an insect breeding cage, and the image of the camera is YOLO neural network learning to learn about insect growth, insect food source status, and excrement. It relates to a factory-type insect breeding system for mass production based on image recognition and breeding cage module, characterized in that it identifies such things.

Fig. 1 is a flowchart of a preferred embodiment of the present invention, Fig. 2 is a flowchart of image recognition and processing, and Fig. 3 is a flowchart of a mobile or web-based visualization module.

As shown in Figures 1 to 3, the present invention is an insect breeding cage with a receiving part that can load multiple layers, a doorway on one side, and a humidification module 31, a heating module 32, and a light source in the konchong breeding cage. The module 33 and the actuator 30 of the camera module 34, the sensor unit 40 of the temperature sensor 41 and the humidity sensor 42, and a movable storage case 10 having a wheel 15 at the lower end )Wow; A data collection unit 51 and an actuator control unit 53 for collecting control and information of the actuator 30, a data storage unit for storing information of the sensor unit 40, and a power supply unit 55 And an insect growth environment collecting device 50 for controlling the insect growth environment by having a communication unit 54; In the breeding cage module-based mass production factory type insect breeding system including; ), building an insect growth activity database with the sensor unit 40 and the camera module 34 (P1); After the data building step (P1), the step of learning the database YOLO neural network (P2); After the YOLO neural network learning step (P2), the environmental identification step (P3) of insects, food, and excrement; After the identification step (P3), the image recognition module step (P4); After the image recognition step (P4), the image discrimination step (P5); characterized in that it comprises a.

6 is a configuration diagram of the movable storage case 10, FIG. 7 is a configuration diagram of automatic excretion discharge, and FIG. 10 is an embodiment of the actuator 30 applied to the breeding gauge.

Referring to the above figure, it is preferable that the movable storage case 10 has an accommodating portion capable of loading multi-layered insect breeding cages so as to accommodate various kinds of insects, and a doorway on one side thereof.

4 is a configuration diagram of a breeding gauge, and FIG. 5 is a schematic diagram of an automatic discharge and a flow chart of a central control system.

The insect breeding cage is provided with a humidification module 31, a heating module 32, and a light source module 33 to provide temperature and humidity suitable for insects, respectively, and a temperature sensor 41 and a humidity sensor 42 for control. ) It is preferable to provide a sensor unit 40 of.

In addition, the insect breeding cage is equipped with an individual camera and a camera module 34 to check the growth status of insects, the status of adults and larvae, the status of the population, the status of the food source, and the status of excreta.

Multiple types of insects may be accommodated in the movable storage case 10, or a single type of insects may be accommodated.

Object detection technology such as insects, food sources, and excrement is important in the image transmitted from the camera. The object detection is an automated technique for distinguishing and identifying an object of interest from a background in a video image, and is one of computer vision technologies.

For correct object detection in video images, you need to set up a bounding box to associate categories of objects that represent objects.

In this case, Deep Learning is used.

Convolutional Neural Network (CNN) is one of the core technologies of computer vision. Convolutional neural network technology enables regression and classification for limited object detection.

Restricted object detection processes a fixed number of objects like the existing computer vision tasks such as Image Recognition, Key Points Detection, and Semantic Segmentation. Model it as a regression problem.

In the present invention, the majority of subjects include insects, food sources, excrement, and classification of adults and larvae.

8 is a picture of insect, food source, and excrement object detection using the YOLO algorithm.

The You Only Look Once (YOLO) is a representative single-step object detection algorithm. The YOLO algorithm divides the original image into grids of equal size. Thereafter, for each grid, the number of bounding boxes designated with a predefined shape centered on the center of the grid is predicted, and reliability is calculated based on this.

The result is whether the image contains objects, or whether the background is alone. Identify object categories by choosing locations with high object confidence.

The number of bounding boxes having a predefined shape is called 'Anchor Boxes', and the anchor boxes are generated from data by a K-means algorithm, and secure prior information on the object size and shape of the data set.

Each anchor is designed to detect objects of different sizes and shapes.

As shown in FIG. 8, it can be confirmed that two anchors are formed in Chinese cabbage, and four anchors are formed in adults. This algorithm detects objects of similar size to the anchor box, but the final prediction is different from the anchor's location or size.

The optimized offset obtained from the result obtained from the feature map of the image is added to the anchor position or size.

The YOLO is characterized by a very fast speed with simple processing. Compared to other real-time vision technologies, it has twice the performance. Classes are distinguished by looking at the entire image at once.

In addition, YOLO can extract objects and the like from real-time videos.

An object of the present invention is to secure a database for optimal growth conditions for each insect by utilizing the YOLO algorithm.

It is desirable to process the insect growth database to visualize the information or to express it in various graphs. Accordingly, the present invention is characterized in that it further comprises; the step of expressing the insect growth environment and growth information in a visualization or graph.

In addition, the present invention is characterized in that it further comprises; transmitting the environmental status of insect growth, abnormal symptoms, and appropriate shipping time to the mobile web of the administrator or user.

The most productive method for insect growth is firstly to quickly identify abnormal signs of growth, secondly, separation of adults and larvae, and thirdly, differentiation of growth conditions for each insect.

To this end, there is a need to quickly deliver the above information to an administrator or a user. Therefore, the signs of icing, the state of growth conditions, the appearance of larvae, etc. should be easily transmitted to the user via mobile as well as wired or wireless.

This achievement is the result of the research performance of the 2019 industry-university-research Collabo R&D project (No.s2736757) supported by the Ministry of SMEs and Startups.

This work was supported by project for Collabo R&D between Industry, Academy, and Research Institute funded Korea Ministry of SMEs and Startups in 2019.(Project No. s2736757)

10: movable storage case, 11: bogie frame, 12: temperature and humidity supply pipe,
13: mist spray nozzle, 15: whirl, 16: barbed wire,
20: control unit, 21: motor drive, 22: transfer motor,
23: transfer device,
30: actuator, 31: humidification module, 32: heating module,
33: light source module, 34: camera module,
40: sensor unit, 41: temperature sensor, 42: humidity sensor,
50: insect growth environment collection device, 51: date collection unit,
52: data storage unit, 53: actuator control unit, 54: communication unit,
55: power unit,
60: communication network, 70: gateway,
100: factory type insect breeding system

Claims (3)

An actuator of a humidifying module 31 , a heating module 32 , a light source module 33 and a camera module 34 in the insect breeding cage with a receiving part capable of stacking multiple layers of an insect breeding cage, and an entrance on one side of the cage (30), the sensor unit 40 of the temperature sensor 41 and the humidity sensor 42, and a movable storage case 10 having a wheel 15 at the lower end thereof; A data collection unit 51 and an actuator control unit 53 for collecting control and information of the actuator 30, a data storage unit for storing information of the sensor unit 40, and a power supply unit 55 and an insect growth environment collecting device 50 for controlling an insect growth environment by having a communication unit 54 with; In the breeding cage module-based mass production factory type insect breeding system including;
building an insect growth activity database with the actuator 30, the sensor unit 40 and the camera module 34 (P1);
After the data building step (P1), the step of learning the database YOLO neural network (P2);
After the YOLO neural network learning step (P2), the environmental identification step (P3) of insects, food, and excrement;
After the identification step (P3), the image recognition module step (P4);
After the image recognition step (P4), the image identification step (P5); Image recognition and breeding cage module-based factory-type insect breeding system for mass production, characterized in that it comprises a. According to claim 1,
Image recognition and breeding cage module-based factory type insect breeding system for mass production, characterized in that it further comprises; expressing the insect growth environment and growth information as a visualization or graph.
According to any one of claims 1 to 2,
Image recognition and breeding cage module-based factory-type insect breeding system for mass production, characterized in that it further comprises; transmitting the environmental status of insect growth, abnormal signs, and appropriate shipment timing to the mobile web of the administrator or user.

Images