阅读说明：本技术 一种害虫识别系统 (Pest identification system ) 是由 王田 王征 于 2020-05-19 设计创作，主要内容包括：本发明涉及病虫害防治领域,具体公开了一种害虫智能识别系统,包括前端信息采集终端、物联网网关、后端云平台和设置在后端云平台内的图像分类识别模型,前端采集终端和物联网网关信号连接,物联网网关和后端云平台信号连接,前端信息采集终端包括若干智能捕虫灯和环境信息采集器,本发明通过智能捕虫灯自行进行捕虫并进行害虫识别,避免了现有技术中通过人工捕捉,并且从始至终进行人工识别,人力成本浪费巨大,识别准确度低,识别效率低且专业人员配备不足的问题。(The invention relates to the field of pest control, and particularly discloses an intelligent pest identification system which comprises a front-end information acquisition terminal, an Internet of things gateway, a rear-end cloud platform and an image classification identification model arranged in the rear-end cloud platform, wherein the front-end information acquisition terminal is in signal connection with the Internet of things gateway, the Internet of things gateway is in signal connection with the rear-end cloud platform, and the front-end information acquisition terminal comprises a plurality of intelligent pest trapping lamps and an environment information collector.)

1. A pest identification system characterized by: the system comprises a front-end information acquisition terminal, an Internet of things gateway, a rear-end cloud platform and an image classification and identification model arranged in the rear-end cloud platform, wherein the front-end information acquisition terminal is in signal connection with the Internet of things gateway, the Internet of things gateway is in signal connection with the rear-end cloud platform, and the front-end information acquisition terminal comprises a plurality of intelligent insect catching lamps and an environment information collector;

the intelligent insect catching lamp comprises a solar panel, a photosensitive sensor, a lampshade, a transparent lamp holder fixedly connected to the inner wall of the lampshade, a bulb arranged on the top surface of the lamp holder, a current sensor arranged on one side of the bulb, an insect receiving chamber and a single chip microcomputer controller which are connected to the lower end of the lampshade in a threaded manner, an insect electromagnetic adsorption device is arranged on the bottom surface of the lamp holder, the insect electromagnetic adsorption device comprises an iron core, a coil wound on the iron core and an electric fence fixedly connected to the free end of the iron core, a scraper capable of magnetically adsorbing the electric fence is connected to the electric fence in a sliding manner, a wireless camera is obliquely arranged on the bottom surface of the lamp holder towards the electric fence, a baffle is further fixedly connected to the inner wall of the lampshade through which is connected with an Internet of things gateway signal, the baffle is positioned below the insect electromagnetic adsorption device, an opening capable of containing an insect corpse is formed, the solar energy collecting lamp is characterized in that a plurality of insect attracting holes for insects to enter are formed in the lamp shade, the insect attracting holes are circumferentially and uniformly distributed in the lamp shade between the lamp holder and the baffle plate, the solar energy plate is electrically connected with the photosensitive inductor, the current inductor, the coil, the wireless camera, the single chip microcomputer controller and the bulb, and the single chip microcomputer controller is in signal connection with the Internet of things gateway;

the environment collector comprises a GPS positioning module, a temperature and humidity acquisition module and an illumination acquisition module, and the GPS positioning module, the temperature and humidity acquisition module and the illumination acquisition module are all in signal connection with the gateway of the Internet of things;

the internet of things gateway comprises a data receiving module and a data sending module, the data receiving module receives images and environmental information collected by the front-end information collection terminal, and the data sending module forwards the images and the environmental information collected by the front-end information collection terminal to the rear-end cloud platform.

The rear-end cloud platform comprises an image classification and identification model, and the image classification and identification model classifies and counts the insect images with environmental information labels acquired by the front-end information acquisition terminal.

2. A pest identification system as claimed in claim 1, wherein: an image feedback information base, an identified picture base and a picture base to be artificially marked are established in the back-end cloud platform, when the back-end cloud platform receives related image information, images are identified and counted by using a pre-trained image classification identification model, the image classification identification model can provide different types for each picture, S is a pest identification type, the confidence of the image classification identification model for identifying the pictures as S is P%, a preset initial value is T, the image classification identification model can be identified as multiple pest types for the same picture, namely, multiple S and multiple P exist, and when P is used, when P is the same picture, multiple S and multiple P exist_maxWhen the picture is more than or equal to T, the picture is accurately identified, the system stores the related information into an image feedback information base, and stores the picture with the environmental information into an identified picture base; when P is present_maxIf the number is less than T, the picture identification is inaccurate, and the system stores the related information into a picture library to be manually marked; after the pictures of the picture library to be manually marked are manually marked and are audited by experts, the system stores the pictures which are manually identified and marked into the identified picture library and stores the related information into the image feedback information library; the system trains the image classification recognition model by using the related picture information of the labeled picture library and updates the model regularly.

3. A pest identification system as claimed in claim 1 or claim 2, wherein: still include the stand, stand one side is equipped with the fastener, and the stand top is equipped with the telescopic link, and solar panel can dismantle to be connected at the telescopic link top, and telescopic link free end fixedly connected with rainshelter, intelligence insect-catching lamp are established in rainshelter below.

4. A pest identification system as claimed in claim 3, wherein: the insect receiving chamber is internally provided with a slide way, an insect receiving disc is arranged below the slide way, the insect receiving disc is connected with the insect receiving chamber through a buckle, and the insect receiving chamber is provided with a plurality of dustproof gauzes.

5. A pest identification system as claimed in claim 4, wherein: the coil is connected with the current inductor in series, the current inductor is connected with the single chip microcomputer controller through signals, the single chip microcomputer controller is electrically connected with the wireless camera, and the coil and the bulb are connected with the solar panel in parallel.

6. A pest identification system as claimed in claim 5, wherein: the telescopic link tip is opened flutedly, rings embedding recess.

7. A pest identification system as claimed in claim 2 or claim 6, wherein: the image classification recognition model can also be trained by adopting a deep learning algorithm or an SVM algorithm or a naive Bayes algorithm.

Technical Field

The disclosure relates to the field of pest control, in particular to a pest identification system.

Background

The pest control is an important link of agricultural production and forest maintenance. Because the pesticide has the highest effect on pests with a certain pest age, the early warning of the occurrence period of the pest corresponding to the pest age is used as an important forecast content and is the key for preventing and controlling the pests. For example, in field investigation, after capturing pests of various pest states and pest ages in the field, manually dividing and grading the pest states and the pest ages indoors, calculating and forecasting the control right period of the pests or dissecting the development level of ovaries of female adults according to the period of each pest state (age) under a certain temperature condition, and judging the development condition of the ovaries, wherein the method can be used for judging the source properties of migratory pests. Because the prior art mainly relies on the manual work to discern the pest state of pest, and catch by hand, the discernment degree of accuracy and efficiency are extremely low, catch by hand, and the human cost is huge to the interference of human factor is big, influences the accuracy that prevents and treats opportune moment or worm source judgement of prediction. In addition, the manual identification is carried out all the time, the requirement on the professional level of the inspectors is high, and proper personnel cannot be equipped in the local for identification. .

Disclosure of Invention

The invention aims to solve the problems that the existing pest identification is manually captured and is manually identified all the time, the labor cost is wasted greatly, the identification accuracy is low, the identification efficiency is low, and the professional personnel are not equipped sufficiently.

In order to achieve the purpose, the basic scheme of the invention provides a pest recognition system which comprises a front-end information acquisition terminal, an internet of things gateway, a rear-end cloud platform and an image classification recognition model arranged in the rear-end cloud platform, wherein the front-end information acquisition terminal is in signal connection with the internet of things gateway, the internet of things gateway is in signal connection with the rear-end cloud platform, and the front-end information acquisition terminal comprises a plurality of intelligent pest trapping lamps and an environment information collector;

the intelligent insect catching lamp comprises a solar panel, a photosensitive sensor, a lampshade, a transparent lamp holder fixedly connected to the inner wall of the lampshade, a bulb arranged on the top surface of the lamp holder, a current sensor arranged on one side of the bulb, an insect receiving chamber and a single chip microcomputer controller which are connected to the lower end of the lampshade in a threaded manner, an insect electromagnetic adsorption device is arranged on the bottom surface of the lamp holder, the insect electromagnetic adsorption device comprises an iron core, a coil wound on the iron core and an electric fence fixedly connected to the free end of the iron core, a scraper capable of magnetically adsorbing the electric fence is connected to the electric fence in a sliding manner, a wireless camera is obliquely arranged on the bottom surface of the lamp holder towards the electric fence, a baffle is further fixedly connected to the inner wall of the lampshade through which is connected with an Internet of things gateway signal, the baffle is positioned below the insect electromagnetic adsorption device, an opening capable of containing an insect corpse is formed, the solar energy collecting lamp is characterized in that a plurality of insect attracting holes for insects to enter are formed in the lamp shade, the insect attracting holes are circumferentially and uniformly distributed in the lamp shade between the lamp holder and the baffle plate, the solar energy plate is electrically connected with the photosensitive inductor, the current inductor, the coil, the wireless camera, the single chip microcomputer controller and the bulb, and the single chip microcomputer controller is in signal connection with the Internet of things gateway;

the environment collector comprises a GPS positioning module, a temperature and humidity acquisition module and an illumination acquisition module, and the GPS positioning module, the temperature and humidity acquisition module and the illumination acquisition module are all in signal connection with the gateway of the Internet of things;

the internet of things gateway comprises a data receiving module and a data sending module, the data receiving module receives images and environmental information collected by the front-end information collection terminal, and the data sending module forwards the images and the environmental information collected by the front-end information collection terminal to the rear-end cloud platform.

The rear-end cloud platform comprises an image classification and identification model, and the image classification and identification model classifies and counts the insect images with environmental information labels acquired by the front-end information acquisition terminal.

Further, an image feedback information base, an identified picture base and a picture base to be artificially marked are established in the back-end cloud platform, when the back-end cloud platform receives related image information, images are identified and counted by using a pre-trained image classification identification model, the image classification identification model can give different types to each picture standard, S is a pest identification type, the confidence of the image classification identification model for identifying the image as S is P%, a preset initial value is T, the image classification identification model can be identified as multiple pest types for the same picture, namely, multiple S and multiple P exist, and when P is P, multiple S and multiple P exist_maxWhen the picture is more than or equal to T, the picture is accurately identified, the system stores the related information into an image feedback information base, and stores the picture with the environmental information into an identified picture base; when P is present_maxIf the number is less than T, the picture identification is inaccurate, and the system stores the related information into a picture library to be manually marked; to the pictures of the picture library to be manually markedAfter manual marking and expert review, the system stores the picture after manual identification marking into an identified picture library and stores related information into an image feedback information library; the system trains the image classification recognition model by using the related picture information of the labeled picture library and updates the model regularly.

Further, still include the stand, stand one side is equipped with the fastener, and the stand top is equipped with the telescopic link, and solar panel can dismantle to be connected at the telescopic link top, and telescopic link free end fixedly connected with rainshelter, intelligence insect-catching lamp are established in rainshelter below.

Furthermore, a slide way is arranged in the insect receiving chamber, an insect receiving disc is arranged below the slide way, the insect receiving disc is connected with the insect receiving chamber through a buckle, and a plurality of dustproof gauzes are arranged on the insect receiving chamber.

Further, the coil is connected with a current inductor in series, the current inductor is connected with a single chip microcomputer controller in a signal mode, the single chip microcomputer controller is electrically connected with the wireless camera, and the coil and the bulb are connected with the solar panel in parallel.

Furthermore, the telescopic link tip is opened flutedly, rings embedding recess.

Furthermore, the image classification recognition model can also be trained by adopting a deep learning algorithm or an SVM algorithm or a naive Bayes algorithm.

The principle and the effect of the invention are as follows:

1. the invention uses the intelligent insect catching lamp to emit light at night to catch insects, after the insects contact the electric fence, the current sensor senses the current change and transmits the information to the single-chip microcomputer controller, the single-chip microcomputer controller controls the wireless camera to photograph the insects, the photographs are transmitted to the rear-end cloud platform after being photographed, the insects are electrocuted by the current, the coil is powered off, the scraper loses the suction force and falls down, the dead bodies of the insects on the electric fence are scraped into the insect receiving chamber, and after the insect photographs are transmitted to the rear-end cloud platform, the rear-end cloud platform classifies and counts the insect images marked with the environmental information collected by the front-end information collecting terminal by using the pre-trained image classification and identification model, thereby realizing effective and intelligent insect identification and monitoring, and being capable of realizing the specific position, temperature, humidity and illumination collected by the environment collector, and the data is transmitted to a rear-end cloud platform, so that further specific positioning analysis is carried out by combining the analyzed insect pictures, and better pest control is achieved.

2. The solar panel is arranged, an external power supply is not needed, solar energy is converted into electric energy in the daytime and stored, the photosensitive sensor can control the bulb to be turned on only at night, and the solar energy-saving lamp is energy-saving and efficient.

3. According to the intelligent insect catching lamp, the intelligent insect catching lamp is used for automatically catching insects and identifying pests, so that the problems that manual catching is adopted in the prior art, manual identification is carried out from beginning to end, manpower cost waste is huge, identification accuracy is low, identification efficiency is low, and professional equipment is insufficient are solved.

Drawings

FIG. 1 is a schematic view of an intelligent insect catching lamp in a pest recognition system according to the present invention;

FIG. 2 is a partial schematic view of an intelligent insect catching lamp in a pest identification system of the present invention;

fig. 3 is a flowchart illustrating an operation of a pest recognition system according to the present invention.

Detailed Description

The following is further detailed by the specific embodiments:

reference numerals in the drawings of the specification include: the intelligent insect catching device comprises a stand column 1, an expansion rod 2, a solar panel 3, a rain shelter 4, an intelligent insect catching lamp 5, a lifting ring 6, a lampshade 7, a ventilation hole 8, a bulb 9, a current inductor 10, a lamp holder 11, a wireless camera 12, an iron core 13, a coil 14, an electric fence 15, a scraping plate 16, a baffle 17, a connecting sleeve 19, a slide way 20, an insect receiving chamber 21, an insect receiving disc 22, an insect attracting hole 23 and dustproof gauze 24.