Automatic inspection system and method for pig farm

Abstract

The invention relates to the technical field of image processing, and particularly discloses an automatic inspection system and method for a pig farm, and the system comprises an edge server and an intra-field camera for collecting videos in a field; the system further comprises an off-site unmanned aerial vehicle. The edge server is used for identifying the live pigs in the abnormal state according to the in-site video, and is also used for judging whether the number of the live pigs in the abnormal state exceeds a threshold value or not, and generating an out-site inspection instruction if the number exceeds the threshold value; the off-site unmanned aerial vehicle is used for receiving the off-site inspection instruction of the edge server, executing an off-site inspection task according to a preset flight route, and sending an inspection video to the edge server; and the edge server is also used for identifying whether an abnormal event exists according to the inspection video and generating a reminding item which comprises an external event reminding generated when the abnormal event is identified or a live pig state abnormity reminding generated when the abnormal event is not identified. By adopting the technical scheme, internal and external linkage monitoring of the farm can be realized, and the cause of the current abnormal state of the live pig can be quickly determined.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a pig farm automatic inspection system and method. BACKGROUND

[0002] In modern livestock farming, especially in large-scale pig farms, the health status of pigs is directly related to the breeding environment safety and the breeding benefit and animal epidemic prevention level. The traditional breeding farm relies on manual inspection and abnormal investigation, which has many disadvantages: manual inspection is low in efficiency and high in cost, and it is difficult to find early abnormal behaviors of pigs, such as abnormal movement and abnormal call; therefore, a video monitoring system for the breeding farm appears, which can automatically identify the state of pigs and warn the breeding personnel whether they are in the breeding farm or not.

[0003] However, the pig farm is not an isolated environment, most of the pig farms are located in rural areas, and the surrounding environment is complex, which is easily disturbed by external factors. For example, the surrounding construction noise, smoke and odor generated by crop or garbage burning, and even the intrusion of wild animals, may cause strong stress reaction of pigs in the breeding farm, and trigger group call abnormality or movement abnormality. These external disturbances are one of the important causes of abnormal state of pigs. However, the current external monitoring of the breeding farm mainly relies on fixed off-site cameras. Due to the wide coverage of the breeding environment and the limited video monitoring, the fixed camera can only capture the situation in a short range. For example, near the gate or fence, there is inevitably a large blind area, which is difficult to trace the external events comprehensively, especially when the breeding personnel are away from the breeding farm, it is difficult to understand the on-site situation, and it is difficult to find a solution as soon as possible.

[0004] Therefore, there is an urgent need for a pig farm automatic inspection system and method which can realize the linkage monitoring of the inside and outside of the breeding farm, and quickly determine the cause of the current abnormal state of pigs. SUMMARY

[0005] One of the purposes of the present application is to provide a pig farm automatic inspection system which can realize the linkage monitoring of the inside and outside of the breeding farm, and quickly determine the cause of the current abnormal state of pigs.

[0006] In order to solve the above technical problems, the present application provides the following technical solutions: A pig farm automatic inspection system, comprising an edge server and an on-site camera for collecting on-site video; further comprising an off-site unmanned aerial vehicle; The edge server is used for identifying pigs in abnormal state according to the on-site video, and is further used for judging whether the number of pigs in abnormal state exceeds a threshold value, and generating an off-site inspection instruction if the threshold value is exceeded; The off-site drone is used to receive off-site inspection instructions from the edge server, perform off-site inspection tasks according to the preset flight route, and send the inspection video to the edge server; The edge server is also used to identify whether there are abnormal events based on the inspection video and generate reminders, including external event reminders generated when abnormal events are identified, or abnormal pig status reminders generated when no abnormal events are identified.

[0007] The basic principles and beneficial effects of the scheme are as follows: This solution achieves intelligent inspection through the collaborative operation of edge servers, in-farm cameras, and off-farm drones, combining internal and external coordination with active and passive methods. The edge server analyzes in-farm video data in real time to identify potential abnormal behaviors in newborn pigs. Once the number of abnormal pigs exceeds a set threshold, the off-farm drone is automatically triggered to perform an inspection mission. The drone acquires on-site video along a preset path and transmits it back to the edge server for analysis and judgment. This effectively identifies whether the abnormal condition of the newborn pigs is caused by internal or external factors, helping farm managers quickly determine the cause of the problem.

[0008] In summary, this solution breaks through the limitations of traditional on-site and off-site monitoring, enabling proactive identification of the status of pigs on-site and combining off-site video analysis to identify the causes of abnormalities. This improves the response speed to emergencies and helps staff quickly determine the causes of current abnormal pig statuses.

[0009] Furthermore, it also includes cloud servers and user terminals; Edge servers are also used to send on-site videos, inspection videos, and reminders to cloud servers; The cloud server is used to push reminders to user terminals; it is also used to receive remote monitoring requests from user terminals and push on-site videos and inspection videos to user terminals.

[0010] Furthermore, the cloud server is also used to receive automatic inspection plans created by user terminals and send the automatic inspection plans to the edge server; the automatic inspection plan includes the inspection route and inspection time. Edge servers are also used to control off-site drones to perform off-site inspection tasks according to the inspection route during inspection time.

[0011] Furthermore, it also includes off-site cameras. The edge server is also used to identify whether there are any pig transport vehicles arriving based on the off-site videos collected by the off-site cameras. If so, it generates disinfection monitoring instructions. Off-site drones are used to receive disinfection monitoring instructions from the edge server, fly to the pig transport vehicle, and film the disinfection process until it is completed. The edge server is also used to analyze the video of the disinfection process to determine whether the disinfection meets the preset standards. If it does not meet the preset standards, it sends a reminder instruction to the drone. Off-site drones are also used to play reminder audio according to reminder instructions.

[0012] During the operation of pig farms, vehicles transporting pigs are high-risk carriers of external pathogens entering the farm area. The viruses and bacteria they carry can easily spread through contact, causing large-scale animal disease outbreaks. Therefore, disinfection management during pig transportation has always been a key and challenging aspect of the livestock disease prevention system. However, traditional methods relying on manual observation or fixed camera recording not only have blind spots, failing to comprehensively cover the disinfection area, but are also prone to oversights or improper human operation, rendering disease prevention measures ineffective and failing to ensure practical results.

[0013] This preferred solution uses off-site cameras to detect the arrival of transport vehicles and trigger subsequent disinfection monitoring procedures. Drones perform close-range, all-angle video capture, flexibly addressing blind spots caused by different vehicle types and parking locations, ensuring complete and clear recording of the entire disinfection process. An edge server intelligently judges whether the disinfection operation conforms to preset specifications. Upon detecting any abnormalities, it immediately plays audio alerts via drones, promptly intervening with operators to prevent potential disease outbreaks due to negligence or violations. This is of great significance for ensuring the biosecurity of farms and blocking disease transmission routes.

[0014] Furthermore, the abnormal state of the pigs includes abnormal vocalizations and abnormal behavior; the abnormal events include abnormal noise and abnormal odor.

[0015] When pigs exhibit abnormal vocalizations or unusual behavior, inspections by drones outside the farm can determine whether there are abnormal noises or odors, making it easier for farmers to determine whether the cause of the abnormalities is internal or external.

[0016] Furthermore, when the inspection time is during the day, the off-site drone inspection video includes visible light images and audio; when the inspection time is at night, the off-site drone inspection video includes infrared images and audio. The edge server is also used to analyze the presence of animals based on infrared images. If there are any, it identifies the animal species and counts them, determines whether the number of animals of a preset species exceeds a number threshold, and generates an animal anomaly alert if the number exceeds the threshold.

[0017] In the daily operation and maintenance of pig farms, in addition to the health and disease prevention management of the pigs themselves, potential animal disturbances in the external environment also need to be prevented. Especially at night, wild animals such as rodents are more likely to enter the vicinity of the farm, becoming a hidden channel for disease transmission, or disturbing the normal routines of the pigs and triggering stress responses. Traditional nighttime inspections mostly rely on manpower or fixed infrared cameras, which are not only inefficient but also have the problem of delayed prevention.

[0018] This preferred solution allows off-site drones to automatically switch to infrared image and audio acquisition mode at night, enhancing target recognition capabilities in low-light or no-light conditions. This enables the edge server to accurately identify whether animals have entered the field of view and further intelligently identify and count the types and numbers of animals. If the number of specific high-risk animals such as rats exceeds a certain threshold, an animal anomaly alert is generated, prompting farmers to intervene promptly or take preventative measures.

[0019] Furthermore, when the edge server detects abnormal pig cries based on the on-site video, it first calls the off-site video collected by the off-site camera to identify whether any registered personnel or vehicles have arrived near the farm. If the edge server detects that a registered person or vehicle has arrived, and the current inspection time is daytime, no off-site inspection instruction will be generated.

[0020] Furthermore, if the edge server detects the arrival of registered personnel, and if the current inspection time is at night, and the edge server has generated an animal abnormality alert based on the nighttime infrared image inspection results, and if the edge server detects through the off-site camera that the breeding personnel are more than the preset safe distance from the farm gate, the edge server generates an escort inspection command to the off-site drone.

[0021] The second objective of this invention is to provide an automatic inspection method for pig farms, comprising the following: S1. Deploy off-site drones at pig farms; S2. Collect videos of pig farms; S3. Identify pigs in abnormal condition based on the on-site video, determine whether the number of pigs in abnormal condition exceeds the threshold, and if it exceeds the threshold, generate an off-site inspection instruction. S4. The off-site drone receives off-site inspection instructions, performs off-site inspection tasks according to the preset flight route, and takes inspection videos; S5. Identify any abnormal events based on the inspection video and generate reminders, including external event reminders generated when abnormal events are identified, or abnormal pig status reminders generated when no abnormal events are identified.

[0022] Furthermore, it also includes: S6. Receive the user's automatic inspection plan, which includes the inspection route and inspection time; the off-site drone performs off-site inspection tasks according to the inspection route during the inspection time. S7. Identify whether there are any pig transport vehicles arriving by collecting off-site videos from off-site cameras; if so, generate a disinfection monitoring command. S8. The off-site drone receives the disinfection monitoring command, flies to the pig transport vehicle, and takes a video of the disinfection process until the disinfection is completed. S9. Analyze the video of the disinfection process to see if the disinfection meets the preset standards. If it does not meet the preset standards, send a reminder instruction to the drone. The drone outside the site will play a reminder audio according to the reminder instruction. Attached Figure Description

[0023] Figure 1 This is a logic block diagram of an embodiment of an automatic inspection system for pig farms. Detailed Implementation

[0024] The following detailed description illustrates the specific implementation method: Example 1 like Figure 1 As shown in the figure, an automatic inspection system for pig farms in this embodiment includes a cloud server, an edge server, an in-farm camera, an off-farm drone, and a user terminal.

[0025] The on-site cameras are used to collect video footage from inside the pig farm.

[0026] Off-site drones are deployed on the roof of the pig farm via a drone airport.

[0027] The edge server is used to identify pigs in abnormal condition based on the video inside the farm, and also to determine whether the number of pigs in abnormal condition exceeds a threshold. If it exceeds the threshold, it generates an off-farm inspection command. The abnormal states of the pigs include abnormal vocalizations and abnormal behaviors. Abnormal vocalizations refer to audio decibels in the video exceeding a set decibel value, while abnormal behaviors refer to panic-induced escape actions. For example, the local pig behavior recognition model on the edge server analyzes the images in the video to identify panic-induced escape actions such as pigs moving rapidly in a short period of time, banging against cages, fences, or walls, shrinking into corners, and avoiding light. In this embodiment, abnormal vocalizations are only detected in decibels, while abnormal behaviors are counted, with a threshold of 5 pigs. The specific time and distance of rapid movement within a short period can be set according to actual conditions.

[0028] The off-site drone is used to receive off-site inspection instructions from the edge server, execute off-site inspection tasks according to a preset flight route, and send the inspection video to the edge server. In this embodiment, the preset flight route is to circle around the pig farm, expanding the radius after each circle, until the preset number of circles is completed.

[0029] The edge server is also used to identify whether there are abnormal events based on the inspection video and generate reminders, including external event reminders generated when abnormal events are identified, or abnormal pig status reminders generated when no abnormal events are identified.

[0030] The abnormal events include noise anomalies and odor anomalies. Noise anomalies refer to external noise levels exceeding the set decibel level, including noise from construction outside the farm, vehicle horns, arguments, etc. Odor anomalies are identified by detecting smoke in images using a local flame and smoke detection model mounted on the edge server, such as smoke from fireworks or burning garbage, crops, or weeds near the farm.

[0031] Edge servers are also used to send on-site videos, inspection videos, and reminders to cloud servers; The cloud server is used to push reminders to user terminals; it is also used to receive remote monitoring requests from user terminals and push on-site videos and inspection videos to user terminals.

[0032] The cloud server is also used to receive automatic inspection plans created by user terminals and send the automatic inspection plans to the edge server; the automatic inspection plan includes the inspection route and inspection time; the edge server is also used to control the off-site drone to perform off-site inspection tasks according to the inspection route during the inspection time.

[0033] Based on the above system, this embodiment also provides an automatic inspection method for pig farms, including the following: S1. Deploy a drone airport on the roof of the pig farm and set up off-site drones; S2. Collect videos of pig farms; S3. Identify pigs in abnormal condition based on the on-site video, determine whether the number of pigs in abnormal condition exceeds the threshold, and generate an off-site inspection instruction if the threshold is exceeded. The abnormal condition of the pigs includes abnormal vocalizations and abnormal behavior. Abnormal vocalizations refer to the audio decibel level in the video exceeding the set vocalization decibel value, and abnormal behavior refers to panic-induced escape actions. S4. The off-site drone receives off-site inspection instructions, performs off-site inspection tasks according to the preset flight route, and takes inspection videos; S5. Identify any abnormal events based on the inspection video and generate alerts, including external event alerts generated when abnormal events are identified, or alerts about abnormal pig status generated when no abnormal events are identified. Abnormal events include abnormal noise and abnormal odor.

[0034] S6. Receive the user's automatic inspection plan, which includes the inspection route and inspection time; the off-site drone performs off-site inspection tasks according to the inspection route during the inspection time.

[0035] In actual pig farming, stress responses are often accompanied by sudden changes in vocalizations or a sudden escape. If not detected and intervened in time, this can lead to group stress, affecting growth and development or causing injury. Traditional reliance on manual inspections often carries the risk of delayed detection. In this solution, on-site cameras collect video data around the clock, and edge servers, through deployed local models, detect abnormal behavior in pigs in real time. For example, if five or more pigs simultaneously exhibit panicked escape behaviors such as huddling together, or if there is an abnormal increase in audio decibels, it will automatically be identified as group stress behavior. An off-site inspection command will be generated, triggering an off-site drone to take off and conduct an expanded inspection of the surrounding environment along a preset path. For instance, if the inspection video, analyzed by the edge server, detects burning garbage on the side of an external road with dense smoke in the image, the system will automatically generate an "external event alert" and push the alert to the farmer's user terminal, such as a mobile phone. After reviewing the real-time video and determining the location of the burning, the farmer can go to the site immediately to handle the situation and improve ventilation within the farm, effectively preventing the incident from escalating.

[0036] Example 2 The difference between this embodiment and Embodiment 1 is that the system in this embodiment also includes an off-site camera. The edge server is also used to identify whether a pig transport vehicle has arrived based on the off-site video collected by the off-site camera. If so, it generates a disinfection monitoring instruction. The off-site drone is used to receive disinfection monitoring instructions from the edge server, fly to the pig transport vehicle, and record video of the disinfection process until disinfection is completed. The edge server is also used to analyze the video of the disinfection process to see if the disinfection meets the preset specifications. If it does not meet the preset specifications, it sends a reminder instruction to the drone. The off-site drone is also used to play reminder audio according to the reminder instruction.

[0037] Specifically, off-site drones follow disinfection personnel to film the disinfection process; the edge server compares and analyzes the changes in color and brightness of the area before and after spraying the video images to identify water marks on the vehicle surface after spraying disinfectant, analyzes whether the water marks cover the required disinfection area of ​​the vehicle, and records the entire disinfection time. If the water marks cover the required disinfection area of ​​the vehicle and the disinfection time is greater than or equal to the preset standard disinfection time, then it meets the preset specifications.

[0038] The method in this embodiment also includes S7, which uses off-site video collected by off-site cameras to identify whether a vehicle transporting live pigs has arrived; if so, a disinfection monitoring instruction is generated. S8. The off-site drone receives the disinfection monitoring command, flies to the pig transport vehicle, and takes a video of the disinfection process until the disinfection is completed. S9. Analyze the video of the disinfection process to see if the disinfection meets the preset standards. If it does not meet the preset standards, send a reminder instruction to the drone. The drone outside the site will play a reminder audio according to the reminder instruction.

[0039] Frequent entry and exit of pig transport vehicles, many of which originate from external regions, pose a significant risk of introducing viruses and bacteria into farms, making them a potential source of major disease outbreaks. Therefore, thorough disinfection of vehicles before entry is the first line of defense in the disease prevention system. However, since disinfection is generally performed manually, issues arise such as inadequate adherence to procedures, skipping areas for convenience, and insufficient time commitment. Traditional cameras also often lack real-time comprehensive monitoring due to limited field of view. This embodiment further enhances the monitoring of this crucial disease prevention node—the disinfection of transport vehicles—by linking off-site cameras with off-site drones. Upon detecting the arrival of a pig transport vehicle, the drone, deployed on the roof, is dispatched to the work area. The drone continuously records video during the disinfection process and, based on instructions from an edge server, automatically follows the disinfection personnel, recording the entire process.

[0040] For example, if the analysis results show that the left side of the vehicle body was not completely sprayed, the system will immediately send a reminder command to the drone and play a voice prompt saying "Please re-spray the left side of the vehicle body", effectively reminding the disinfection personnel to re-spray and ensuring that the disinfection procedures are followed properly.

[0041] Example 3 The difference between this embodiment and Embodiment 1 is that, in this embodiment, the inspection time is during the day, and the off-site drone inspection video includes visible light images and audio; when the inspection time is at night, the off-site drone inspection video includes infrared images and audio; the daytime and nighttime times are determined according to the sunrise and sunset times of the location.

[0042] The edge server is also used to analyze the infrared images to determine the presence of animals. If animals are present, the server identifies and counts the species, and determines whether the number of animals of a preset species exceeds a threshold. If the threshold is exceeded, an animal anomaly alert is generated. In this embodiment, rats are primarily identified and counted. The threshold is set based on the environment and size of the pig farm. If the land outside the pig farm is hardened and the farm is small, a smaller threshold can be set; conversely, a larger threshold can be set to reduce alert sensitivity.

[0043] In traditional livestock farm management, rodent activity typically occurs at night, making them highly concealed, destructive, and prone to carrying diseases. Rats not only directly gnaw on feed and damage electrical wires or equipment, but more importantly, they are vectors for various diseases. In the past, rodent control relied mainly on manually setting rat traps or periodically administering pesticides, which was reactive and difficult to detect and intervene in a timely manner.

[0044] This embodiment achieves day-night adaptive video acquisition, enabling all-weather monitoring of the risk of small animal intrusion into pig farms, with a particularly enhanced nighttime rodent control capability. By setting different rodent thresholds, the sensitivity can be flexibly adjusted according to the farm's location (e.g., rural open fields or urban fringe), size, and infrastructure sealing, achieving dynamic monitoring tailored to local conditions. Compared to traditional surveillance cameras that only detect rodents when they reach the farm's edge, this embodiment expands the detection range and allows for earlier detection, facilitating timely application of pesticides by farm staff. This provides significantly better prevention than when rodents have already multiplied and invaded the farm's boundaries.

[0045] Example 4 The difference between this embodiment and Embodiment 3 is that, in order to further improve the accuracy of the system's early warning, reduce false alarms caused by the daily activities of farm workers, and enhance the safety protection for personnel entering the farm at night, this embodiment also includes: When the edge server detects abnormal pig noises based on the on-site video, it does not immediately generate an off-site inspection command. Instead, it first retrieves off-site video footage from cameras to identify whether any registered personnel or vehicles have arrived near the farm. In this embodiment, "nearby" refers to the identifiable area. Registered personnel or vehicles include, for example, farm workers and their vehicles.

[0046] If the edge server detects that a registered person or vehicle has arrived, and the current inspection time is during the day, it will determine that the abnormal noise is most likely due to the pigs' excitement and agitation caused by familiarity with the farm staff (e.g., they are about to be fed). The edge server will not generate an off-site inspection command, and the off-site drone will not take off, thus avoiding unnecessary false alarms during inspections.

[0047] If the current inspection time is at night, and the edge server has already generated an animal anomaly alert based on the nighttime infrared image inspection results (in this embodiment, it is to identify that the number of rodents exceeds the threshold), then if the edge server identifies a farmer walking by through the off-site camera, and the farmer is more than 30 meters away from the farm gate, the edge server generates an escort inspection command.

[0048] An off-site drone takes off upon receiving an escort and inspection command, flies to the vicinity of the pedestrian, switches to infrared image acquisition mode, and follows the farmer to the farm gate. During this process, the edge server continuously analyzes the infrared images transmitted by the drone to determine if snakes are present in the vicinity of the farmer (in this embodiment, the ground). If a snake is detected, the edge server sends an alert to the off-site drone; the drone then plays an audio alert, such as "Watch your step, snake spotted, please maintain a safe distance," to warn the farmer.

[0049] During testing, it was found that in some farms with small spaces or poor sound insulation, pigs, with their sensitive hearing, are easily excited by the arrival of farm workers, leading to false alarms, once they become familiar with the sounds of their footsteps or vehicles. This embodiment addresses this by using off-site cameras to identify registered personnel and vehicles before triggering drone inspections. This effectively filters out abnormal excited vocalizations in pigs caused by daily activities (especially daytime feeding), reducing the false alarm rate. In addressing this issue, this embodiment also fully considers the complex ecological environment of farms. As the ecosystem improves, feed and excrement in farms easily attract rodents, and the increased rodent population attracts their predators (such as snakes). At night, when abnormal rodent activity has been detected, the safety risks of encountering snakes during foot patrols increase. This solution utilizes drones to actively escort patrol personnel at night using infrared imaging. It can promptly detect snakes hiding in the dark and issue alarms, greatly ensuring the personal safety of farmers and increasing the system's visibility during their activities. It also enhances the farmers' experience and provides emotional value, making it easier to promote and apply.

[0050] The above are merely embodiments of the present invention. The invention is not limited to the fields covered by these embodiments. Commonly known structures and characteristics in the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are able to access all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A pig farm automatic inspection system, comprising an edge server and an in-field camera collecting video in the field; characterized in that, The field outside unmanned aerial vehicle is also included; The edge server is used for identifying abnormal state of live pigs according to the field inside video, and is also used for judging whether the number of live pigs in abnormal state exceeds a threshold value, and if the threshold value is exceeded, a field outside inspection instruction is generated; The field outside unmanned aerial vehicle is used for receiving the field outside inspection instruction of the edge server, performing a field outside inspection task according to a preset flight route, and sending the inspection video to the edge server; The edge server is also used for identifying whether there is an abnormal event according to the inspection video, and generating a reminder, including an external event reminder generated by identifying an abnormal event, or a live pig state abnormality reminder generated by not identifying an abnormal event.

2. The automatic inspection system for pig farm according to claim 1, characterized in that: The cloud server and the user terminal are also included; The edge server is also used for sending the field inside video, the inspection video, and the reminder to the cloud server; The cloud server is used for pushing the reminder to the user terminal; The cloud server is also used for receiving an automatic inspection plan created by the user terminal, and sending the automatic inspection plan to the edge server; 3. The automatic inspection system for pig farm according to claim 2, characterized in that: The automatic inspection plan includes an inspection route and an inspection time; The edge server is also used for controlling the field outside unmanned aerial vehicle to perform the field outside inspection task according to the inspection route at the inspection time. The field outside camera is also included, and the edge server is also used for identifying whether a live pig transport vehicle arrives according to field outside video collected by the field outside camera, and if the live pig transport vehicle arrives, a disinfection monitoring instruction is generated; 4. The automatic inspection system for pig farm according to claim 3, characterized in that: The field outside unmanned aerial vehicle is used for receiving the disinfection monitoring instruction of the edge server, flying to the live pig transport vehicle, shooting a video of a disinfection process, and stopping until the disinfection is completed; The edge server is also used for analyzing whether the disinfection conforms to a preset specification according to the video of the disinfection process, and if the disinfection does not conform to the preset specification, sending a reminder instruction to the unmanned aerial vehicle; The field outside unmanned aerial vehicle is also used for playing a reminder audio according to the reminder instruction. The abnormal state of the live pig includes abnormal call and abnormal action; and the abnormal event includes abnormal noise and abnormal smell.

5. The automatic inspection system for pig farm according to claim 4, characterized in that: The inspection time is daytime, and the inspection video of the field outside unmanned aerial vehicle includes a visible light image and an audio; the inspection time is nighttime, and the inspection video of the field outside unmanned aerial vehicle includes an infrared image and an audio; 6. The automatic inspection system for pig farm according to claim 5, characterized in that: The edge server is also used for analyzing whether there is an animal according to the infrared image, identifying an animal species and counting if there is an animal, judging whether the number of animals of a preset species exceeds a number threshold value, and generating an animal abnormality reminder if the number threshold value is exceeded. When the edge server identifies the abnormal call of the live pig according to the field inside video, the edge server first calls the field outside video collected by the field outside camera, and identifies whether a person or a vehicle recorded in a record arrives near the breeding farm; 7. The automatic inspection system for pig farm according to claim 6, characterized in that: If the edge server identifies that the recorded person or vehicle arrives, and the current inspection time is daytime, the field outside inspection instruction is not generated. When the edge server identifies that the recorded person arrives, if the current inspection time is nighttime, and the edge server has generated the animal abnormality reminder according to the nighttime infrared image inspection result, if the edge server identifies that the breeder is more than a preset safety distance away from the gate of the breeding farm through the field outside camera, the edge server generates an escort inspection instruction to the field outside unmanned aerial vehicle.

8. The automatic inspection system for pig farm according to claim 7, characterized in that: The following content is included:

9. A method for automatically inspecting a pig farm, characterized by, S1, deploying a field outside unmanned aerial vehicle in a live pig breeding farm; ​ S2, collect the video in the pig farm; S3, identify the abnormal state of live pigs according to the video in the field, judge whether the number of abnormal state of live pigs exceeds the threshold value, if it exceeds the threshold value, generate off-site inspection instruction; S4, the off-site unmanned aerial vehicle receives the off-site inspection instruction, according to the preset flight route, executes the off-site inspection task, and shoots the inspection video; S5, according to the inspection video, it is judged whether there is an abnormal event, and a reminder is generated, including an external event reminder generated by identifying an abnormal event, or a live pig state abnormality reminder generated by not identifying an abnormal event.

10. The method of claim 9, wherein the method further comprises: Also includes: S6, receive the automatic inspection plan of the user, the automatic inspection plan includes the inspection route and the inspection time; The off-site unmanned aerial vehicle executes the off-site inspection task according to the inspection route at the inspection time; S7, through the off-site video collected by the off-site camera, it is judged whether there is a live pig transport vehicle, if there is, a disinfection monitoring instruction is generated; S8, the off-site unmanned aerial vehicle receives the disinfection monitoring instruction, flies to the live pig transport vehicle, shoots the video of the disinfection process, and stops until the disinfection is completed; S9, according to the video analysis of the disinfection process, it is judged whether the disinfection conforms to the preset specification, if it does not conform to the preset specification, a reminder instruction is sent to the unmanned aerial vehicle; the off-site unmanned aerial vehicle plays the reminder audio according to the reminder instruction.

Images