(i) Overview of artificial intelligence in agriculture
1. Concept
Artificial intelligence technology (Artificial Intelligence) is a new technology science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. The research areas of artificial intelligence include robotics, language recognition, image recognition, natural language processing, and expert systems.
The so-called agricultural human intelligence is a field that uses this new technology to improve the quality and efficiency of agricultural production, and most of the smart agriculture that can often be heard now refers to the application of artificial intelligence in the agricultural field.
2. Composition
Because the language recognition, image recognition, etc. in artificial intelligence technology are mainly chip technology, the correlation for agricultural enterprises is very low, so the components of artificial intelligence are mainly introduced in the agricultural field of artificial intelligence products and applications. Roughly speaking, the current products of artificial intelligence in the agricultural field mainly include intelligent monitoring, intelligent agricultural drones, intelligent robots, and intelligent tractors.
1) Intelligent agricultural monitoring equipment
An important application of artificial intelligence in the agricultural field is intelligent monitoring, and the core of intelligent monitoring is machine learning algorithms and video analysis technology.
The intelligent monitoring here does not refer to a simple camera, but a monitoring system based on artificial intelligence and machine learning. Identify animal or human violations by getting a live video feed of each crop area and send alerts immediately and automatically.
The main application scenario of this artificial intelligence monitoring is the "less people and more land" area, that is, large farms, so it is widely used in Canada and the United States. In the United States, for example, the biggest damage to crops is animals, so they use artificial intelligence and machine learning to reduce the possibility of livestock and wildlife accidentally destroying or breaking into crop areas, effectively preventing them from stealing food from remote locations.
Driven by AI and machine learning algorithms, video analytics technology is rapidly advancing, allowing every agricultural participant to protect the perimeter of their fields and buildings. AI and machine learning video surveillance systems make large-scale agricultural operations as easy as individual farm operations.
Of course, this machine learning-based surveillance system can now be programmed or trained for business management, through artificial intelligence monitoring to identify employees and vehicles, as well as identify accidental intruders and then alarm them.
Relying on artificial intelligence and machine learning algorithms to identify people and vehicles is currently simplifying remote operations for agribusinesses across the globe.
2) Smart agricultural drones
In recent years, the development of UAVs has been very fast, and it has been widely used in various fields, and UAVs can often be seen in this year's Russian-Ukrainian war, which can be detected and bombed, very hot.
In addition to the military field, smart drones are also very many applications in the agricultural field, through the real-time sensor data and visual analysis data of drones to predict, farmers can increase crop yields, reduce the time and effort invested, and significantly obtain the maximum return on investment.
At present, there are many types of UAVs in the world, but there are roughly two classification methods used in the field of agriculture:
1) According to the power division: mainly according to the source of power used by the drone to classify, just like the current car market divided into gasoline vehicles and electric vehicles, drones can also be divided into oil-powered drones (engines as power units), electric drones (motors as power units), but now, cars still burn more oil, but most of the drones use batteries.
2) According to the structure of the model: it can be divided into fixed-wing UAV, single-rotor UAV, and multi-rotor UAV.
Fixed-wing aircraft are mainly used in farmland information collection and farmland remote sensing, and have the characteristics of large capacity, fast flight speed and high operational efficiency. The operation generally adopts ultra-low altitude flight, 5-7 meters away from the crop canopy, which has high requirements for the terrain of the operation area, and is generally widely used in open farms.
The volume and load of single-rotor and multi-rotor UAVs are relatively small, flexible operation, high operating efficiency, suitable for operations in more scattered farmland blocks, compared with the agricultural characteristics of the mainland, the practicality of this UAV is high.
And there is another point, the price of fixed-wing aircraft is high, and the operation also requires professionals, that is, it needs to be licensed, and the entry threshold for rotary wing drones is very low.
3) Intelligent agricultural robots
Today's shortage of agricultural workers, especially in remote areas where agricultural workers are difficult to find, intelligent agricultural robots and other intelligent machinery based on AI and machine learning have become a viable option for agricultural cultivation. Today's agricultural robots are mainly used for crop cultivation, crop picking and crop processing.
Large agribusinesses couldn't find enough employees and turned to robotics to manage hundreds of acres of crops. In addition, artificial intelligence technology can also provide security for the perimeter of remote areas. By programming autonomous robotic devices, smart agricultural robots can use an array of sensors and artificial intelligence to analyze the location and maturity of crops, observe their surroundings in a 3D environment using intelligent motion sensing technology, and they are able to spread fertilizer, pick, carry and more for each row of crops. As a result, operating costs can be reduced and yields and efficiency further increased. Agricultural robots are rapidly becoming more sophisticated, allowing a machine to determine the best route to pick vines, leaves and other immature crops to pick its target.
Agricultural robots have proven to be able to capture valuable data that can be used to fine-tune artificial intelligence and machine learning algorithms to further improve crop yields.
4) Smart tractor
An important product of artificial intelligence in the field of agriculture is the intelligent tractor. This intelligent tractor mainly uses VRT (variable rate technology), which can use artificial intelligence recognition technology to distinguish crops and weeds, and then carry out weeding, fertilization and other work, intelligent tractor through machine learning to amplify the machine's own advantages: precision.
The smart tractor can take pictures of crops at a speed of 5,000 plants per minute as it drives through a field, and use algorithms and machine vision to determine whether each plant is a crop or a weed. The image processing chip can recognize a photo in just 0.02 seconds, with an accuracy of 1/4 inch, and can accurately identify each weed and spray them with pesticides during operation.
Smart tractors can usually achieve the following functions:
1) Accurate navigation, providing the best ridge excavation navigation path, maximizing the use of light and heat resources, advanced automatic driving system can improve navigation accuracy in complex terrain and environment, and reduce the problem of agricultural tool deviation;
2) Operation memory and sharing of multiple agricultural machinery journey information to avoid repeated operations or omissions;
3) Autonomous driving, providing high-precision positioning, automatic steering, automatic navigation, repeated control, etc.;
4) Automatic spray bar control device, which can improve the accuracy of seed and fertilizer delivery.
(2) Characteristics of agricultural artificial intelligence
Specifically, agricultural artificial intelligence has the following four characteristics:
1. Permeability characteristics
As a universal, basic and enabling digital technology, artificial intelligence has the potential to integrate with all links in the agricultural field, that is, artificial intelligence can be used regardless of agricultural production, circulation, wholesale, and retail, and this characteristic that can be widely used in various fields of agriculture is defined as the penetration of universal technology.
2. Synergy characteristics
In the field of agricultural production, the application of artificial intelligence can improve the matching degree between capital, labor, technology and other elements, strengthen the collaboration between upstream technology research and development, midstream engineering implementation, downstream application feedback and other production links, so as to improve operational efficiency. In the field of agricultural product consumption, artificial intelligence can realize the automatic portrait of users' consumption habits and consumption demand, complete the intelligent matching of personalized demand and professional supply, and further release consumption potential. In general, the synergistic characteristics of artificial intelligence are reflected in the improvement of economic operation efficiency.
3. Substitution characteristics
Agricultural artificial intelligence can realize the direct replacement of agricultural personnel. From simple work to complex work, agricultural artificial intelligence will continue to play a substitution effect, while accumulating as an independent factor, it can replace other capital factors and labor factors, and its supporting role in economic development will continue to strengthen.
4. Innovative features
Production automation can replace some high-intensity and difficult continuous labor, and artificial intelligence raises concerns about employment prospects because it can replace human mental work and creative activities. For example, agricultural artificial robots are not only replacing human labor, but can also analyze and process complex agricultural labor.
(3) The development history of artificial intelligence
The concept of artificial intelligence first appeared in the 50s of the last century, and it has been more than 70 years now. According to the development process of artificial intelligence, it can be roughly divided into six stages:
1. The first stage: the initial development period (50s - early 60s of the 20th century)
In the 50s, the concept of artificial intelligence began to appear. 1950 An American named Claude Elwood Shannon published a paper in Scientific American called "The Method of Time-Human-Machine Game", which is considered to be the starting point for the development of artificial intelligence technology.
In 1954, Dr. Turing, who used machines to crack German codes in World War II, proposed the famous "Turing test", the core of which is that if a third party cannot distinguish the difference between humans and artificial intelligence machines, the machine can be judged to have artificial intelligence. For the first time, machines and intelligence were combined, so Turing is also known as the father of artificial intelligence.
Then in 1956, a group of scientists held a seminar at Dartmouth, at which the concept of artificial intelligence was formally proposed and a special discipline was established, so many official documents now designate 1956 as the year of the emergence of artificial intelligence.
In 1957, a man named Frank Rosenblatt simulated a neural network model called a perceptron on a computer, which simply means that the machine learns things on its own.
At that time, after the concept of artificial intelligence was proposed, the market was excited, so there were many research results around this concept, among which the more famous is the program of using machines to play checkers, as well as machine theorem proofs, etc., which is the first climax of artificial intelligence development.
2. Stage 2: Rethinking the Development Period (60s-early 70s of the 20th century)
Some of the first advances in artificial intelligence made people at the time very excited, thinking that since machines could play checkers, they could do more.
At that time, the United States was not fighting the Vietnam War, many American soldiers were tossed in the jungle with psychological diseases, American psychologists were not enough, so in 1966 MIT developed a "human-computer dialogue" program, that is, a software program called ELIZA was installed in the computer, the main purpose is to imitate psychologists in clinical treatment, which is the earliest chatbot.
However, the effect of these studies is very poor, the chat program basically can't understand anything, the successive failures of expert systems and the failure of expected goals have made the development of artificial intelligence enter a trough in this period.
3. The third stage: the period of application development in the professional field (early 70s to mid-90s of the 20th century)
In the 70s of the 20th century, an important part of artificial intelligence appeared: "expert systems".
The earliest expert systems used computers to help chemists infer molecular structures based on mass spectrometry data, with the main purpose of studying drugs. This thing is a breakthrough in artificial intelligence from theory to practical application, because all technology is like this, if there is no practical application, there is no future at all.
On the theoretical front, there has also been a lot of progress during this time.
In 1975, Marvin Minsky proposed a framework theory for "knowledge representation" in artificial intelligence, which is actually mainly aimed at people's mental models when understanding things or events.
1976 Heuristic search appeared. At the time, it was mainly used in mathematical research, but later search engines adopted this technique. In the same year, David Marr proposed the concepts of computer vision and computer neurology, which means that he began to study computers as he studied humans.
4. The fourth stage: the period of steady development (mid-90s of the 20th century - 2010)
In the late 90s of the 20th century, because of the collapse of the Soviet Union, the original military information system of the United States was converted to civilian use, which caused the great development of network technology at that time, and many famous Internet companies appeared at that time. Because the cost of these systems was borne by the government, Internet companies at that time were particularly profitable. With money, it has accelerated the innovative research of artificial intelligence, and the development of artificial intelligence has also begun to develop from the professional field to the civilian field.
The most famous AI event of that time period was IBM's Deep Blue supercomputer defeating chess world champion Kasparov in the New York Chess Tournament in 1997. A total of 6 rounds were played, and in the end, the machine won 3 wins and 1 draw, and "Deep Blue" became the first computer to defeat the chess world champion.
In this environment of hot events, in 1998 the semantic web appeared, which simply means that computers can understand the language, that is, data. In 2003, the LDA theme model appeared, that is, computers can analyze large documents and extract them for people to use, similar to computer secretaries.
In 2005, when too many people died in Afghanistan in the United States, the government was burned out, so it paid Boston Dynamics to develop robots that can replace soldiers to work, this quadruped robot is called "big dog" (see figure below), in the mountains of Afghanistan can travel more than 20 kilometers, carrying 170 kilograms, which was very sensational at the time, which can be regarded as opening a new era of robot research.
Big dog robot
In 2008, IBM proposed the concept of "smart earth", which is to use artificial intelligence technology to transform the earth, and now the concepts of "smart city", "smart transportation" and "smart agricultural approval" are extended from the "smart earth". The "wisdom" refers to artificial intelligence.
5. Phase V: Booming period: 2010 to present.
During this period, the Internet bubble has burst, so the capital that Europe and the United States ran out of the Internet has entered these fields such as artificial intelligence and big data, so both concepts and applications have developed rapidly during this period.
To make a digression, at that time, the mainland's capital mainly ran from the Internet to real estate, so it developed poorly in artificial intelligence, because the core of artificial intelligence is "chips", that is, things that have been "stuck in the neck" now. But at that time, those who were able to enter the field of artificial intelligence are now very profitable, and the best example is the DJI drone.
Returning to the development of artificial intelligence, the largest artificial intelligence application in the civilian field of unmanned driving began at that time. On October 9, 2010, Google announced its self-driving car plan, and in 2012 it also obtained a legal license, but in 2016, when the self-driving car and the truck collided, Google separated the driverless from Google and renamed it Waymo.
In addition to driverless technology, artificial intelligence in 2010 was represented by IBM Watson's challenge to Jeopardy, the strongest in history.
Watson is a question-and-answer computer system that can answer questions raised by natural language, and can also analyze people's emotions, tone and whatever, and the final result is that the computer wins.
To insert here, IBM has not been well-known over the years, but it has always attached great importance to scientific research, and there are constantly scientific and technological innovations, which is also a reason why after the US interest rate hike in May 2022, stocks such as Apple and Amazon have plummeted, but IBM's stock is still strong.
By 2015, a bunch of Silicon Valley tycoons had joined forces to establish an artificial intelligence nonprofit organization, OpenAI, led by Musk (who quit in 2017). The goal of this organization is to research and build "universal" bots and chatbots that use natural language. Later, in 2019, Microsoft also participated and invested $1 billion.
The most famous research result of OpenAI is the GPT series, which is a very powerful language generation model. What does it mean, that is, computers can help people write documents? For example, once a student in California used this model to write an article about productivity, and finally won an award, but no one knew that this article was actually written by AI intelligence, and people's job is to give a title and direction! So this thing is now very controversial, and everyone dares not apply it on a large scale.
At that time, there was another world-renowned event in the field of artificial intelligence, that is, the Sophia robot.
In 2016, Hansen Robotics launched Sophia, a human robot that closely resembles humans. This robot not only looks like a human, but can also chat like a human. It was even granted citizenship by Saudi Arabia in 2017. In 2018, it was also promoted in China's CCTV "Dialogue" column, and this robot is like the following.
In fact, these were not too unusual at that time, when the world caused a sensation in a TV show, when the host asked the robot "Hey, Sophia, do you want to destroy humanity?" Sophia responded without thinking, "Yes, I will destroy humanity." ”！
It is this sentence that scares people enough, and many later film and television works about robot rebellion originated from this Sophia.
It is also because of this that this Sophia robot has been unable to be mass-produced because of boycotts, until the end of 2021, because of the epidemic, when the communication between people was restricted, Hansen began to mass-produce the Sophia robot, but it has not yet been produced and the price has not been announced.
It can be seen from the above that the technological development of artificial intelligence is mainly in Europe and the United States, and the artificial intelligence technology of the mainland has developed after 2010. Although it is very hot now, but the mainland's artificial intelligence is mainly concentrated at the application level, and the research on the underlying technology is relatively backward.
The better thing to do in artificial intelligence is face recognition brought about by electronic payment, this technology is relatively leading, and there is drones, which was mentioned earlier.
However, after 2018, due to the trade war and science and technology station launched by the United States, the "neck" of science and technology has gradually increased, so in recent years, the mainland's attention to artificial intelligence has gradually increased.
In 2019, artificial intelligence was officially included in the list of newly approved undergraduate majors in the mainland, in order to increase the number of professionals in artificial intelligence technology.
In March 2020, in the context of the pandemic, the "new infrastructure" launched by the government also included artificial intelligence infrastructure. On August 5, 2020, the Standardization Administration of China, the Cyberspace Administration of China, the National Development and Reform Commission, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, and five departments jointly issued the Guidelines for the Construction of the National New Generation Artificial Intelligence Standard System. The guide puts forward specific ideas and construction contents for the construction of the national new generation of artificial intelligence standard system, and attaches a detailed list of the development direction of artificial intelligence standards, further standardizes the application system of artificial intelligence at the national level, and clarifies its development direction.
In the field of agriculture, the application of artificial intelligence in mainland China is relatively late.
Generally speaking, the mainland's agricultural science and technology are started with the government as the main body, and it was not until July 2017, when the State Council issued the "New Generation of Artificial Intelligence Development Plan" that artificial intelligence at the market level, some enterprises have carried out intelligent agriculture attempts, and positive progress has been made in the application of some fields.
In terms of artificial intelligence equipment, because China is a big manufacturing country, the development of equipment is generally good, so intelligent agricultural machinery equipment is more prominent in the agricultural field.
In 2016, after the Sino-US friction began, the mainland proposed the concept of "intelligent manufacturing in China", and China was not a stock market crash that year, and the government hoped that capital would be transferred to rural areas, which led to the development of intelligent agricultural machinery and equipment.
However, the demand for capital for intelligent agricultural machinery equipment is very large, so it is better to do it or state-owned enterprises, such as Lovol Heavy Industry, Zoomlion, etc., the agricultural machinery equipment developed is mainly tractor automatic driving system and precision level system, which are mainly used in large-scale planting areas.
(4) Application fields of agricultural artificial intelligence
The application of artificial intelligence technology in the field of mainland agriculture is mainly the use of intelligent agricultural machinery and equipment research and development for pest identification, plant protection drone upgrade, non-destructive testing of agricultural products, etc.
These include intelligent robots such as farming, sowing and picking, tractors that can automatically identify and kill sick plants, and AI phones that can automatically notify farmers of crop diseases. There are also intelligent identification systems such as intelligent soil detection, detection of pests and diseases, climate disaster warning, satellites that automatically detect traces of drought, and smart wearable products for livestock used in the livestock industry.
1. Useful for soil measurement and analysis
Before planting crops, farmers should ensure that the soil is fertile. The data they collect from soil samples can give them in-depth information about how much fertilizer they need, which crops are growing better, and how much water they need.
However, manual monitoring, collection, and analysis of soil samples is simply not an ideal option. Therefore, drones can effectively take on this work, and agricultural drones equipped with high-definition cameras and advanced sensors, capable of creating accurate three-dimensional maps of plots and soil analysis, and making detailed planning for planting, play a crucial role in the initial stage of agricultural production. At the same time, agricultural drones in this category can also closely monitor the condition of the plot, providing powerful data support for later irrigation and soil nitrogen content management.
Agricultural drone mapping soil information
2. Helps with crop sowing
Through GPS and RTK technology, the UAV seeding system can locate the field sowing position to the centimeter level and the sowing amount is accurate. The system simultaneously injects the seed into the soil with all the nutrients it needs to grow. This technology increases the absorption of nutrients from seeds by 75% and reduces sowing costs by 85%. It not only ensures the survival rate of seeds, but also relatively reduces the consumption of other resources, such as the use of drones to sow rice, which reduces the breeding and planting links, saving time, labor and effort.
Agricultural seeding drones
3. Contributes to crop growth monitoring and management
For agricultural products, after sowing, the most critical step before harvest is the need to monitor the crop regularly. But large land areas and inefficient crop monitoring have always been another major pain point in agricultural production. As global temperatures rise and climate change becomes more volatile, managing and maintaining land conditions becomes more difficult.
Except in cases where there is very little land, it is practically impossible to manually monitor the health of crops. In the past, farmers mainly relied on satellite imagery to assist in crop monitoring, but this was too expensive and required to be pre-purchased, and the efficiency of the daily average service was difficult to provide real-time and accurate crop monitoring information.
As a result, agricultural drones and tractors equipped with real-time imaging systems have emerged, and frequent inspections can help farmers obtain real-time information on water scarcity, crop diseases and moisture levels. It provides a reliable and effective guarantee for farmers to monitor crop growth more accurately and adopt timely and effective prevention and control measures. At the same time, the vast amount of data captured by smart sensors and drones is combined into a live video stream, giving agricultural experts an entirely new set of data they have never had access to before.
In the middle of agricultural production, using smart sensors in combination with the drone's visual data stream, agricultural AI applications can now detect areas with the most severe pests and diseases in the planting area. Supervised machine learning algorithms can then be used to determine the optimal mix of pesticides that can reduce the threat, spread and infect healthy crops. Optimizing the right mix of pesticides and fertilizers and applying them only to field areas where treatment is needed to reduce costs and increase yields is one of the most common uses of AI and machine learning in agriculture today. The data shows that with smart tractors, farmers can reduce their pesticide use by 90%.
The right amount of fertilizers and pesticides is essential for proper crop growth. Current fertilization methods are the use of tractors or manual spraying of fertilizers and pesticides. However, it is impossible for tractors to reach every corner of the field, and manual work is too expensive. In addition, we are not sure whether humans can do their jobs correctly.
Therefore, agricultural drones equipped with ultrasound, radar, and advanced spraying systems are widely used in fertilizer and pesticide spraying operations to help farmers spray the right amount of pesticide or fertilizer. Also commonly known as plant protection drones. This type of UAV can continuously adjust the flight altitude according to the geographical terrain, independently identify and avoid obstacles, and rely on the advanced spraying system to implement accurate and uniform spraying operations on crops. This technology greatly reduces the amount of pesticides and well reduces the pollution of water and the environment by plant protection operations.
Another key benefit of crop spraying drones is that they can be operated without anyone, reducing cost, time and effort. According to expert estimates, compared with traditional machinery, the operation efficiency of plant protection UAVs is as much as 5 times, which will greatly reduce the consumption of manpower and material resources.
Plant protection drones spraying pesticides
Airborne or satellite remote sensing has not been successfully applied to directly identify and locate insects. By combining drone thermal imaging camera data with sensors that monitor the relative health of plants, agricultural management teams can predict and identify pests before they occur with the help of AI, so that pests can be detected in time for chemical treatment, which is more cost-effective.
With the help of artificial intelligence analysis, farms can dispose of infected leaves in a timely manner, reducing losses as early as possible. In the past, the detection of pests and diseases required manual inspections, and if they were not detected in time, they would lead to the death of large areas of crops. Manual inspections are time-consuming, laborious, and can be overlooked. The introduction of artificial intelligence can provide continuous monitoring and forecasting, reducing losses caused by pests and diseases.
The prediction of pests and diseases mainly uses deep learning technology in artificial intelligence
Deep learning is a powerful algorithm in which programmers no longer explicitly tell a computer a target, but instead train it to recognize several patterns. The way to do this is to provide the computer with pictures of plant leaves labeled healthy and diseased. From these photos, it can learn how to distinguish healthy and diseased leaves, and then determine whether other leaves are healthy. Through large-scale crop pest image training, the types of pests and diseases in crop pictures can be automatically identified.
At present, the algorithm has realized the identification of hundreds of diseases and pests of dozens of crops such as apples, potatoes, and peanuts. The algorithm can help crop growers monitor crop disease status, and quickly, conveniently and accurately determine the type of disease. Unclear diseases can also be initially identified and coping strategies can be provided to further increase crop yields.
4. Contribute to crop health assessment
Assessing the health of crops and detecting pests and diseases in a timely manner is one of the necessary guarantees for quality agricultural production. Equipped with visual infrared and near-infrared emitting equipment, agricultural drones can accurately analyze the amount of green and near-infrared light reflected by crops and create multispectral images to track changes and health of crops. Farmers can make rapid and effective control measures based on the corresponding data to ensure that crops are free from diseases and pests, thereby significantly improving the efficiency of agricultural production.
Agricultural drones equipped with hyperspectral, multispectral, and thermal sensors can accurately analyze and identify arid areas of land parcels, providing powerful data support for precision irrigation operations. In addition, during the growth of crops, this type of drone can also calculate the vegetation index, analyze crop density and health status through the heat emitted by the crops themselves.
Agricultural smart drones crop health monitoring
The application of drones in the agricultural field will be more and more, from the current situation, the application of drones is mainly in the production link, in the transportation and retail of agricultural products, although there are also SF, Jingdong these enterprises have tried, but in general, still stay in the level of playing the concept, and have not formed a real business application.
In addition, the application of drones in the agricultural field still has problems such as network security problems, high costs, and the need for professionals to operate. However, over time, when all the current issues surrounding drones are fully resolved, drones will be more widely adopted in agriculture.
5. Helps ensure the health of livestock
AI can be used to monitor the health of livestock, including vital signs, daily activity levels, and food intake. Ensuring the health of livestock is currently one of the fastest growing aspects of artificial intelligence and machine learning in agriculture.
It is important to understand how each animal reacts to diet and boarding conditions in order to optimally handle them in the long term. Using AI and machine learning, agricultural experts can know what determines the emotions of cows and adjust them appropriately to increase milk production. For many farms with cattle and other livestock themes, this emerging technology opens up entirely new horizons on how farms can improve profits.
6. Helpful for sorting and processing of agricultural products
Another important application area of artificial intelligence technology in the field of agriculture is quality inspection.
The quality inspection of agricultural products includes agricultural product processing, quality control and component analysis, which is an important link in the circulation and consumption of agricultural products and an important step to ensure the safety of consumer consumption.
The traditional quality inspection of agricultural products mainly relies on manual means, which is not only inefficient, but also affected by human subjective and objective factors, and the accuracy and stability of the detection results are poor.
Using machine vision and artificial neural networks in artificial intelligence, the quality and quality of agricultural products can be accurately and quickly detected, which not only saves manpower, but also greatly improves work efficiency and detection accuracy. The practice of using artificial neural networks for the detection and application of agricultural products in mainland China has also made progress, and the detection objects mainly include fruits, tea, cotton, poultry and meat products, and the detection content includes the size, shape, texture, color, visual defects, etc. of agricultural products.
In terms of quality inspection, it has a relatively large correlation with the agricultural batch market. In recent years, the transformation of the agricultural batch market wants to improve the sorting and processing capacity, but most of the current use is still manual sorting and processing methods, and a very small number of agricultural batch markets only use mechanical fruit pickers such as a single testing equipment, including our new Shenyang sorting and processing project is also using this mechanical equipment.
The sorting and processing system based on artificial intelligence, in fact, mainly uses machine vision technology in artificial intelligence, with this technology can be sorted without contact with agricultural products, we all know that fresh food is very afraid of contact, easy to break, and for the non-standard shape of fruits and vegetables can also be detected and sorted.
(5) Agricultural artificial intelligence industry
1. Industry scale
u Chinese industrial intelligence industry scale: 404.1 billion yuan (2021 data)
u Agricultural artificial intelligence industry scale: 6.013 billion yuan (Lao Feng calculation data)
According to reports, in 2021, the scale of the mainland's artificial intelligence industry reached 404.1 billion yuan.
In terms of agricultural artificial intelligence, Lao Feng did not find the industrial scale data in 2021, but could only find that the scale of the artificial intelligence industry in the mainland agricultural field in 2016 was 432.2 million US dollars, which was converted into RMB 3.0254 billion at the 1:7 exchange rate, and then according to the compound annual growth rate of 13.57% reported by the Ministry of Agriculture and Rural Affairs, it can be roughly calculated that the scale of the mainland's agricultural artificial intelligence industry in 2021 is 6.013 billion. According to this scale, the industrial scale of agricultural artificial intelligence accounts for 1.49% of the overall artificial intelligence industry in the mainland. (However, this data should be higher than the actual industry scale, because an international think tank report shows that the global agricultural artificial intelligence market size in 2021 is only 6.672 billion yuan, and the compound annual growth rate (CAGR) is expected to be 16.4% during 2022-2028)
2. The situation of the industrial chain
The artificial intelligence industry chain is basically divided into three levels: basic layer, technology layer, and application layer.
The basic layer is the foundation of the artificial intelligence industry, mainly research and development of hardware and software. It mainly includes computing hardware (AI chips), computing system technology (big data, cloud computing and 5G communication) and data (data collection, annotation and analysis) to provide data and computing power support for artificial intelligence;
The technical layer is the core of the artificial intelligence industry, and the technical path is constructed from the starting point of simulating the intelligence-related characteristics of humans;
It mainly includes algorithm theory (machine learning), development platform (basic open source framework, technology open platform) and application technology (computer vision, machine vision, intelligent speech, natural language understanding).
The application layer is an extension of the artificial intelligence industry, integrating one or more types of artificial intelligence basic application technologies to form software and hardware products or solutions for specific application scenarios. Including robots, smart healthcare, smart transportation, smart finance, smart home, smart education, wearable devices, security and other aspects.
Chinese the structure and value of the intelligent industrial chain
At present, the gross profit margin of infrastructure enterprises in the basic support layer is moderate, and the gross profit margin of data service enterprises is relatively high, maintaining the overall level of more than 44%. The gross profit margin of applied technology enterprises in the technical layer is moderate, and the overall gross profit margin of algorithm theory and platform framework enterprises is high, at a level of more than 23%; The gross profit margin of application-layer enterprises is relatively high, and the gross profit margin of solution enterprises and product service enterprises remains above 35%.
(6) There are problems with artificial intelligence in agriculture
1. Lack of AI talent
Lack of AI technicians. According to the comparison of talents in algorithm research in China and the United States by Oxford University in 2018 with advanced countries in the world, China's current talents in algorithm research account for only 13.1% of the world's underlying artificial intelligence technology research, while the proportion of algorithm talents in the United States is 26.2%. In terms of the number of artificial intelligence professional colleges, there are less than 30 university research laboratories in China focused on artificial intelligence, which is far from meeting the employment needs of artificial intelligence enterprises, and the talent who masters both artificial intelligence technology and agricultural production technology is even rarer
Lack of people using artificial intelligence. The current agricultural workers on the mainland are basically farmers in their forties and fifties, most of whom have not received higher education and have little reserve of knowledge. According to the survey, the average education level of more than 800 million peasants on the mainland has reached less than seven years, and of the 490 million agricultural laborers, only 13 percent have received high school education or above, less than 49 percent have received junior high school education, and the remaining 38 percent have only received primary education, or even none at all. At present, only a small number of talents have received professional agricultural knowledge training, and most agricultural workers cannot adapt to high and new technologies, and it is difficult to use artificial intelligence to develop smart agriculture.
2. The smallholder economic model is not conducive to AI applications
Agricultural scale is the key condition for the development of modern artificial intelligence, the current domestic agriculture in the development, household production contract responsibility system is still the main model, most farmers due to personal in technology, experience and other aspects of the deficiencies, the development of artificial intelligence to show obvious understanding of the problem in place, this model for the improvement of their own economic effect of the role can not be accurately appreciated, and even there is that the development of artificial intelligence may lead to their own loss of original economic income. In addition, the conditions affecting artificial intelligence are also farmers' disagreement, and it is difficult to unite.
At this stage, agriculture is still the most labor-intensive industry in the mainland, but the overall available land in China is relatively scattered, which leads to the very obvious dispersion of domestic labor in the process of use. At this stage, the scale of domestic agricultural development is obviously affected by the low level of agricultural production, with the continuous expansion of the scope of application of artificial intelligence in China, the scale of agricultural production is required to be higher and higher, otherwise it will restrict the rapid development of artificial intelligence, and economic development will inevitably slow down.
3. It is difficult to collect information resources in the field of agriculture
At this stage, there are many websites that can effectively query agricultural information in China, but the managers behind these websites are the government, and their role is to provide various types of agricultural information, and the content of the website section is opened in accordance with the requirements of the superior, most of which are policy and repetitive content. There is also some agricultural information that is a small-scale survey conducted by commercial enterprises that are the suppliers of agricultural information, and the acquisition of such information requires high costs and is not suitable for ordinary people. These very scattered sources of official and unofficial agricultural information come from a wide range of sources, with very different collection environments, messy structures, and lack of unified standard data, which is not conducive to the development of artificial intelligence.
4. Insufficient commercialization of AI in agriculture
At this stage, in the process of smart agricultural technology development, the problem of lagging commercial development is prominent, which also leads to the fact that various types of artificial intelligence agricultural technologies urgently needed by farmers cannot be well satisfied.
Because artificial intelligence products require large investment and large cycles, they now need government guidance in most cases, without sufficient capital investment, without self-sufficient long-term planning, and without the strategy of using surplus valuable products to enable capital to be circularly developed.
In general, the agricultural application of mainland artificial intelligence technology is relatively rudimentary, not extensive and deep.
The above is some of the contents of mainland agricultural artificial intelligence, welcome to the axe, thank you