Right now, as the temperature rises, strawberries, which are the absolute mainstream of the market, will be unveiled and will be seen again in the autumn and winter. And in shanghai, several saber plants are exploring strawberry freedom throughout the year. While traditional large-shed strawberry cultivation relies mostly on the experience of senior fruit farmers, plant plants can generate crop adjustment programmes based on ai algorithms that allow each strawberries to enjoy customized scientific care and grow steadily over four seasons。
What's the smell of strawberries grown with al? What else would the cyber plant bring to china's agricultural future
In the open area of shanghai's agricultural valley, four 50 square metres of “containers” are not left in caves — one of the strawberries in a stereotee is growing in a fully artificially lighted environment, some of which have produced fresh red or pink ripe fruit. This is at the final stage of the 5th multi-agricultural science and technology competition, where four strawberry plants, consisting of teams from universities, scientific institutions and agricultural science and technology enterprises, are produced. Over the next six months, teams will be engaged in the ultimate fight with ai's lifeline, challenging high-yielding, high-quality and low-energy-consumption domestic strawberries。
How do i grow strawberries with al? The "berry cube" team at shanghai university of transport has developed an image recognition smart diagnostic system. An example of the problem of “leaf-colour inequality” in strawberry strains in the second class of the shanghai university of transportation, school of agriculture and biology, may be caused by a variety of causes, often involving multiple attempts to identify the problem, each of which will take a week or even longer to remove the jamming factor from a wide range of variables, such as air moisture, ambient temperature and water fertility. The development of the "ai-view prescription" system is about rapid diagnosis and treatment of strawberry disorders in a more efficient manner。
He shiwei: currently, agricultural production is based on eye-to-eye diagnosis of disease. The pathological system has precedents, but it is more a “expert system” where experts judge through images and give feedback on the next concrete operation. And the purpose of our system was to have a precise diagnosis of strawberries inside plants, which is a digital expression. If strawberries are missing, it is difficult to see through the surface, and we want to use the ai tool to diagnose images in real time, using a number of condensed neural network models, and, depending on this data, we can do well with the later planting and environmental strategy。
Specifically, the "ai-view opener" means that by capturing the growth state of strawberry strains in real time by cameras, and automatically comparing systems that are sent to in-depth learning, it is possible to calculate the nitrogen, potassium, etc. Content of the plant in an instant, and to rapidly generate targeted planting adjustments. The team member, a student at the college of agriculture and biology, shanghai transport university, studied the example of the darkness of the strawberries' leaves colour that we see in the system, which can be a precise colour, colour and prescription。
Lu ming: we see plants as "deep green" and "light green," but the "rgb" colour system has a gradient range of 0-255 on the computer. For example, “r” is red, which is 255, while “g” (green) and “b” (blue)” are 0, which is shown as pure red. The same is true for green, where one can see only a little greener, or less green, but if the picture is passed on to a computer, there is a standard. We judge the amount of nitrogen in the leaves by color to guide us to the next fertilization。
If traditional cultivation is “a flower that depends on the weight of the crop”, then the ai strawberries are “what is missing” and can help the average grower to deal effectively with the pain of over-fertilization or delayed diagnosis. From decision-making to implementation, this “ai-view-based” system not only provides evidence for cultivation, but also directly affects the nutrient ratio and drip irrigation of strawberry strains, with automatic process adjustments。
After four months of data collection, the “small model” developed by the berry cube team on its own is currently over 90 per cent diagnostic accuracy. Based on ai algorithms, team members have set the target for the sown "powder" strawberries: single production exceeding 0. 4 kg and sweetness stabilizing above 12 degrees. The team guidance teacher and a professor at the college of agriculture and biology at shanghai transport university, chang liying, presented their desire to explore the depth of the combination of artificial intelligence and agricultural technology at the saber plant。
Chang liing: strawberry has a particularly long reproductive period, and in this process we need to acquire environmental data from this scene, crop growth surface images, plant images, fruit images to further train our models. This is equivalent to an intelligent body, which, in order to keep these data sets inside, can make decision-making management more accurate in such contexts and varieties. Next, we are going to challenge shanghai's highest temperatures in july and august, with these algorithms for crop growth management that can help with strawberry production at high summer temperatures。
Strawberry yields continue to be compared to tastes in the "sebo plant" in shanghai agricultural science valley. In the town of white crane, in the qingura region of shanghai, more than 60 kilometres away, a 345-square square metre “twisted strawberry plant” was formally introduced into the operational phase, the first strawberries from mid-may to be marketed, and the year-round supply of these “smart strawberries” from ai will be brought to the popular table for the first time。
The vice-researcher of the forest fruit institute of the shanghai academy of agricultural sciences informed the chinese news reporter-in-chief that the research team had previously placed unmanned surveying robots in the strawberry plant for experimental use, that they had multi-dimensional collection of plant data, and that they had helped scientists to construct the crop maturity recognition model for strawberry, the plant form growth model and the environmental response model, and that these “submodels” had become important tools to guide the production of the “twilight strawberry plant”。
Square: like a fruit maturity recognition model, confidence is around 85 per cent. With this model, we'll be able to automate all the robots in the future. The robot also needs to be implanted into this model, which, once it has identified and made its decision, determines that the fruit can be harvested and can automatically operate its arm。
White crane's "twilight strawberry plant"
Equivalent revealed that the harder-to-build diagnostic models of physical health, pest identification models and product quality prediction models were still being developed and that, in the future, when these “submodels” were integrated, a large multi-data-driven strawberry growth model would enter production practice from the research site. Of course, with the exception of strawberries, fruits and vegetables of different kinds can follow the same logic towards intelligent production。
Zero: we need regular training to guide regular production. Large growth models are designed for plant plants, i. E. For relatively manageable environments in facilities, and can then be further expanded in this area if all environmental conditions in smart greenhouses are more precise, with applications gradually being extended to connected glass greenhouses and then to linked film greenhouses。
According to yang, the chief scientist and researcher of the institute of urban agriculture of the chinese academy of agricultural sciences, agricultural cultivation ranges from feeling to looking at data, and ai is transforming agricultural production, as is being done at the cyber botanical plant。
Chief yang: agriculture has two production methods, like medical chinese and western medicine. Some of the country's agricultural production patterns are now known as “suspecting” and then deciding how to diagnose and treat them. The future direction of agricultural development will be judged by human vision, the brain, and ultimately by the refinement of some data and models, allowing computers and plants to “dialogue” and lead to intelligent decision-making and control。
This year's central document emphasizes the need to develop new quality agricultural productivity and promote the integration of artificial intelligence with agricultural development, adapted to local conditions. Chief yang told the china voice reporter at the station that in the area of facilities agriculture, ai would not only be productive, but would also create more surprises for consumers。
Chief yang: the first is that the entire production process of the product is safe; the second is also likely to be an industry direction in the future that gives the product taste, taste, sugar and functionality. For example, some people want high sugar, others choose low sugar, and some consumers expect products to be complementary to certain health functions, and if ai can be used well, it can be achieved in plant factories and greenhouse conditions。
