Artificial intelligence (AI) in agriculture will help detect pests, plant diseases and nutritional deficiencies on farms. AI sensors can identify and target weeds before deciding which herbicide to use in an area. Precision agriculture, often known as AI systems, helps improve the overall quality and accuracy of harvesting.
The role of computer vision
It takes a lot of land to feed billions of people. It is impossible to cultivate by hand these days. At the same time, crop failures are often caused by pest infestations and plant diseases. These damages are difficult to detect and stop given the scale of modern agricultural operations.
This is a new application of computer vision technology. Aerial photography is used by growers to identify early indicators of plant diseases or pests at the macro level and crop diseases using close-up pictures of leaves and plants at the micro level. A common method of computer vision used in this study is Convolutional Neural Networks.
The term ‘computer vision’ is used quite broadly here. Images are often not the most reliable source of information. The best way to study many important aspects of plants is by other methods. It is possible to collect hyperspectral images with special sensors or perform 3D laser scanning to better understand plant health. In agriculture, these technologies are increasingly being applied thanks to AI.
In general, this high-resolution data type is more like a medical image than a photograph. AgMRI is one of the field monitoring systems. Specialized models are required to process this data, but convolutional neural networks can be used because of the spatial organization of the data.
Research into plant phenotyping and imaging of plant traits shaped by genetic and environmental influences is receiving millions of dollars. The main task at hand is to collect significant data sets (in the form of pictures or three-dimensional images) about crops and collate phenotypic information with plant genotypes. Research findings and information can be applied to advance global agricultural technology.
How are robots being used in agriculture?
According to Dataconomy, a data science publication, Prospero (robot farmer) and other autonomous agricultural robots can dig holes in the ground and plant seeds while adhering to established basic patterns and taking into account the unique characteristics of the region. have the ability The robot can also manage the growth process and interact with each plant individually. Robots harvest and process plants at the perfect time.
Herd farming is the basis of Prospero. Imagine a bunch of small Prosperos crawling over a crop, leaving behind a neatly organized plant. Prospero first appeared in 2011, before the current deep learning revolution reached its peak. Thanks to the rapid penetration of robots today, more and more tasks in agriculture can be automated
Small and agile atomizer drones can deliver hazardous materials more accurately than larger aircraft. Aerial photos taken with atomizer drones can also be used to collect data for computer vision algorithms.
A growing number of robots designed specifically for harvesting are being built and deployed. Combine harvesters have been in use for a long time. But making a robot that picks strawberries, for example, has only recently become possible with advances in computer vision and robotics. Individual weeds can be recognized and mechanically removed by robots such as Hortibot. Using manipulators to distinguish between weeds and beneficial plants and to interact with smaller plants was previously impossible, but thanks to advances in modern robotics and computer vision, it has become possible.
While many agricultural robots are still prototypes or small-scale testing, it is already clear that machine learning (ML), AI and robotics can work well in agriculture. It is also clear that agricultural activities will be mechanized in the near future.
Currently, more applications for AI in agriculture are being developed. The Neuromation pilot project, for example, applies computer vision to the livestock sector, an area that has not yet attracted much attention in the deep learning community.
Machine Learning and AI in Agriculture
Of course, there have been attempts to utilize livestock tracking data for machine learning. Pakistani company Cowlar, for example, has launched a collar under the motto ‘FitBit for Cows’, which wirelessly monitors cattle activity and temperature. French researchers are working on facial recognition technology in cows.
There are also attempts to use AI in pig farming, a hitherto under-utilized field with a market value of hundreds of billions of dollars. Pigs are raised in relatively small groups on modern farms. Feed is a major cost of pig production, so the main goal of modern pig production is to maximize the fattening process.
If farmers had a comprehensive knowledge of pig weight gain, they could have solved this problem. Animals are often weighed only twice in their lifetime. When to start and when to end. Experts can design a unique combination of fattening regimens and feed additives for each pig if they know how each piglet is gaining weight. This will greatly increase production.
It’s not particularly difficult to put pigs on a scale, but it puts a lot of stress on them and the stressed pigs lose weight. A new AI study seeks to create a new approach to weighing animals. The pig’s weight is inferred from photo and video data by neuromation using a computer vision model. These estimates are integrated into existing analytic machine learning models to improve the fattening process.
What is the future of AI in agriculture?
Agriculture and animal husbandry are sometimes considered obsolete occupations. Today, however, AI in agriculture is becoming a common tool on many farms. The primary reason is that there are many simultaneous operations in agriculture. They are too complex to be automated using deep learning and modern AI. The grown plants and pigs did not come from the same assembly line. Human intervention was absolutely necessary until recently, as each tomato bush and pig requires a unique method.
We can use AI to solve challenges while automating technologies that interact with plants and animals and take into account their unique characteristics. Weighing a pig is simpler than learning how to pass the Turing test, which determines whether a machine has human-level intelligence, and driving a tractor in a large field is like driving a high-traffic car. Simpler.
Since agriculture is still one of the largest and most important businesses in the world, even small improvements in efficiency can yield significant benefits.
AI Times member Chan Park cpark@aitimes.com
[관련기사]Agricultural robots for rice planting and harvesting… The upcoming ‘smart agriculture’
[관련기사]’Endless potential for agricultural growth through AgTech’… How to use it?
