Applying AI in Vietnam’s agriculture

February 19, 2019 | 10:00
(0) user say
With its strong performance over the past few years, Vietnam’s agricultural sector is surging to become smarter. Nguyen Ky Tai, researcher at the University of Southern Queensland in Australia, provides insights on the role of Artificial Intelligence and Big Data in this sector.
applying ai in vietnams agriculture

Climate change has been bringing more and more natural disasters, causing bad impacts on ­agriculture, one of the most complex fields, containing many uncertainties and ­uncontrollable factors of the environment. As an agriculture country, in recent years Vietnam has been steadily ­applying Artificial Intelligence (AI) and Big Data in the sector to limit the negative impact of climate change. However, there remains many obstacles.

In 2018, for the first time, the government invited 100 overseas intellectuals to take part in the Programme of Connecting Vietnam’s Innovation Network. At the event, hosted by the ministries of Planning and Investment, Science and Technology (MoST), Agriculture and Rural Development (MARD), and Information and Communications, Prime Minister Nguyen Xuan Phuc emphasised the importance of hi-tech and AI in agriculture along with other areas. Through this programme, many have partaken in two projects on applying AI and Big Data in agriculture.

The first is Digitalised Vietnamese Knowledge System, which is being led by Deputy Prime Minister Vu Duc Dam and MoST Deputy Minister Bui The Duy. The project will build up Big Data sources for developing AI applications in agriculture, transport, medicine, and more. A Big Data platform in agriculture is being built and integrated by a research team from the MoST, the MARD, the Vietnam Academy for Water Resources (VAWR), the Ho Chi Minh City Space Technology Application Centre, and other agriculture companies.

The second venture is a collaborative project for applying Big Data and AI in agriculture between the Ho Chi Minh Open University and the University of Southern Queensland, with research funding from the Agriculture Hi-tech Park in Ho Chi Minh City. This project will focus on the methods to collect data to develop AI models for training and comparing impacts of irrigation and fertiliser control systems, and soil types on crop yield.

AI requires huge amounts of data for training but databases in Vietnam have been unable to answer the demand, and costs are high. Therefore, it is expected that costs would be cheaper if many households or enterprises applied the same, or a similar, system.

It is necessary to understand that AI and new ­technologies still have to be based on the real background of agriculture. Tech just helps ­increase added value, lowers expenses and risks, and ­predicts uncertain elements. Sometimes, we do not need to invest in hi-tech but simply complete all basic stages, and with that we can make big changes in Vietnam’s ­agriculture.

As it stands, the Vietnamese government has not developed any long-term plan to orient, invest, develop and apply AI in agriculture. Enterprises and localities do this by ­themselves, and therefore there is little synchronisation and ­expenses increase.

In addition, human resources (HR) is also a big problem when most ­Vietnamese agricultural ­officers and experts cannot work with AI and Big Data.

Australia’s agriculture AI

In Australia, by applying a range of data and scenarios, farmers are able to receive ­insights into probable crop production outcomes and make more informed decisions around their farming practices.

Recognised as one of the biggest cotton-producing countries in the world, Australia uses water three times more efficiently than the global average thanks to industry-standard soil-water sensors, which contact with the soil and provide information based on a single point in the field. While this enables farmers to have some oversight of soil-water conditions, it does not account for the spatial variability found in the entire field.

Such AI crop models can determine the current and predict the future status daily soil-water, nitrogen and fruit load of cotton plants. This can be based on the day of the year, weather data, soil electric conductivity, soil moisture, vegetation indices, plant density and canopy size, and visual plant response is captured using cameras.

Alongside this, with predictions of higher temperatures and changes to rainfall rates in Australia, the country’s wheat producing faces an inevitable decrease in yield. Therefore, it is imperative for farmers to be able to eliminate other factors which may affect crop outcomes like crown rot, a fungal disease found in a variety of winter cereals, and which is of great concern to farmers. The disease is passed down from crop to crop through infected stubble and can have a huge impact, costing Australian wheat growers an average of $80 million a year.

To handle this, Tai and his team in Australia have used multispectral, thermal and visible cameras to improve identification and data collection in infected research plots. The images are analysed to extract the differences between the crops with higher levels of partial resistance and those which are more susceptible to crown rot.

Limit negative impacts

Big Data, AI, machine learning, and deep learning are the breakthrough application of science and technology in agriculture to enable farmers to have more insights about the consequences of their actions and take a much better and informed decision on farming practices.

Big Bata can further provide a wealth of information about soil, seeds, livestock, crops, costs, farm equipment and the use of water and fertilizer. Using Big Data, AI models can analyse and provide insight into how to optimise yield, improve planning, and make smart decisions in order to minimise waste and increase yields.

Applying AI in the sector

In the agricultural industry, the most popular AI applications focus on four main groups. The first is agricultural robots, which are being developed and programmed automatically to handle essential agricultural tasks. These tasks can include collecting data, identifying and spraying weeds, irrigation, and more.

Robot control technology uses rapidly-growing AI to help improve efficiency and address the challenges facing the agricultural industry, ­including crop productivity, soil health and herbicide ­resistance.

The second is monitoring of crops and soil. Technology for monitoring land, the ­environment and crops will also become important applications in the future as ­climate change continues to be ­studied and evaluated for ­environmental impact.

Large amounts of information collected every day from satellites, cameras, and sensors are a challenge for data integration, processing and analysis.

Third is predictive analysis involving machine-learning models, cloud computing, and the Internet of Things. Remote sensing techniques are developed to monitor and to predict different environmental impacts to crop yields. The benefits of improving ­accurate weather forecasts are determining the best time for planting, fertilising, pesticide spraying, irrigation and crop harvesting to maximise ­farming profits.

Finally, AI applications focus on precision agriculture, which will help farmers save on the costs of using water, fertilisers and pesticides. AI apps will improve management in agriculture, and detect pests and nutritional deficiencies of plant crops on fields. Crop information, soil conditions, pests, and weather will be digitised and managed with mapped layers that can interact with each other.

Automated AI applications are being developed to help improve crop productivity, which can help farmers know exactly where to place the ­fertiliser, and where to water and spray, leading to more ­efficiency and ­potentially greater results.

What the stars mean:

★ Poor ★ ★ Promising ★★★ Good ★★★★ Very good ★★★★★ Exceptional