How AI/ML technologies are increasing agricultural productivity and profitability

The Covid19 pandemic has already had a damaging effect on agriculture and allied sectors across the globe. While local ecosystems have encountered severe disruption, global supply chains have completely crashed. The crisis will soon pass but one of its most critical impacts will be – firstly, faster adoption of digital technologies and secondly, increased mechanization across the value chains. This is where data science combined with artificial intelligence and machine learning (AI/ML) will come increasingly into play.

The whole concept of smart farming, which is making agriculture more efficient and sustainable, and thus profitable, is largely driven by AI/ML technologies. These technologies can be used in crop and water management, pest and disease detection, crop health monitoring and yield estimation, cultivating and harvesting by smart tractors without drivers as well as other types of forecasts and predictive analytics.

These technologies combined with others like remote sensing and big data bring about data-intensive processes in agriculture, which increases the efficiency and productivity of agriculture at a time when it has become imperative to produce more with less.  Here are a few ways in which AI/ML can boost agricultural productivity.

Improving harvest quality and managing disease

A key application of AI has been helping in identifying pests and diseases. Custom databases for specific crops and helps farmers identify pests and plant diseases with nothing but just a mobile phone. This saves human intervention, cost of hiring an expert and, most importantly, there is no delay in diagnosis.

Sensors are also being used to detect and target weeds. In some instances, robots are used to uproot weeds and in others, it helps in targeted application of pesticides. One research team that used AI technology to detect disease in cassava plants in Tanzania found that AI was able to detect disease with 98 percent accuracy.  Instead of spraying pesticides uniformly over the entire cropping area which is an expensive proposition for the farmer, ML can aid in targeting the inputs precisely in terms of time, place and affected plants. This can reduce the chemicals used and improve the quality of produce, and save cost.

Weather forecasting and advisory services

AI/ML is playing a significant role in advancing hyper-local weather predictions.  Using massive data coming from weather satellites combined with continuously expanding weather stations and IoT sensors on the ground, more accurate hyper localized weather predictions are becoming possible.  Some models go as granular as 4km resolution. This type of hyper-local weather data is increasingly utilized to provide targeted advisories in a given cluster of villages.  Using the historic weather, forecasted weather, type of crop and the stage of the crop various crop scenarios can be built using AI/ML technologies and those scenarios can be utilized to deliver targeted and precise actionable advisories to farmers at the village level.

Crop management

The timely information on when to sow the seed can make all the difference between a profitable year and a failed harvest. ICRISAT in collaboration with Microsoft has used a predictive analytics tool to determine the precise date for sowing for maximum yield.

Another important aspect of crop management is yield mapping and estimation, which helps in matching supply and demand. Using remote sensing technologies, NDVI models and climatic conditions, the Machine Learning models can be built for specific crops in a given region to predict accurate yield estimations. These types of advances in yield estimations are extensively being used by governments, thus eliminating or minimizing the costly crop cutting experiments.

Quality analysis

This is one of the significant use cases of AI/ML and will soon become mandatory across value chains for evaluating quality and aid in grading of agricultural commodities, both field crops such as cereals and pulses as well as horticulture crops such as fruits and vegetables. Imagine passing turmeric through a sensor and determining curcumin content in real time! Using hyper spectral imaging, AI/ML algorithms can be trained to learn the various quality parameters just by scanning the produce. This type of rapid quality testing is already there for many crops and is all set to transform how we trade and evaluate commodities.

Water management

No resource is as precious as water. Machine learning models, combined with IoT sensors on the ground, enables estimation of daily, weekly and monthly use of water and optimizes water usage. This is a significant jump in innovation from the currently used drip irrigation models and is transforming specific sectors already – such as vineyards, for example

Managing price fluctuation

Price fluctuation and lack of market insights have been damaging farmers’ incomes for decades. Thankfully, AI/ML models are now being developed that take into consideration historical trends, current cropping patterns and demand estimations to arrive at price estimates for different crops. This will empower growers to run farming like most modern businesses are run today.

Conclusion

AI/ML technologies have definitely helped in turning agriculture into a more scientifically managed activity, with the ability to assess input needs and predict output. These technologies are the absolute need of the hour, at a time when the agricultural system is getting more complex, and the pressure on producing more with less has never been higher.  It also helps farms of all sizes operate and function in an efficient way. Farmers will finally have the tools and the data to get the most from every acre.

By Dr. VenkatMaroju, CEO, SourceTrace