Crop disease detection for smallholder farmers

e-Agri in Developing countries

Applying close-proximity hyperspectral imaging of the early onset of crop diseases to minimise preventable losses in emerging food production via the application of contemporary concepts into a new low-cost mobile attachment.

There is a significant yield gap for agricultural crops in developing countries versus those in the west, where mechanisation and agricultural technologies have a far greater adoption. Pest and fungal disease can have a significant impact on these yields and overall profitability, resulting in many cases to adversely affected livelihoods due to poor decision making in the lack of effective knowledge.

Paired with a rising uptake of new inexpensive smartphones in the developing market, due to availability and farming trade information access. This makes for an exceptional framework on which to develop new technologies to better manage fungal and bacterial diseases that may reduce future harvests.

People

Dr Bruce Grieve, Director of the e-Agri Sensors Centre

Mr Charles M Veys, Research Engineer

The challenges

It can be very difficult to correctly identify the infiltration and spread of a fungal pathogen in time to prevent major economic impact.  Given the scale and diversity of crops, varieties and diseases even the most experienced of farmers can incorrect diagnosis. This typically results in one of two negative outcomes such as

  • scheduled pesticide spraying where it is not needed
  • late spraying to minimise yield loss.

With an automated diagnosis, compared against a growing database of known pests/pathogens, this problem could be resolved. However, it would be impracticable to develop the breadth of data in laboratories to be able to effectively diagnose in the field. Thus initial trials using a data mining version would be necessary, both initially and periodically to update the results.

Smartphone to fight crop disease

Our research

Supported by EPSRC, the School of Electrical and Electronic Engineering is partnering with a small Anglo-Indian company, Barefoot Lightning Ltd (BfL), to develop a very low-cost sensor system for close-proximity hyperspectral imaging of the early onset of crop diseases.

It is closely linked with the work done in conjunction with the University of Bonn, where we managed to replicate results by replacing a high cost spectrometer with low cost narrow band emitters.

How it works

A low cost tool will be able to scan for disease signatures, i.e. spectral and spatial signals, and the results compared against known values.

In the field it can be put in the hands of field extension workers, semi-technical staff employed by agronomy companies, NGOs and/or regional government bodies.

Barefoot Lightning’s libraries

Extension workers may then scan plants for disease signals already stored in Barefoot Lightning’s libraries or, where new disease signals are found, they can be added to the libraries, thus enabling BfL to build up the database and simultaneously providing an early warning system.

Barefoot Lightning, will also make use of the same technology to look for plant disorders such as nutrient deficiencies, so creating a high-value dataset relating crop-stress to hyperspectral data, which may then be exploited further by academic groups and/or marketed to agronomy companies.

Other applications

Disease signals expanded libraries can also then be used in the simpler ‘farmer versions’ to allow them to search for early warning signs of disease in their fields.

Barefoot Lightning Ltd, will also make use of the same technology to look for plant disorders such as nutrient deficiencies, so creating a high-value dataset relating crop-stress to hyperspectral image data. The latter may then be exploited further by academic groups and/or marketed to agronomy companies.

Future development

Although the pilot market is in India due to existing industrial infrastructure, the smartphone based technology has the potential to impact global farming communities, food supply and even animal wellbeing.

Plant virus control in Sub-Saharan African cassava farming

In particular the e-Agri team are looking into extending the sensors for use in plant virus control in Sub-Saharan African cassava farming, alongside colleagues in the University of Greenwich and the Bill & Melinda Gates Foundation.

Animal and human health assessment

Broader implications for animal and human health assessment are also being developed, notably with the University of Liverpool  and The University of Leeds. In the longer term there are plans to bring the technology into the hands of growers in the UK to help the growing community of home producers of food.

The opportunities are very broad and the team are always working on new ideas.

▲ Up to the top