Moving towards smart and sustainable agriculture - a roundup from the SFN+ 5th Annual Conference on 3rd and 4th July 2023
Scaling up food production to feed the growing population would be a daunting challenge by itself, but somehow this must be achieved in the face of climate change, degraded natural resources, and increasing pressures on land, water and energy use. However, as the SFN’s 2023 Annual Conference demonstrated, projects supported by the network are redefining how we grow and produce our food – offering hope for a future of ‘Smart and Sustainable Agriculture.’
Keynote talk: The Odisha Millets Mission experience, Dr. Arabinda Kumar Padhee
Technological advancement may be a key driver of agricultural progress, but according to Dr Padhee, even greater transformational change could be achieved by switching to climate-resilient crops. Millet, for instance, requires significantly less water and land than many crops and has a low carbon footprint, yet is highly nutritious. Historic low investment, however, has resulted in millets being underappreciated. The Odisha Millets Mission is fighting to change this by working with a wide range of stakeholders, from farmers and producer organisations, to NGOs and women’s self-help groups. Their approach includes improving productivity of millet-based crop systems, conserving and promoting millet landraces, and developing processing enterprises. Simultaneously, the project is applying various creative means to increase consumer demand for millet products, such as sponsoring sports teams, partnering with chefs, engaging schools, promoting millet-based snacks, and even composing a song for the UN’s International Year of Millets.
“The Odisha Millets Mission is using a fork-to-farm behaviour transformation approach, with a focus on integrating millets in rural and urban food systems and value chains, by involving primary producers, mostly smallholders, women self-help groups, and other stakeholders."- Dr Arabinda kumar Padhee, Principal Secretary, Agriculture and Farmer’s’ Empowerment, Government of Odisha, India
The featured projects:
AI-AMMS: AI coupled with Aerial iMaging and Mobile Sensing by Dr Kang Liu , The University of Sheffield
Could an AI-informed imaging system help farmers to avoid excessive pesticide use? AI-AMMS is investigating whether a drone-mounted hyperspectral imaging system could detect wheat diseases based on changes in light reflected from the plants. Their aim is to develop a deep-learning based model that can detect diseases before visible symptoms are apparent, then recommend whether pesticides are needed based on known regional crop disease tolerance levels. The project has already developed a model with an accuracy of around 95%, generated interest from key stakeholders, and secured additional follow-on funding.
‘Our project has shown the potential for adaptive AI-enabled mobile intelligence for climate-smart pest management that can optimise sustainable and resilient farming.’ Dr Kang Liu
AIKPlatform - Preservation and Visualization of African Indigenous Knowledge for Resilient Food Systems 2.0 by Dr Steven Sam, Brunel University London
Most food in Africa is produced by small-scale, traditional farmers but the penetration of modern agriculture, rural-urban migration and land grabbing are causing valuable indigenous knowledge to be lost. The AIKPlatform project is countering this by building a digital platform to capture, store, and share this knowledge for learning and research. Drawing on STFC capabilities in citizen science, the team worked with farmers to co-design a platform that integrates farmer knowledge and experiences (for instance, captured by self-recorded videos) with quantitative data. In the future, the team hope to add an interface for food-related datasets and to develop learning resources for schools.
‘The AIKPlatform is promoting resilient food systems by bridging the gap between indigenous and scientific knowledge; facilitating knowledge sharing and learning; and promoting gender equality and inclusive agricultural development.’ Dr Steven Sam
Corn yield prediction via integration of remote sensing, machine learning and crop modelling by Mr Vivatvong Vichit-Vadakan , SkyVIV
Sweetcorn is a major cash crop and export product for Thailand, but once harvested it must be processed within 24 hours. With no system to optimise deliveries to processing plants, this results in frequent times of over- or under-supply. This project is investigating two potential methods to help predict sweetcorn yields: a multispectral imaging system coupled with machine learning algorithms, and a crop modelling software based on input variables (such as corn cultivar, amount of fertiliser, temperature, etc.). So far, the team have achieved around 90% accuracy with both methods, and are now investigating whether they can be combined to improve this even further.
‘Being able to predict yields of sweetcorn crops would result in eliminating waste, as well as optimising processing plant operations. We anticipate that this approach can be applied to other key economic crops of Southeast Asia, particularly rice, palm oil, coffee, and high value tropical fruits.’ Mr Vivatvong Vichit-Vadakan
Evaluating influence of moisture in controlling release of nutrients in novel green fertiliser using neutron imaging and muonic X-rays by Professor Ruben Sakrabani , Cranfield University
Organic fertilisers offer a more sustainable alternative to conventional, mineral-based fertilisers, but their higher variability makes it difficult for farmers to consistently apply the right levels of nutrients. Following a successful SFN Proof of Concept project, Professor Sakrabani is leading follow-on work using neutron computed tomography and muonic X rays to investigate how moisture content affects the physical integrity and chemical composition of a novel pellet-based organic fertiliser. So far, they are working to determine the critical point of moisture absorption that causes the pellets to buckle and release nutrients, which could help to optimise spreading regimes.
Read more here in this SFN blog post.
‘Besides reducing our reliance on chemical fertilisers and lowering our carbon footprint, optimal use of organic fertilisers will help farmers to apply the right nutrients to their crops at the right time.’ Professor Ruben Sakrabani
Monitoring tropical pollinators in conventional and organic fruit orchards in central Thailand by Dr Maria Anastasiadi, Cranfield University
Pesticide use is known to be a major cause of declining pollinators in Western countries, but very little is known about how these are affecting tropical bee species. Using deep learning methods, this project has developed an automated application for bee monitoring based on data captured by time-lapse cameras in guava orchards in central Thailand. For most species, the model has an average accuracy of over 90%, performing well on both static images and video footage. The team have developed the tool into a user-friendly interface that allows bee species diversity and abundance to be readily compared between organic and conventional agricultural systems.
‘Collecting data on pollinators can be labour intensive and time-consuming, meaning that new and effective ways to assess their activity were urgently needed.’ Dr Maria Anastasiadi
Non-disruptive in situ root imaging to investigate the role of soil microbes in cowpea drought stress-adaptive responses by Dr Steven Chivasa , Durham University
As a high-protein crop, cowpea can be a nutritious food source for smallholder farmers in Zimbabwe. However, high levels of soil degradation limit the crop’s productivity and make it more vulnerable to drought. By carrying out non-invasive imaging at STFC’s Diamond Light Source facility, this project is investigating whether a lack of soil microbes is the key reason for this. This confirmed that soil microbes are essential for cowpea to develop a good root system and to withstand drought. The group are now using genetic approaches, such as DNA sequencing, to try to identify the specific microbes responsible for these effects.
‘There is increasing interest within the agrotechnology industry in ‘bioprospecting’ for microbes that we can use to protect crops against drought and other stresses.’ Dr Steven Chivasa
Species identification using sRGB image technology for field-based biomonitoring of agricultural ecosystem by Dr Rajneesh Dwevedi , The University of Delhi
Soil-dwelling organisms play many essential roles in promoting healthy soils where crops can thrive, but there are few accessible tools to monitor soil biodiversity. This project aims to develop a machine learning-powered app for Indian farmers that can identify species from uploaded images. After collecting sixty soil samples from across North India, the team developed a prototype model for identifying organisms present on the soil surface that has already revealed significant differences between organic and intensive farms . The team are now working to expand the range of species the model can identify, including those that dwell deep within the soil.
Read more here in this SFN blog post.
‘Ultimately, we hope this tool will encourage farmers to switch to sustainable agricultural practices that are not just beneficial to us, but also friendly to soil fauna as well.’ Dr Rajneesh Dwevedi
Artificial Neural Network based Segmentation to detect objects in Hyperspectral image data captured for agriculture by Ms Neetu Sigger , The University of Buckingham
Hyperspectral imaging has immense potential for real-time monitoring of agricultural systems, but the massive amounts of data make it difficult to identify relevant features. Drawing on STFC capabilities in artificial intelligence, remote sensing, and crop image analysis, this project is investigating methods to efficiently extract data from hyperspectral images to aid agricultural decision making. Using an approach based on diffusion modelling, they were able to extract both spectral and spatial dimensions, and map land use (?) from images with over 81% accuracy. The team now intend to expand the model, for instance to identify crop varieties, monitor plant growth, and detect crop diseases.
‘Our model also has many potential commercial applications such as in urban planning, disaster management, and forensic analysis.’ Ms Neetu Sigger
You can watch the session ‘Smart and Sustainable Agriculture’ here (Keynote between 9:00- 31:00; project presentations from 32:20 onwards)
Over the next few months, we will be showcasing more of the sessions from the SFN+ 2023 Annual Conference. To keep updated about new posts, you can sign up to receive the monthly SFN+ email newsletter.
September 2022 - Caroline Wood, Freelance Science Writer