Innovative techniques for measurement
The list below highlights some of the problems associated with traditional data collection methods. So there are clear reasons for adopting new technologies that help reduce the measurement burden.
- Time-consuming
- Labour-intensive
- Subjective and non-standardised
- Limited scope and scale
Figure 1: Rotten apples on the floor. ©Peter Neumann via Unsplash.
This Step explores some of the new technologies currently under development within the context of the FOLOU project. Figure 2 outlines some of the transformative potential of precision agriculture technologies (e.g. drones with sensors and imaging technology), cold chain storage, AI-powered monitoring systems and blockchain for supply chain transparency in relation to reducing food and production losses. Two of these examples are explored in further detail after the table.
| Level | Technology and potential application |
|---|---|
| Food loss in agriculture | Tractor and drone-mounted camera images analysed with Artificial Intelligence (AI) technology. – Assess crop maturity and health – Identify best time to harvest – Provide food loss estimates – FOLOU project example |
| Production loss in agriculture | Time series of multispectral camera images from drone and satellite data – Model crop growth – Provide yield and production loss estimates – FOLOU project example |
| Production loss in aquaculture | Multispectral camera technology and AI analysis – Conduct automatic counting – Detect non-viable fish eggs in the hatching systems – Enable non-viable eggs to be automatically sorted and removed – FOLOU project example |
| Production loss and food loss in aquaculture | Blockchain and distributed ledger technology – Secure and traceable food loss data in cold chain (temperature-controlled storage and transportation) systems – FOLOU project example |
| Predicting changes in consumer behaviour (indirect method) | Market demand tool based on social media data – Predict food demand and consumption trends – Opportunity to better align supply with demand – FOLOU project example |
Figure 2: The potential technologies that could reduce production loss and food loss and how they are being applied within the FOLOU project
Remember that production loss occurs before a crop is ready to be harvested. Food loss starts from the point at which a crop is harvest-mature. Look again at Step 2.2 to see the difference between production loss and food loss.
Case 1
Applying deep learning (Artificial Intelligence) technology to measure pre-harvest and harvest losses using high-resolution images.
Here tractor or drone-mounted cameras take high resolution images which are analysed using AI to assess the maturity and health of a crop in the field. This helps identify the best time to harvest and also provides an automated estimate of food loss (for example any food that unharvestable for various reasons).
Within the FOLOU project this technology is being applied to cauliflower production. Cauliflower is a good study crop because it is a high value crop which is grown in many countries across Europe and a given field may be subject to multiple harvests over a period of time. Countries including Spain, Italy and France are key producers, supplying 150,000, 130,000 and 110,000 tonnes respectively in 2020. So the opportunity to both monitor the growing crop and also reduce losses is considerable and could lead to reduced food loss and potentially better financial margins for farmers.
Detailed images of the study fields are captured by drones or tractor-mounted cameras. Computer-based systems are then trained to recognise the degree of plant maturity, size and diseased cauliflowers. Initially the images captured for each cauliflower are marked manually. Semi-autonomous algorithms are used to replicate this manual marking on subsequent picture frames and over time the system “learns” to identify individual plants and their “status”. As the technology further develops the aim is that the images will provide near real time estimates of crop yield and harvest potential, as well as estimating loss in the field. Figures 3 and 4 illustrate aspects of the technology in use.
Figure 3: High resolution imagery being captured by drones ©dilepix
Figure 4: Identification of individual plants and their status using learning algorithms ©dilepix
Case 2
Automated fish egg sorting using multispectral camera technology in trout aquaculture.
In 2019, the world trout production was around 940,000 tons (FAO) and has been increasing since 2015 (+21 % in volume between 2015 and 2019). The main species farmed is the rainbow trout (Oncorhynchus mykiss) which accounted for 97 % of the total volume in 2019. The EU is the second largest producer in the world (183,000 tons in 2019: 20 % of world production). The rainbow trout farming industry is well-established and many aspects rely on highly efficient, proven systems. However, current research and development is continually attempting to increase production efficiency and sales.
Currently, early embryo development is monitored by visually inspecting the upwelling incubators (a special kind of incubator that moves nutrient-rich water towards the surface), manually removing dead embryos (white), and using chemicals to disinfect the whole batch. This is time-consuming and not sufficiently accurate to prevent the spread of bacteria and fungi.
In this case, machine learning can be used to detect infected fish eggs. Multispectral cameras can detect differences in embryo colours. This identifies the status of fish eggs which in turn allows infected fish eggs to be automatically sorted and removed. Although the technology still at the ‘proof of concept’ stage, it is likely to have a significant efficiency and financial benefits to aquaculture systems. You can find more details on the FOLOU website including pictures, graphics and illustrations of the technology.
Reach your personal and professional goals
Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.
Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.
