The Food Crisis – Resolving the problem of food waste and food loss
24/01/2023
Carbon Farming
10/04/2023
Machine Vision in Agriculture
What is Machine Vision?
You might already be familiar with the concept of “Machine Learning”, but what about Machine Vision? What does the concept bring to mind? As the name suggests, Machine Vision refers to the visual aspect or the ‘eyes’ of a machine, allowing it to visualise objects that appear in front of it. The system functions using digital input which is captured by a camera for processing and informing the machine of what procedure needs to follow. Machine vision has been contributing significantly to industrial automation and manufacturing, mainly by performing an automated inspection as a part of quality control procedures. Indeed, it has been practiced in real operational processes since the 1950’s and began gaining traction within the manufacturing industry between the 1980’s and 1990’s.
Let’s look at a simple example of a fill-level inspection system at a brewery to understand the technology better.
An inspection sensor detects the presence of the beer bottle passing through, triggering the vision system to cast a light on that specific area and capture an image of the bottle. A Frame Grabber translates the image that was taken into a digital format. The next step is storing the digital file in the machine’s memory to be analysed by the system’s software. A direct comparison is then made between the file and the predetermined criteria to identify defects. If an incorrectly filled bottle is detected, a failed response is delivered signalling a diversion to reject the bottle. The operator can also view the discarded bottle and real-time data on display.
This example demonstrates the usefulness of machine vision in automating a daily inspection procedure that is usually carried out by human workers. It boosts daily productivity and brings significant improvements to operational profit. A system such as this one is only possible with the combination of software and hardware utilising machine vision technology. The components that are typical to such a system include:
- Sensors
- Frame-grabber
- Cameras
- Lighting
- Computer and software
- Output screen or relevant mechanical components
- 1D Vision System:Instead of looking at the whole image at once, 1D Vision analyses a digital signal one line at a time. The technique usually detects defects in materials manufactured in a continuous process, such as paper cardboard and plastics.
- 2D Vision System:Mostly involves inspections that requires a range of measurements such as area, perimeter, shape, resolution, centre of gravity etc.
- 3D Vision System:Made up of multiple cameras or one or more laser displacement sensors. The latter allows the measurement of volume, shape, surface quality, and 3D shape matching.
There are currently three categories of measurement for machine vision systems: 1D, 2D and 3D.
Machine Vision Applications in Agriculture
Within the agricultural sector, machine vision technology holds various benefits. By utilising the ability of machines to capture and process visual information for analysis, machine vision can improve the effectiveness of the monitoring systems used to oversee agricultural assets such as crops and livestock.
Crop Monitoring
Machine vision has significant applications with regards to the monitoring of crops. By integrating machine vision into robotics and drone systems, farm owners can closely monitor the condition of crops, capturing detailed physical information which informs them on how to apply better management practices.
The combination of machine vision technologies with drone systems has become widely popular in farming operations over the past five to ten years. The open aerial manoeuvrability of drones allows them to be deployed across large swaths of farmland, collecting data on crops which can be analysed to improve practices. These systems also allow more advanced automated application to take place, such as seeding and pesticide application.
Specialised imaging technologies which detect colour signals that relay information on plant hydration can be used to analyse crop health and allow farmers to apply the necessary treatment to plants accordingly. Sime Darby Plantations in Malaysia have utilised these crop monitoring technologies in Malaysia and reported yield improvements of up to 15% as a result.
In China, XAG Co Ltd, China’s largest agricultural drone manufacturer, has developed a series of drones and self-driving utility vehicles which are able to perform automated seeding, fertilisation and pesticide application tasks supported by machine vision technology.
Livestock and Poultry
Machine vision systems are highly beneficial with regards to the monitoring of livestock and poultry. Cameras equipped with machine vision technology can closely monitor the condition of farm animals in a particular enclosure, allowing them to receive important information pertaining to the health of the animals.
This is a particularly important function, as these technologies allow farmers to ensure that animals receive the best treatment while remaining free from overly invasive monitoring methods. As animal welfare on farms has become an increasingly vital consideration for farmers, having these monitoring systems in place is highly beneficial.
Machine vision can monitor details such as the physical condition of the animals, behavioural aspects, as well as detect issues such as disease and pestilence in advance. This allows farmers to put effective measures in place to safeguard the health of their livestock.
Machine vision technologies have also been combined with automated systems to provide more seamless livestock management. In Japan for instance, a project by the farming company Mega Farms has incorporated machine vision into a robotic milking system for cows, allowing it to identify the cow, detect the location of the nipples and conduct a disinfection before milking.
With regards to poultry, machine vision applications have been used to oversee the health of chickens as well as to better monitor the conditions of their eggs, allowing farm owners to get a better gauge on the fertility of eggs and predict when they will hatch. Machine vision can also support the sorting and management of eggs when integrated with robotics systems.
Fish Farming
Machine Vision applications also extend towards fish farming operations. Automatic fish detection using computer vision allows farmers to monitor the number of fishes in a particular area, while information received on the weight, length of fish, as well as fish behaviour allows for the creation of intelligent feeding systems. These systems provide the necessary information required to reduce costs and boost production, while keeping the health and integrity of the fish well managed.
Phenotyping
Another application of Machine Vision with regards to the agricultural industry is within the practice of phenotyping. Phenotyping refers to the scientific monitoring of organic life to determine important genetic attributes about the subject in question. With regards to plant life, phenotyping allows researchers to study how the plants grow, what environments are most conducive to optimise crop health and other factors which can improve both the quality and yield of future produce.
This process has been conventionally carried out using manual observation methods, however, more recently, it has been done with the assistance of machine vision technologies. By closely analysing plants using machine vision assisted systems, scientists can breed and cultivate plant life and produce more productive and climate resilient species. Doing so allows farmers to select plants which will be most likely to grow productively and thrive in a sustainable manner.
The application of Machine Vision technology affords numerous benefits to farming operations. Chief among these is that it enables a more accurate and detailed monitoring of farming assets such as crops and livestock. Machine vision enabled cameras are able to analyse crop and livestock conditions in detail, providing a better understanding of their physical conditions which can be leveraged to improve practices and decision making.
The technology also allows farmers to improve the efficiency of processes. As with the egg management system, for instance, farmers are able to deploy these systems and improve the management of the eggs with minimal dependency on human labour. This effectively improves the speed of these processes tremendously.
Improvements in efficiency necessarily translate into cost reductions. As processes' efficiencies are improved, fewer resources are required to complete tasks. For instance, a better understanding of crop conditions allows farmers to apply water and pesticide in more precise ways, avoiding unnecessary wastage.
Machine Vision also allows farmers to observe livestock and crop conditions more closely, which improves the health of these farming assets. Better operational practices allow farmers to cultivate a better quality of crop and better safeguard the health of their livestock, which translates into a healthier final output.
The improvement to productivity is another benefit afforded by machine vision systems, as greater efficiency translates into improvements in the rate of output, and therefore an increase in overall productivity.
In addition to the above, machine vision technology enables better safety on farming operations, as the automated systems that are supported by machine vision reduce human exposure to hazardous elements.
Another considerable advantage to having these systems is that they improve animal welfare, as machine vision systems allow the close monitoring of animal conditions which improves the ability of farm owners to safeguard animal health and safety.
Potential Future Applications
While machine vision has already seen various possible applications within the area of crop and livestock monitoring, the potential for these technologies to be applied on a larger scale remains possible.
Research is presently underway to extend machine vision systems to major agricultural industries such as the oil palm, pineapple and durian industries, to name a few.
With respect to the palm oil industry, research is being carried out to improve the detection of FFB ripeness in the field. Machine vision technologies are being used as means of detecting FFB ripeness through detailed colour, spectral and thermal imaging, relaying information to farmers as to the maturity of the palm oil crop in question.
The ripeness of the palm oil fruit is a significant factor in determining the standard of oil that will be produced. A riper fruit allows more oil to be extracted and produces a higher quality of palm oil. These factors influence the price at which the oil can be sold at, as a higher grade of oil will find greater demand from buyers. Machine Vision therefore holds the potential to significantly improve both the productivity output and the quality of palm oil that reaches the market.
There is also the potential for machine vision to support weed detection and management with respect to the oil palm crop. These systems are presently carried out manually, however with the use of machine vision technology, plantation managers would be able to detect weeds and apply the appropriate level of herbicide and pesticide, reducing the associated costs. This would also have a favourable impact on palm oil yields.
Similar approaches have been taken with respect to pineapple and durian crops. Research is presently underway to support efforts to determine the maturity of these crops and determine the best time for harvesting using machine vision technology.
Conclusion
Machine Vision technology indeed holds the potential to provide significant benefits to the agricultural industry. Improving the oversight over crops and livestock through machine vision assisted monitoring equipment allows farmers to get a detailed understanding of crop and livestock conditions.
Machine Vision technology indeed holds the potential to provide significant benefits to the agricultural industry. Improving the oversight over crops and livestock through machine vision assisted monitoring equipment allows farmers to get a detailed understanding of crop and livestock conditions.
This information allows farm owners to apply better farming practices, which can improve process efficiencies, reduce costs, and improve overall farming productivity.
The benefits of machine vision extend towards the quality and health of crops and livestock, which is ultimately carried forward to consumers, who consume more nutritious products.
As machine vision technology continues to bring new improvements to agricultural processes, initiatives such as those mentioned above are underway to discover additional applications for the technology in relevant agricultural industries.
With the advent of 4IR technology systems and further technological advancements in the agricultural industry, machine vision will play an integral part in the development of comprehensive technological solutions that support the improvement of agricultural processes and food production operations around the world.
