Tackling Food Adulteration With Artificial Intelligence: Scope and Limitations
For a long time now, food adulteration has had a significant impact on the food processing industry. In developing countries, especially, the laws and regulations to prevent adulteration of food items have not been stringent. This not only negatively impacts the health of the consumers but also contributes to the recession of the economy of the country.
More than 60 million tonnes of food in India, which is worth thousands of crores of Rupees, go to waste every year. The majority of this wastage includes agricultural produce, dairy products, and other processed food items. The wastage may take place during the production, processing, packaging, consumption, or distribution stage. A considerable amount of this wastage can be attributed to the deterioration in the quality of the food items due to the addition of certain chemicals or adulterants in them.
Food adulteration essentially involves adding certain chemicals to the food products, which can prove to be hazardous to the health of the consumer. In fact, the adulteration of food items has become a business for many. With a view of earning more profits, many authentic products such as milk, spices, wheat, and so on are adulterated. This yearns for the demand for a technology that can detect and help prevent the addition of chemicals in food items that may deteriorate their quality.
Challenges in Traditional Methods of Detecting Food Adulteration
The most common food items which are adulterated include dairy items such as milk, curd, etc. Usually, the substances used to contaminate these products include urea, salt, and chemicals such as ammonium sulfate. The traditional way to detect these contaminants includes conducting specific chemical tests. These tests, however, are time-consuming, require a lot of effort and many resources to carry out effectively. For instance, some tests involve using a strip that changes its colour when dipped in dairy products in the presence of chemicals. These strips, however, take 5-10 minutes to change their colour and are also expensive. This imposes a significant barrier to the quick detection of contaminants. Moreover, since these tests are specific to particular chemicals, there is a chance that some other chemicals may go undetected. Moreover, manual detection of colour changes may be error prone as the colour difference may sometimes be hard to distinguish. In addition, quantifying the presence of chemicals may also not be possible.
Due to these reasons, the chemical testing method was soon substituted by electronic testing. However, the per-unit cost of the electronic devices and their maintenance involved large amounts of expenditure. Keeping these machines clean was yet another tedious task, and a failure to do so led to inaccurate results.
Tackling Food Adulteration With Artificial Intelligence
All the problems mentioned above in chemical testing and using electronic machines led to the need for alternate monitoring and control systems. The accurate detection of adulterated food items requires a robust monitoring system. Research done in the field of AI has led to the design of mathematical tools and models which are based on algorithms such as artificial neural networks (ANNs). These algorithms not only help in the detection of adulterants in food items but also identify the type of chemical/adulterant added to it. In addition, they can also compute the percentage of adulterants added, which helps in its quantification. Artificial Intelligence essentially helps reduce the reliability of visual monitoring, which would otherwise have been a source of erroneous results.
Another primary aspect of food processing, that is, the supply chain, can be monitored using AI. The supply chain is essential in determining the quality of the final product. Using AI to track the distribution of products and testing food items at different points in the supply chain can help detect the exact source at which the food became adulterated. This would not only lead to an increase in accountability but also ensure that the various participants in the supply chain take precautionary measures to prevent any adulteration.
Apart from visual indicators, using AI and electric sensors can also yield more accurate results as compared to manual analysis of detecting contaminants by ‘smelling’ the product. This will lead to faster detection of those chemicals that produce an odour different from that of the original product. Lastly, considering that another use case of AI is to identify any toxins or pathogens in the product, the algorithms can be programmed in such a manner that they not only detect the external substances but also identify any impure toxins in the food item which may have crept in during its production.
Machine Learning and Deep Learning for Detection of Adulterants
There are several machine learning algorithms that have been utilized to detect food adulterants. The results have shown a promising direction in the use of machine learning and deep learning to improve food quality. A report prepared by the International Journal of Applied Engineering Research shows the results of a test conducted on a sample of saffron to detect its constituents.
Due to the high cost of saffron, adulteration frequently occurs in the local market. An electronic nose system recognized the fragrance and fingerprints of saffron, yellow saffron, safflower and teased maize in the test. Characteristics of these signals acquired from the nose system had been retrieved and analyzed for analysis.
The primary component analysis and results were confirmed using artificial neural backpropagation networks. The results showed that the system could adequately distinguish the adulteration of saffron. To conclude, the electronic nose, with the use of artificial neural networks, identified the adulterants in saffron with 100% accuracy. Similarly, a distinction between safflower and adulterants was made with 86.87 per cent precision. Where the rate of adulteration was over 10%, the electric nose was successfully able to differentiate between the adulterated and non-adulterated saffron. The technique used for this detection involved image processing which was based on machine learning methods like OpenCV.
Indian Startups That Have Developed AI-based Food Adulteration Detection Technology
Many companies in India, particularly startups, have started offering AI-based solutions to tackle various challenges throughout different sectors. As for the food processing industry, some startups are working on developing products/technologies that can detect chemicals and other adulterants in food items.
RAAV Techlabs, an enterprise based in Delhi, has developed two products to perform qualitative analysis on food products. The first product can be used to detect the presence of adulterants in milk and other dairy items. Additionally, the second device can be used to measure and detect the chemical composition of fruits and vegetables.
So how exactly do these devices do this? The device which detects the chemical composition is a portable handheld one, which takes a time period of fewer than 5 seconds after coming in contact with the fruit or vegetable to ascertain its quality. The basis for this ascertainment is the measurement of the quantity of chlorophyll, brix, acidity, and moisture in the food product. The measured amounts then determine its ripeness, durability, sweetness, or sourness. This ultimately helps in finding out which products are not fit to reach the end consumer.
The presence of impurities like water, milk powder, urea, melamine, detergents, and paint is analyzed by the milk analyzer developed in RAAV Techlabs, which informs on the quality of milk. The fat, protein, and SNF content in milk is also analyzed. The results of these tests are then published on the application installed in the user’s handheld device. The process of testing uses Artificial Intelligence and machine learning-based algorithms, which are used to measure the interaction of electromagnetic radiation with the molecules.
Rahul Kumar, Co-founder of RAAV Technologies, told YourStory, “We have developed two devices capable of generating real time-data with regard to spoilage as well as the nutritional value of food items so that processes like storage, logistics, and harvesting can be planned and executed efficiently without wastage. These devices are capable of reducing food wastage in India by 50 per cent, provided they are utilised consistently for two to three years by farmers, middlemen, as well as consumers.”
AgNext Technologies is another Indian Startup that has developed a solution to detect adulteration, specifically in milk. This startup has used Near-Infrared (NIR) spectroscopy based on the principles of ultraviolet and infrared rays in its device to provide for a quality assessment of milk and ascertain its contents.
In about 40 to 45 seconds, it tests for adulterants which include, among others, melamine, vegetable oil, maltodextrin, detergent, urea, starch, sugar, and salt. AgNext claims that it can sample and test milk samples with up to 99% accuracy, which would give results that are quicker and more efficient than traditional laboratory tests.
Taranjeet Singh Bhamra, founder and CEO of AgNext Technologies, told The Better India, “The testing equipment is an IoT-enabled, AI-powered, handheld, battery operated device. The data is fed into Qualix—AgNext’s cloud-based SaaS platform—which helps provide a collective visualization of information across locations, allowing businesses to fast track data-led procurement decisions. This augments and replaces the intuitive and sensory method of milk testing with an unbiased scientific measurement.”
Limitations in Implementation of AI to Detect Adulterants
Despite its various benefits, in India, the adoption of AI comes with multiple implementation issues. A few of these issues are:
Lack of Stringent Laws and Regulations – The use of Artificial Intelligence and other data-centric technologies do not have stringent regulations governing them. This increases the chances of manipulation and access to unauthorized data, which means that adulterated products might go undetected, and people may take advantage of the lack of regulations.
Lack of Proper Definitions – While certain chemicals and substances used in adulteration are appropriately defined, there are specific terms in the analysis of data that have no standard definition, which might lead to different interpretations of data. Thus, a well-defined nomenclature of terms needs to be defined to provide for a robust and objective analysis of the tests.
Challenge in obtaining authentic samples – Since the adulteration tests cannot be done on every food item, an authentic sample representative of the quality of a batch of goods is required for accurate results. Identification of an accurate sample, however, is a limitation. The traceability of these samples in their supply chain is another necessity to achieve authenticity.
At present, AI is becoming pervasive in our daily lives. However, when it comes to incorporating AI in any sector, the general notion is that the existing jobs will be at risk. This notion, however, is false as AI also creates new job opportunities. When it comes to the food processing industry, the usage of AI becomes essential to ensure that the final products that consumers consume are both safe and natural. With the right regulations in place, the adoption of AI will perhaps help us fight food adulteration and bring about an increase in the quality of processed food items.