Machines that learn by themselves. This is what the branch of artificial intelligence known as “Machine Learning” is all about. At AINIA we are already applying it to quality control and food safety systems, from the control of Petri dishes in food analysis laboratories, to the analysis of data to optimize production processes in the industry. In the article we give the keys.
The complexity of the products and the manufacturing processes of the food industry, as well as the volume and disparity of data that are generated, force a decisive advance in the application of automatic learning techniques (“Machine Learning”) to extract information that helps decision making.
These types of techniques make it possible to classify the huge amounts of data information in real time and, therefore, make full use of the potential of industry 4.0. If we train machines so that, like neural systems, they solve complex problems such as defect detection or pattern identification in real time, the potential of digital image processing is immense.
These techniques require great computing power to process the large amount of data using algorithms that allow parallelization. Machine Learning uses the high calculation capacity of GPUs (Graphic Processing Units) allowing analysis to be carried out in real time. The previous graph (Source NVIDIA) represents the increase in speed when using a GPU in the processing.
Digital image processing is a discipline that integrates various areas of knowledge such as computing, electronics and physics. This discipline allows the analysis and processing of digital images through the use of computers, in order to extract useful information from them to carry out a particular task. In our case, we apply it to quality control and online food safety, in different production sectors (nuts and snacks, fruit and vegetable products, meats, wines…).
Our experience allows us to affirm that digital image processing is an area of great importance in the industry, since it helps:
- Detection of defects in processes
- Automation of production operations
- Product classification
- Quality assessment
At AINIA we have been developing applications in this field for decades and now we have taken another step by incorporating Machine Learning techniques into digital image processing, with truly surprising results.
Can Machine Learning improve the microbiological control of food?
The use of Machine Learning techniques in the automatic reading of Petri dishes in AINIA’s food control laboratories is allowing, in its own R&D work, to obtain an overall success rate of 95% compared to the 68 percent obtained using conventional pattern recognition techniques.
In the image you can see the results obtained in traditional plate count, with a hit rate of 46/53 colonies detected, while the same sample processed using machine learning obtained a hit rate of 53/53.
Experiences in the application of machine learning in the food industry and in the field of health
The detection of objects in real time is another of the fields of application of this outstanding discipline of artificial intelligence. Google is already doing it, for example, in the classification and distinction of different materials and objects in real time and with high precision using the machine learning reading of complex databases fed through electromagnetic waves that leave a unique imprint.
The diagnosis of diseases from medical images is another field of application where there are already demonstrable experiences. An example is the self-diagnostic tests carried out by Harvard University.
In the food industry, these techniques are very useful for quality control. A clear example of this is the application of maching learning techniques to evaluate the internal quality of the fruit, to predict the quality of wine or to classify meat products.
At AINIA we are specialized in the development of technologies applied to Industry 4.0 in the field of food and health. If you think that we can cooperate with you to ensure that artificial intelligence can help you optimize your processes and improve your value contribution to the market and to the final consumer, call us, we will be happy to tell you what we have done and how we can help you.