Fetal timeline: AI will ensure high quality tangerines on the shelves

Russian specialists have created a program complex with a computer vision system to control the quality of fruits and vegetables. The developers are already using the invention to sort tangerines and apples for research purposes. Unlike analogs, the system monitors the fruit throughout its entire journey along the selection line, which allows to assess its condition from all angles. According to experts, the use of artificial intelligence is an important step forward in such technologies. However, the question remains whether it will be able to assess the quality of products as accurately as it does now with the help of samples.

Analyzing fruits from all angles

Skoltech specialists have developed a software system for analyzing the quality of fruit and vegetable products based on artificial intelligence. Unlike analogs, the system is able to detect defects on the entire surface of the object, not only on the side that first comes into the field of view of the camera. This is done due to the fact that AI distinguishes each individual fruit or vegetable among others and accompanies it in the process of movement along the sorting line. During this process, the fruit changes its position and the computer sees its state from different angles. With the help of a multispectral lens, the development detects rotting processes before they can be detected by the naked eye. At the same time, the system not only detects the fact of product spoilage, but can classify it: reject it or send it for recycling.

- Our system is able to use a multispectral camera to detect signs of rot on fruit before they are visible to the human eye. It operates in the near infrared range with a wavelength of 900-950 nanometers. If more sensitive equipment becomes available, we can do it even earlier. To recognize fruits and vegetables, we use artificial intelligence, which we train on images, because in fact video is a high sequence of frames," said Nikita Stasenko, junior research engineer at the Skoltech Agrotechnology Project Center.

The developers have already tested the system's capabilities on apples and are now adapting it to work with citrus fruits, particularly tangerines. It is assumed that the complex will include several cameras, which are placed above the sorting line, and a computer. At the current stage, the system processes images at 17 frames per second. If there is something wrong with the fruit, the AI signals it to the technologist monitoring the process. Now the creators are training the artificial intelligence to assess the degree of damage to objects according to checklists that are used to sort fruits and vegetables at the bases. For example, fruits with mold are rejected, and if they show only the first signs of rotting, they are sent for recycling.

- We plan to present the first sample of the system ready for practical use in three to four months," said Nikita Stasenko.

We need high accuracy of quality assessment

According to Natalia Shishkina, head of the department of refrigerated processing of fruits and vegetables at the All-Russian Research Institute of Canning and Vegetable Processing Industry, technologies of automatic sorting of fruit products are very necessary for the industry, but their implementation is limited by the issue of financing.

- In principle it is interesting. The industry needs such systems, but we don't have enough money, so we have to use much simpler methods of control. But instrumental methods are the future," she told Izvestia.

Russian fruit and vegetable producers are actively introducing automatic sorting systems of Western manufacture. Their willingness to use domestic developments will depend on their quality, said Pavel Tyurin, chief technologist of the Plodovoye society.

- Such systems are very useful. Many fruit diseases cannot be detected by eye. For example, apples have radial and lateral vitreousness, external rotting, worms. And new technologies allow us to remove such products, and the network is already more qualitative products. Now we are installing such a line manufactured in the Netherlands. We are ready to use Russian equipment, if it will show high quality of work, easy to maintain, at an acceptable price and by the totality of parameters will be better than competitors, - he told "Izvestia".

Easy Entry: RF Requests Tariff Relief on Imports of Fruit and Vegetable Products

Why the innovation may be adopted and how the price of goods will change

The development of Skoltech specialists is extremely relevant at the present stage, but raises questions as to how accurately the invention will be able to determine the signs of rotting fruits and vegetables without sampling, which is still used in the reception, said Associate Professor Regina Gurina of the Agrarian and Technological Institute of PFUR.

- In order to create a program complex based on AI, it is necessary to study tens of thousands of samples of various fruits and vegetables, to create an up-to-date database and constantly replenish it. Some exotic fruits began to appear on the shelves in our stores not so long ago and still appear, suppliers change. Now the method is used to conduct experiments to assess the quality of post-harvest fruits and vegetables, which will allow us to create a database and process them. But for now it is a test system," said Regina Gurina.

The advantage of the complex, compared to existing solutions, is that the creators are actively implementing AI to assess not only by appearance, but also with the ability to look at the processes inside fruits and vegetables without penetrating them, the specialist noted.