The idea belongs to first-year master's student Denis Mezentsev from the Institute of Computer Technologies and Information Security at Southern Federal University (SFEDU). "Due to the high percentage of incorrect sexing (determining the gender) of chicks, poultry farms and hatcheries waste time on chick care and, consequently, lose profits. The industry is primarily interested in hens that lay eggs and their meat. Our software-hardware complex allows for the automatic or automated determination of a chick's gender with high probability at an early stage of development – literally within a day after hatching," Denis Mezentsev explained.
The principle of operation is straightforward: a camera system uses computer vision to scan a box of chicks on a moving platform and sends a signal to the platform's microcomputer. The microcomputer processes the information and, depending on the gender, creates a route for the moving part. Traditional methods of determining gender (based on the bird's reactions and behavior, feathering, and coloration) are not always reliable, require at least 1-2 weeks for verification, and do not eliminate the so-called human factor.
The startup from SFEDU's ICTIB has already received support from the Innovation Promotion Fund. The developed software-hardware complex has scalable architecture capabilities.
"The year 2024 has been groundbreaking for the team working on the project. In addition to analyzing existing solutions worldwide, we assembled a dataset for training the neural network, designed the housing for the software-hardware complex, tested the system, and prepared the necessary documentation to obtain a patent," added Denis Mezentsev.
In December 2024, the startup was presented at the project-investment session of Sfedu Exponent and received high marks. The expert community noted that the potential market for the product includes 500 poultry farms in Russia and about 1200 in the CIS. Moreover, the net profit of Russian poultry farms for 2022 amounted to 5,967,929,000 rubles.
The project will be implemented at the Institute of Computer Technologies and Information Security at Southern Federal University, which provides scientific support and facilities for the system's operation.
"The team's plans include not only expanding the dataset but also automating the chick output. We also want to make the route flexible to accommodate landscape structure, and develop software with a user interface to allow the selection of not just hens, but also ducks, thereby expanding the functionality for sexing other birds.
We are already working on determining the gender of the chick not after hatching but while still in the egg," concluded Denis Mezentsev.